Joe Verbout
/
main
Important changes to repositories hosted on mbed.com
Mbed hosted mercurial repositories are deprecated and are due to be permanently deleted in July 2026.
To keep a copy of this software download the repository Zip archive or clone locally using Mercurial.
It is also possible to export all your personal repositories from the account settings page.
Embed:
(wiki syntax)
Show/hide line numbers
calib3d.hpp
00001 /*M/////////////////////////////////////////////////////////////////////////////////////// 00002 // 00003 // IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING. 00004 // 00005 // By downloading, copying, installing or using the software you agree to this license. 00006 // If you do not agree to this license, do not download, install, 00007 // copy or use the software. 00008 // 00009 // 00010 // License Agreement 00011 // For Open Source Computer Vision Library 00012 // 00013 // Copyright (C) 2000-2008, Intel Corporation, all rights reserved. 00014 // Copyright (C) 2009, Willow Garage Inc., all rights reserved. 00015 // Copyright (C) 2013, OpenCV Foundation, all rights reserved. 00016 // Third party copyrights are property of their respective owners. 00017 // 00018 // Redistribution and use in source and binary forms, with or without modification, 00019 // are permitted provided that the following conditions are met: 00020 // 00021 // * Redistribution's of source code must retain the above copyright notice, 00022 // this list of conditions and the following disclaimer. 00023 // 00024 // * Redistribution's in binary form must reproduce the above copyright notice, 00025 // this list of conditions and the following disclaimer in the documentation 00026 // and/or other materials provided with the distribution. 00027 // 00028 // * The name of the copyright holders may not be used to endorse or promote products 00029 // derived from this software without specific prior written permission. 00030 // 00031 // This software is provided by the copyright holders and contributors "as is" and 00032 // any express or implied warranties, including, but not limited to, the implied 00033 // warranties of merchantability and fitness for a particular purpose are disclaimed. 00034 // In no event shall the Intel Corporation or contributors be liable for any direct, 00035 // indirect, incidental, special, exemplary, or consequential damages 00036 // (including, but not limited to, procurement of substitute goods or services; 00037 // loss of use, data, or profits; or business interruption) however caused 00038 // and on any theory of liability, whether in contract, strict liability, 00039 // or tort (including negligence or otherwise) arising in any way out of 00040 // the use of this software, even if advised of the possibility of such damage. 00041 // 00042 //M*/ 00043 00044 #ifndef __OPENCV_CALIB3D_HPP__ 00045 #define __OPENCV_CALIB3D_HPP__ 00046 00047 #include "opencv2/core.hpp" 00048 #include "opencv2/features2d.hpp" 00049 #include "opencv2/core/affine.hpp" 00050 00051 /** 00052 @defgroup calib3d Camera Calibration and 3D Reconstruction 00053 00054 The functions in this section use a so-called pinhole camera model. In this model, a scene view is 00055 formed by projecting 3D points into the image plane using a perspective transformation. 00056 00057 \f[s \; m' = A [R|t] M'\f] 00058 00059 or 00060 00061 \f[s \vecthree{u}{v}{1} = \vecthreethree{f_x}{0}{c_x}{0}{f_y}{c_y}{0}{0}{1} 00062 \begin{bmatrix} 00063 r_{11} & r_{12} & r_{13} & t_1 \\ 00064 r_{21} & r_{22} & r_{23} & t_2 \\ 00065 r_{31} & r_{32} & r_{33} & t_3 00066 \end{bmatrix} 00067 \begin{bmatrix} 00068 X \\ 00069 Y \\ 00070 Z \\ 00071 1 00072 \end{bmatrix}\f] 00073 00074 where: 00075 00076 - \f$(X, Y, Z)\f$ are the coordinates of a 3D point in the world coordinate space 00077 - \f$(u, v)\f$ are the coordinates of the projection point in pixels 00078 - \f$A\f$ is a camera matrix, or a matrix of intrinsic parameters 00079 - \f$(cx, cy)\f$ is a principal point that is usually at the image center 00080 - \f$fx, fy\f$ are the focal lengths expressed in pixel units. 00081 00082 Thus, if an image from the camera is scaled by a factor, all of these parameters should be scaled 00083 (multiplied/divided, respectively) by the same factor. The matrix of intrinsic parameters does not 00084 depend on the scene viewed. So, once estimated, it can be re-used as long as the focal length is 00085 fixed (in case of zoom lens). The joint rotation-translation matrix \f$[R|t]\f$ is called a matrix of 00086 extrinsic parameters. It is used to describe the camera motion around a static scene, or vice versa, 00087 rigid motion of an object in front of a still camera. That is, \f$[R|t]\f$ translates coordinates of a 00088 point \f$(X, Y, Z)\f$ to a coordinate system, fixed with respect to the camera. The transformation above 00089 is equivalent to the following (when \f$z \ne 0\f$ ): 00090 00091 \f[\begin{array}{l} 00092 \vecthree{x}{y}{z} = R \vecthree{X}{Y}{Z} + t \\ 00093 x' = x/z \\ 00094 y' = y/z \\ 00095 u = f_x*x' + c_x \\ 00096 v = f_y*y' + c_y 00097 \end{array}\f] 00098 00099 Real lenses usually have some distortion, mostly radial distortion and slight tangential distortion. 00100 So, the above model is extended as: 00101 00102 \f[\begin{array}{l} 00103 \vecthree{x}{y}{z} = R \vecthree{X}{Y}{Z} + t \\ 00104 x' = x/z \\ 00105 y' = y/z \\ 00106 x'' = x' \frac{1 + k_1 r^2 + k_2 r^4 + k_3 r^6}{1 + k_4 r^2 + k_5 r^4 + k_6 r^6} + 2 p_1 x' y' + p_2(r^2 + 2 x'^2) + s_1 r^2 + s_2 r^4 \\ 00107 y'' = y' \frac{1 + k_1 r^2 + k_2 r^4 + k_3 r^6}{1 + k_4 r^2 + k_5 r^4 + k_6 r^6} + p_1 (r^2 + 2 y'^2) + 2 p_2 x' y' + s_3 r^2 + s_4 r^4 \\ 00108 \text{where} \quad r^2 = x'^2 + y'^2 \\ 00109 u = f_x*x'' + c_x \\ 00110 v = f_y*y'' + c_y 00111 \end{array}\f] 00112 00113 \f$k_1\f$, \f$k_2\f$, \f$k_3\f$, \f$k_4\f$, \f$k_5\f$, and \f$k_6\f$ are radial distortion coefficients. \f$p_1\f$ and \f$p_2\f$ are 00114 tangential distortion coefficients. \f$s_1\f$, \f$s_2\f$, \f$s_3\f$, and \f$s_4\f$, are the thin prism distortion 00115 coefficients. Higher-order coefficients are not considered in OpenCV. 00116 00117 In some cases the image sensor may be tilted in order to focus an oblique plane in front of the 00118 camera (Scheimpfug condition). This can be useful for particle image velocimetry (PIV) or 00119 triangulation with a laser fan. The tilt causes a perspective distortion of \f$x''\f$ and 00120 \f$y''\f$. This distortion can be modelled in the following way, see e.g. @cite Louhichi07. 00121 00122 \f[\begin{array}{l} 00123 s\vecthree{x'''}{y'''}{1} = 00124 \vecthreethree{R_{33}(\tau_x, \tau_y)}{0}{-R_{13}(\tau_x, \tau_y)} 00125 {0}{R_{33}(\tau_x, \tau_y)}{-R_{23}(\tau_x, \tau_y)} 00126 {0}{0}{1} R(\tau_x, \tau_y) \vecthree{x''}{y''}{1}\\ 00127 u = f_x*x''' + c_x \\ 00128 v = f_y*y''' + c_y 00129 \end{array}\f] 00130 00131 where the matrix \f$R(\tau_x, \tau_y)\f$ is defined by two rotations with angular parameter \f$\tau_x\f$ 00132 and \f$\tau_y\f$, respectively, 00133 00134 \f[ 00135 R(\tau_x, \tau_y) = 00136 \vecthreethree{\cos(\tau_y)}{0}{-\sin(\tau_y)}{0}{1}{0}{\sin(\tau_y)}{0}{\cos(\tau_y)} 00137 \vecthreethree{1}{0}{0}{0}{\cos(\tau_x)}{\sin(\tau_x)}{0}{-\sin(\tau_x)}{\cos(\tau_x)} = 00138 \vecthreethree{\cos(\tau_y)}{\sin(\tau_y)\sin(\tau_x)}{-\sin(\tau_y)\cos(\tau_x)} 00139 {0}{\cos(\tau_x)}{\sin(\tau_x)} 00140 {\sin(\tau_y)}{-\cos(\tau_y)\sin(\tau_x)}{\cos(\tau_y)\cos(\tau_x)}. 00141 \f] 00142 00143 In the functions below the coefficients are passed or returned as 00144 00145 \f[(k_1, k_2, p_1, p_2[, k_3[, k_4, k_5, k_6 [, s_1, s_2, s_3, s_4[, \tau_x, \tau_y]]]])\f] 00146 00147 vector. That is, if the vector contains four elements, it means that \f$k_3=0\f$ . The distortion 00148 coefficients do not depend on the scene viewed. Thus, they also belong to the intrinsic camera 00149 parameters. And they remain the same regardless of the captured image resolution. If, for example, a 00150 camera has been calibrated on images of 320 x 240 resolution, absolutely the same distortion 00151 coefficients can be used for 640 x 480 images from the same camera while \f$f_x\f$, \f$f_y\f$, \f$c_x\f$, and 00152 \f$c_y\f$ need to be scaled appropriately. 00153 00154 The functions below use the above model to do the following: 00155 00156 - Project 3D points to the image plane given intrinsic and extrinsic parameters. 00157 - Compute extrinsic parameters given intrinsic parameters, a few 3D points, and their 00158 projections. 00159 - Estimate intrinsic and extrinsic camera parameters from several views of a known calibration 00160 pattern (every view is described by several 3D-2D point correspondences). 00161 - Estimate the relative position and orientation of the stereo camera "heads" and compute the 00162 *rectification* transformation that makes the camera optical axes parallel. 00163 00164 @note 00165 - A calibration sample for 3 cameras in horizontal position can be found at 00166 opencv_source_code/samples/cpp/3calibration.cpp 00167 - A calibration sample based on a sequence of images can be found at 00168 opencv_source_code/samples/cpp/calibration.cpp 00169 - A calibration sample in order to do 3D reconstruction can be found at 00170 opencv_source_code/samples/cpp/build3dmodel.cpp 00171 - A calibration sample of an artificially generated camera and chessboard patterns can be 00172 found at opencv_source_code/samples/cpp/calibration_artificial.cpp 00173 - A calibration example on stereo calibration can be found at 00174 opencv_source_code/samples/cpp/stereo_calib.cpp 00175 - A calibration example on stereo matching can be found at 00176 opencv_source_code/samples/cpp/stereo_match.cpp 00177 - (Python) A camera calibration sample can be found at 00178 opencv_source_code/samples/python/calibrate.py 00179 00180 @{ 00181 @defgroup calib3d_fisheye Fisheye camera model 00182 00183 Definitions: Let P be a point in 3D of coordinates X in the world reference frame (stored in the 00184 matrix X) The coordinate vector of P in the camera reference frame is: 00185 00186 \f[Xc = R X + T\f] 00187 00188 where R is the rotation matrix corresponding to the rotation vector om: R = rodrigues(om); call x, y 00189 and z the 3 coordinates of Xc: 00190 00191 \f[x = Xc_1 \\ y = Xc_2 \\ z = Xc_3\f] 00192 00193 The pinehole projection coordinates of P is [a; b] where 00194 00195 \f[a = x / z \ and \ b = y / z \\ r^2 = a^2 + b^2 \\ \theta = atan(r)\f] 00196 00197 Fisheye distortion: 00198 00199 \f[\theta_d = \theta (1 + k_1 \theta^2 + k_2 \theta^4 + k_3 \theta^6 + k_4 \theta^8)\f] 00200 00201 The distorted point coordinates are [x'; y'] where 00202 00203 \f[x' = (\theta_d / r) x \\ y' = (\theta_d / r) y \f] 00204 00205 Finally, conversion into pixel coordinates: The final pixel coordinates vector [u; v] where: 00206 00207 \f[u = f_x (x' + \alpha y') + c_x \\ 00208 v = f_y yy + c_y\f] 00209 00210 @defgroup calib3d_c C API 00211 00212 @} 00213 */ 00214 00215 namespace cv 00216 { 00217 00218 //! @addtogroup calib3d 00219 //! @{ 00220 00221 //! type of the robust estimation algorithm 00222 enum { LMEDS = 4, //!< least-median algorithm 00223 RANSAC = 8, //!< RANSAC algorithm 00224 RHO = 16 //!< RHO algorithm 00225 }; 00226 00227 enum { SOLVEPNP_ITERATIVE = 0, 00228 SOLVEPNP_EPNP = 1, //!< EPnP: Efficient Perspective-n-Point Camera Pose Estimation @cite lepetit2009epnp 00229 SOLVEPNP_P3P = 2, //!< Complete Solution Classification for the Perspective-Three-Point Problem @cite gao2003complete 00230 SOLVEPNP_DLS = 3, //!< A Direct Least-Squares (DLS) Method for PnP @cite hesch2011direct 00231 SOLVEPNP_UPNP = 4 //!< Exhaustive Linearization for Robust Camera Pose and Focal Length Estimation @cite penate2013exhaustive 00232 00233 }; 00234 00235 enum { CALIB_CB_ADAPTIVE_THRESH = 1, 00236 CALIB_CB_NORMALIZE_IMAGE = 2, 00237 CALIB_CB_FILTER_QUADS = 4, 00238 CALIB_CB_FAST_CHECK = 8 00239 }; 00240 00241 enum { CALIB_CB_SYMMETRIC_GRID = 1, 00242 CALIB_CB_ASYMMETRIC_GRID = 2, 00243 CALIB_CB_CLUSTERING = 4 00244 }; 00245 00246 enum { CALIB_USE_INTRINSIC_GUESS = 0x00001, 00247 CALIB_FIX_ASPECT_RATIO = 0x00002, 00248 CALIB_FIX_PRINCIPAL_POINT = 0x00004, 00249 CALIB_ZERO_TANGENT_DIST = 0x00008, 00250 CALIB_FIX_FOCAL_LENGTH = 0x00010, 00251 CALIB_FIX_K1 = 0x00020, 00252 CALIB_FIX_K2 = 0x00040, 00253 CALIB_FIX_K3 = 0x00080, 00254 CALIB_FIX_K4 = 0x00800, 00255 CALIB_FIX_K5 = 0x01000, 00256 CALIB_FIX_K6 = 0x02000, 00257 CALIB_RATIONAL_MODEL = 0x04000, 00258 CALIB_THIN_PRISM_MODEL = 0x08000, 00259 CALIB_FIX_S1_S2_S3_S4 = 0x10000, 00260 CALIB_TILTED_MODEL = 0x40000, 00261 CALIB_FIX_TAUX_TAUY = 0x80000, 00262 // only for stereo 00263 CALIB_FIX_INTRINSIC = 0x00100, 00264 CALIB_SAME_FOCAL_LENGTH = 0x00200, 00265 // for stereo rectification 00266 CALIB_ZERO_DISPARITY = 0x00400, 00267 CALIB_USE_LU = (1 << 17), //!< use LU instead of SVD decomposition for solving. much faster but potentially less precise 00268 }; 00269 00270 //! the algorithm for finding fundamental matrix 00271 enum { FM_7POINT = 1, //!< 7-point algorithm 00272 FM_8POINT = 2, //!< 8-point algorithm 00273 FM_LMEDS = 4, //!< least-median algorithm 00274 FM_RANSAC = 8 //!< RANSAC algorithm 00275 }; 00276 00277 00278 00279 /** @brief Converts a rotation matrix to a rotation vector or vice versa. 00280 00281 @param src Input rotation vector (3x1 or 1x3) or rotation matrix (3x3). 00282 @param dst Output rotation matrix (3x3) or rotation vector (3x1 or 1x3), respectively. 00283 @param jacobian Optional output Jacobian matrix, 3x9 or 9x3, which is a matrix of partial 00284 derivatives of the output array components with respect to the input array components. 00285 00286 \f[\begin{array}{l} \theta \leftarrow norm(r) \\ r \leftarrow r/ \theta \\ R = \cos{\theta} I + (1- \cos{\theta} ) r r^T + \sin{\theta} \vecthreethree{0}{-r_z}{r_y}{r_z}{0}{-r_x}{-r_y}{r_x}{0} \end{array}\f] 00287 00288 Inverse transformation can be also done easily, since 00289 00290 \f[\sin ( \theta ) \vecthreethree{0}{-r_z}{r_y}{r_z}{0}{-r_x}{-r_y}{r_x}{0} = \frac{R - R^T}{2}\f] 00291 00292 A rotation vector is a convenient and most compact representation of a rotation matrix (since any 00293 rotation matrix has just 3 degrees of freedom). The representation is used in the global 3D geometry 00294 optimization procedures like calibrateCamera, stereoCalibrate, or solvePnP . 00295 */ 00296 CV_EXPORTS_W void Rodrigues( InputArray src, OutputArray dst, OutputArray jacobian = noArray() ); 00297 00298 /** @brief Finds a perspective transformation between two planes. 00299 00300 @param srcPoints Coordinates of the points in the original plane, a matrix of the type CV_32FC2 00301 or vector<Point2f> . 00302 @param dstPoints Coordinates of the points in the target plane, a matrix of the type CV_32FC2 or 00303 a vector<Point2f> . 00304 @param method Method used to computed a homography matrix. The following methods are possible: 00305 - **0** - a regular method using all the points 00306 - **RANSAC** - RANSAC-based robust method 00307 - **LMEDS** - Least-Median robust method 00308 - **RHO** - PROSAC-based robust method 00309 @param ransacReprojThreshold Maximum allowed reprojection error to treat a point pair as an inlier 00310 (used in the RANSAC and RHO methods only). That is, if 00311 \f[\| \texttt{dstPoints} _i - \texttt{convertPointsHomogeneous} ( \texttt{H} * \texttt{srcPoints} _i) \| > \texttt{ransacReprojThreshold}\f] 00312 then the point \f$i\f$ is considered an outlier. If srcPoints and dstPoints are measured in pixels, 00313 it usually makes sense to set this parameter somewhere in the range of 1 to 10. 00314 @param mask Optional output mask set by a robust method ( RANSAC or LMEDS ). Note that the input 00315 mask values are ignored. 00316 @param maxIters The maximum number of RANSAC iterations, 2000 is the maximum it can be. 00317 @param confidence Confidence level, between 0 and 1. 00318 00319 The functions find and return the perspective transformation \f$H\f$ between the source and the 00320 destination planes: 00321 00322 \f[s_i \vecthree{x'_i}{y'_i}{1} \sim H \vecthree{x_i}{y_i}{1}\f] 00323 00324 so that the back-projection error 00325 00326 \f[\sum _i \left ( x'_i- \frac{h_{11} x_i + h_{12} y_i + h_{13}}{h_{31} x_i + h_{32} y_i + h_{33}} \right )^2+ \left ( y'_i- \frac{h_{21} x_i + h_{22} y_i + h_{23}}{h_{31} x_i + h_{32} y_i + h_{33}} \right )^2\f] 00327 00328 is minimized. If the parameter method is set to the default value 0, the function uses all the point 00329 pairs to compute an initial homography estimate with a simple least-squares scheme. 00330 00331 However, if not all of the point pairs ( \f$srcPoints_i\f$, \f$dstPoints_i\f$ ) fit the rigid perspective 00332 transformation (that is, there are some outliers), this initial estimate will be poor. In this case, 00333 you can use one of the three robust methods. The methods RANSAC, LMeDS and RHO try many different 00334 random subsets of the corresponding point pairs (of four pairs each), estimate the homography matrix 00335 using this subset and a simple least-square algorithm, and then compute the quality/goodness of the 00336 computed homography (which is the number of inliers for RANSAC or the median re-projection error for 00337 LMeDs). The best subset is then used to produce the initial estimate of the homography matrix and 00338 the mask of inliers/outliers. 00339 00340 Regardless of the method, robust or not, the computed homography matrix is refined further (using 00341 inliers only in case of a robust method) with the Levenberg-Marquardt method to reduce the 00342 re-projection error even more. 00343 00344 The methods RANSAC and RHO can handle practically any ratio of outliers but need a threshold to 00345 distinguish inliers from outliers. The method LMeDS does not need any threshold but it works 00346 correctly only when there are more than 50% of inliers. Finally, if there are no outliers and the 00347 noise is rather small, use the default method (method=0). 00348 00349 The function is used to find initial intrinsic and extrinsic matrices. Homography matrix is 00350 determined up to a scale. Thus, it is normalized so that \f$h_{33}=1\f$. Note that whenever an H matrix 00351 cannot be estimated, an empty one will be returned. 00352 00353 @sa 00354 getAffineTransform, getPerspectiveTransform, estimateRigidTransform, warpPerspective, 00355 perspectiveTransform 00356 00357 @note 00358 - A example on calculating a homography for image matching can be found at 00359 opencv_source_code/samples/cpp/video_homography.cpp 00360 00361 */ 00362 CV_EXPORTS_W Mat findHomography( InputArray srcPoints, InputArray dstPoints, 00363 int method = 0, double ransacReprojThreshold = 3, 00364 OutputArray mask=noArray(), const int maxIters = 2000, 00365 const double confidence = 0.995); 00366 00367 /** @overload */ 00368 CV_EXPORTS Mat findHomography( InputArray srcPoints, InputArray dstPoints, 00369 OutputArray mask, int method = 0, double ransacReprojThreshold = 3 ); 00370 00371 /** @brief Computes an RQ decomposition of 3x3 matrices. 00372 00373 @param src 3x3 input matrix. 00374 @param mtxR Output 3x3 upper-triangular matrix. 00375 @param mtxQ Output 3x3 orthogonal matrix. 00376 @param Qx Optional output 3x3 rotation matrix around x-axis. 00377 @param Qy Optional output 3x3 rotation matrix around y-axis. 00378 @param Qz Optional output 3x3 rotation matrix around z-axis. 00379 00380 The function computes a RQ decomposition using the given rotations. This function is used in 00381 decomposeProjectionMatrix to decompose the left 3x3 submatrix of a projection matrix into a camera 00382 and a rotation matrix. 00383 00384 It optionally returns three rotation matrices, one for each axis, and the three Euler angles in 00385 degrees (as the return value) that could be used in OpenGL. Note, there is always more than one 00386 sequence of rotations about the three principle axes that results in the same orientation of an 00387 object, eg. see @cite Slabaugh . Returned tree rotation matrices and corresponding three Euler angules 00388 are only one of the possible solutions. 00389 */ 00390 CV_EXPORTS_W Vec3d RQDecomp3x3( InputArray src, OutputArray mtxR, OutputArray mtxQ, 00391 OutputArray Qx = noArray(), 00392 OutputArray Qy = noArray(), 00393 OutputArray Qz = noArray()); 00394 00395 /** @brief Decomposes a projection matrix into a rotation matrix and a camera matrix. 00396 00397 @param projMatrix 3x4 input projection matrix P. 00398 @param cameraMatrix Output 3x3 camera matrix K. 00399 @param rotMatrix Output 3x3 external rotation matrix R. 00400 @param transVect Output 4x1 translation vector T. 00401 @param rotMatrixX Optional 3x3 rotation matrix around x-axis. 00402 @param rotMatrixY Optional 3x3 rotation matrix around y-axis. 00403 @param rotMatrixZ Optional 3x3 rotation matrix around z-axis. 00404 @param eulerAngles Optional three-element vector containing three Euler angles of rotation in 00405 degrees. 00406 00407 The function computes a decomposition of a projection matrix into a calibration and a rotation 00408 matrix and the position of a camera. 00409 00410 It optionally returns three rotation matrices, one for each axis, and three Euler angles that could 00411 be used in OpenGL. Note, there is always more than one sequence of rotations about the three 00412 principle axes that results in the same orientation of an object, eg. see @cite Slabaugh . Returned 00413 tree rotation matrices and corresponding three Euler angules are only one of the possible solutions. 00414 00415 The function is based on RQDecomp3x3 . 00416 */ 00417 CV_EXPORTS_W void decomposeProjectionMatrix( InputArray projMatrix, OutputArray cameraMatrix, 00418 OutputArray rotMatrix, OutputArray transVect, 00419 OutputArray rotMatrixX = noArray(), 00420 OutputArray rotMatrixY = noArray(), 00421 OutputArray rotMatrixZ = noArray(), 00422 OutputArray eulerAngles =noArray() ); 00423 00424 /** @brief Computes partial derivatives of the matrix product for each multiplied matrix. 00425 00426 @param A First multiplied matrix. 00427 @param B Second multiplied matrix. 00428 @param dABdA First output derivative matrix d(A\*B)/dA of size 00429 \f$\texttt{A.rows*B.cols} \times {A.rows*A.cols}\f$ . 00430 @param dABdB Second output derivative matrix d(A\*B)/dB of size 00431 \f$\texttt{A.rows*B.cols} \times {B.rows*B.cols}\f$ . 00432 00433 The function computes partial derivatives of the elements of the matrix product \f$A*B\f$ with regard to 00434 the elements of each of the two input matrices. The function is used to compute the Jacobian 00435 matrices in stereoCalibrate but can also be used in any other similar optimization function. 00436 */ 00437 CV_EXPORTS_W void matMulDeriv( InputArray A, InputArray B, OutputArray dABdA, OutputArray dABdB ); 00438 00439 /** @brief Combines two rotation-and-shift transformations. 00440 00441 @param rvec1 First rotation vector. 00442 @param tvec1 First translation vector. 00443 @param rvec2 Second rotation vector. 00444 @param tvec2 Second translation vector. 00445 @param rvec3 Output rotation vector of the superposition. 00446 @param tvec3 Output translation vector of the superposition. 00447 @param dr3dr1 00448 @param dr3dt1 00449 @param dr3dr2 00450 @param dr3dt2 00451 @param dt3dr1 00452 @param dt3dt1 00453 @param dt3dr2 00454 @param dt3dt2 Optional output derivatives of rvec3 or tvec3 with regard to rvec1, rvec2, tvec1 and 00455 tvec2, respectively. 00456 00457 The functions compute: 00458 00459 \f[\begin{array}{l} \texttt{rvec3} = \mathrm{rodrigues} ^{-1} \left ( \mathrm{rodrigues} ( \texttt{rvec2} ) \cdot \mathrm{rodrigues} ( \texttt{rvec1} ) \right ) \\ \texttt{tvec3} = \mathrm{rodrigues} ( \texttt{rvec2} ) \cdot \texttt{tvec1} + \texttt{tvec2} \end{array} ,\f] 00460 00461 where \f$\mathrm{rodrigues}\f$ denotes a rotation vector to a rotation matrix transformation, and 00462 \f$\mathrm{rodrigues}^{-1}\f$ denotes the inverse transformation. See Rodrigues for details. 00463 00464 Also, the functions can compute the derivatives of the output vectors with regards to the input 00465 vectors (see matMulDeriv ). The functions are used inside stereoCalibrate but can also be used in 00466 your own code where Levenberg-Marquardt or another gradient-based solver is used to optimize a 00467 function that contains a matrix multiplication. 00468 */ 00469 CV_EXPORTS_W void composeRT( InputArray rvec1, InputArray tvec1, 00470 InputArray rvec2, InputArray tvec2, 00471 OutputArray rvec3, OutputArray tvec3, 00472 OutputArray dr3dr1 = noArray(), OutputArray dr3dt1 = noArray(), 00473 OutputArray dr3dr2 = noArray(), OutputArray dr3dt2 = noArray(), 00474 OutputArray dt3dr1 = noArray(), OutputArray dt3dt1 = noArray(), 00475 OutputArray dt3dr2 = noArray(), OutputArray dt3dt2 = noArray() ); 00476 00477 /** @brief Projects 3D points to an image plane. 00478 00479 @param objectPoints Array of object points, 3xN/Nx3 1-channel or 1xN/Nx1 3-channel (or 00480 vector<Point3f> ), where N is the number of points in the view. 00481 @param rvec Rotation vector. See Rodrigues for details. 00482 @param tvec Translation vector. 00483 @param cameraMatrix Camera matrix \f$A = \vecthreethree{f_x}{0}{c_x}{0}{f_y}{c_y}{0}{0}{_1}\f$ . 00484 @param distCoeffs Input vector of distortion coefficients 00485 \f$(k_1, k_2, p_1, p_2[, k_3[, k_4, k_5, k_6 [, s_1, s_2, s_3, s_4[, \tau_x, \tau_y]]]])\f$ of 00486 4, 5, 8, 12 or 14 elements. If the vector is empty, the zero distortion coefficients are assumed. 00487 @param imagePoints Output array of image points, 2xN/Nx2 1-channel or 1xN/Nx1 2-channel, or 00488 vector<Point2f> . 00489 @param jacobian Optional output 2Nx(10+<numDistCoeffs>) jacobian matrix of derivatives of image 00490 points with respect to components of the rotation vector, translation vector, focal lengths, 00491 coordinates of the principal point and the distortion coefficients. In the old interface different 00492 components of the jacobian are returned via different output parameters. 00493 @param aspectRatio Optional "fixed aspect ratio" parameter. If the parameter is not 0, the 00494 function assumes that the aspect ratio (*fx/fy*) is fixed and correspondingly adjusts the jacobian 00495 matrix. 00496 00497 The function computes projections of 3D points to the image plane given intrinsic and extrinsic 00498 camera parameters. Optionally, the function computes Jacobians - matrices of partial derivatives of 00499 image points coordinates (as functions of all the input parameters) with respect to the particular 00500 parameters, intrinsic and/or extrinsic. The Jacobians are used during the global optimization in 00501 calibrateCamera, solvePnP, and stereoCalibrate . The function itself can also be used to compute a 00502 re-projection error given the current intrinsic and extrinsic parameters. 00503 00504 @note By setting rvec=tvec=(0,0,0) or by setting cameraMatrix to a 3x3 identity matrix, or by 00505 passing zero distortion coefficients, you can get various useful partial cases of the function. This 00506 means that you can compute the distorted coordinates for a sparse set of points or apply a 00507 perspective transformation (and also compute the derivatives) in the ideal zero-distortion setup. 00508 */ 00509 CV_EXPORTS_W void projectPoints( InputArray objectPoints, 00510 InputArray rvec, InputArray tvec, 00511 InputArray cameraMatrix, InputArray distCoeffs, 00512 OutputArray imagePoints, 00513 OutputArray jacobian = noArray(), 00514 double aspectRatio = 0 ); 00515 00516 /** @brief Finds an object pose from 3D-2D point correspondences. 00517 00518 @param objectPoints Array of object points in the object coordinate space, 3xN/Nx3 1-channel or 00519 1xN/Nx1 3-channel, where N is the number of points. vector<Point3f> can be also passed here. 00520 @param imagePoints Array of corresponding image points, 2xN/Nx2 1-channel or 1xN/Nx1 2-channel, 00521 where N is the number of points. vector<Point2f> can be also passed here. 00522 @param cameraMatrix Input camera matrix \f$A = \vecthreethree{fx}{0}{cx}{0}{fy}{cy}{0}{0}{1}\f$ . 00523 @param distCoeffs Input vector of distortion coefficients 00524 \f$(k_1, k_2, p_1, p_2[, k_3[, k_4, k_5, k_6 [, s_1, s_2, s_3, s_4[, \tau_x, \tau_y]]]])\f$ of 00525 4, 5, 8, 12 or 14 elements. If the vector is NULL/empty, the zero distortion coefficients are 00526 assumed. 00527 @param rvec Output rotation vector (see Rodrigues ) that, together with tvec , brings points from 00528 the model coordinate system to the camera coordinate system. 00529 @param tvec Output translation vector. 00530 @param useExtrinsicGuess Parameter used for SOLVEPNP_ITERATIVE. If true (1), the function uses 00531 the provided rvec and tvec values as initial approximations of the rotation and translation 00532 vectors, respectively, and further optimizes them. 00533 @param flags Method for solving a PnP problem: 00534 - **SOLVEPNP_ITERATIVE** Iterative method is based on Levenberg-Marquardt optimization. In 00535 this case the function finds such a pose that minimizes reprojection error, that is the sum 00536 of squared distances between the observed projections imagePoints and the projected (using 00537 projectPoints ) objectPoints . 00538 - **SOLVEPNP_P3P** Method is based on the paper of X.S. Gao, X.-R. Hou, J. Tang, H.-F. Chang 00539 "Complete Solution Classification for the Perspective-Three-Point Problem". In this case the 00540 function requires exactly four object and image points. 00541 - **SOLVEPNP_EPNP** Method has been introduced by F.Moreno-Noguer, V.Lepetit and P.Fua in the 00542 paper "EPnP: Efficient Perspective-n-Point Camera Pose Estimation". 00543 - **SOLVEPNP_DLS** Method is based on the paper of Joel A. Hesch and Stergios I. Roumeliotis. 00544 "A Direct Least-Squares (DLS) Method for PnP". 00545 - **SOLVEPNP_UPNP** Method is based on the paper of A.Penate-Sanchez, J.Andrade-Cetto, 00546 F.Moreno-Noguer. "Exhaustive Linearization for Robust Camera Pose and Focal Length 00547 Estimation". In this case the function also estimates the parameters \f$f_x\f$ and \f$f_y\f$ 00548 assuming that both have the same value. Then the cameraMatrix is updated with the estimated 00549 focal length. 00550 00551 The function estimates the object pose given a set of object points, their corresponding image 00552 projections, as well as the camera matrix and the distortion coefficients. 00553 00554 @note 00555 - An example of how to use solvePnP for planar augmented reality can be found at 00556 opencv_source_code/samples/python/plane_ar.py 00557 - If you are using Python: 00558 - Numpy array slices won't work as input because solvePnP requires contiguous 00559 arrays (enforced by the assertion using cv::Mat::checkVector() around line 55 of 00560 modules/calib3d/src/solvepnp.cpp version 2.4.9) 00561 - The P3P algorithm requires image points to be in an array of shape (N,1,2) due 00562 to its calling of cv::undistortPoints (around line 75 of modules/calib3d/src/solvepnp.cpp version 2.4.9) 00563 which requires 2-channel information. 00564 - Thus, given some data D = np.array(...) where D.shape = (N,M), in order to use a subset of 00565 it as, e.g., imagePoints, one must effectively copy it into a new array: imagePoints = 00566 np.ascontiguousarray(D[:,:2]).reshape((N,1,2)) 00567 */ 00568 CV_EXPORTS_W bool solvePnP( InputArray objectPoints, InputArray imagePoints, 00569 InputArray cameraMatrix, InputArray distCoeffs, 00570 OutputArray rvec, OutputArray tvec, 00571 bool useExtrinsicGuess = false, int flags = SOLVEPNP_ITERATIVE ); 00572 00573 /** @brief Finds an object pose from 3D-2D point correspondences using the RANSAC scheme. 00574 00575 @param objectPoints Array of object points in the object coordinate space, 3xN/Nx3 1-channel or 00576 1xN/Nx1 3-channel, where N is the number of points. vector<Point3f> can be also passed here. 00577 @param imagePoints Array of corresponding image points, 2xN/Nx2 1-channel or 1xN/Nx1 2-channel, 00578 where N is the number of points. vector<Point2f> can be also passed here. 00579 @param cameraMatrix Input camera matrix \f$A = \vecthreethree{fx}{0}{cx}{0}{fy}{cy}{0}{0}{1}\f$ . 00580 @param distCoeffs Input vector of distortion coefficients 00581 \f$(k_1, k_2, p_1, p_2[, k_3[, k_4, k_5, k_6 [, s_1, s_2, s_3, s_4[, \tau_x, \tau_y]]]])\f$ of 00582 4, 5, 8, 12 or 14 elements. If the vector is NULL/empty, the zero distortion coefficients are 00583 assumed. 00584 @param rvec Output rotation vector (see Rodrigues ) that, together with tvec , brings points from 00585 the model coordinate system to the camera coordinate system. 00586 @param tvec Output translation vector. 00587 @param useExtrinsicGuess Parameter used for SOLVEPNP_ITERATIVE. If true (1), the function uses 00588 the provided rvec and tvec values as initial approximations of the rotation and translation 00589 vectors, respectively, and further optimizes them. 00590 @param iterationsCount Number of iterations. 00591 @param reprojectionError Inlier threshold value used by the RANSAC procedure. The parameter value 00592 is the maximum allowed distance between the observed and computed point projections to consider it 00593 an inlier. 00594 @param confidence The probability that the algorithm produces a useful result. 00595 @param inliers Output vector that contains indices of inliers in objectPoints and imagePoints . 00596 @param flags Method for solving a PnP problem (see solvePnP ). 00597 00598 The function estimates an object pose given a set of object points, their corresponding image 00599 projections, as well as the camera matrix and the distortion coefficients. This function finds such 00600 a pose that minimizes reprojection error, that is, the sum of squared distances between the observed 00601 projections imagePoints and the projected (using projectPoints ) objectPoints. The use of RANSAC 00602 makes the function resistant to outliers. 00603 00604 @note 00605 - An example of how to use solvePNPRansac for object detection can be found at 00606 opencv_source_code/samples/cpp/tutorial_code/calib3d/real_time_pose_estimation/ 00607 */ 00608 CV_EXPORTS_W bool solvePnPRansac( InputArray objectPoints, InputArray imagePoints, 00609 InputArray cameraMatrix, InputArray distCoeffs, 00610 OutputArray rvec, OutputArray tvec, 00611 bool useExtrinsicGuess = false, int iterationsCount = 100, 00612 float reprojectionError = 8.0, double confidence = 0.99, 00613 OutputArray inliers = noArray(), int flags = SOLVEPNP_ITERATIVE ); 00614 00615 /** @brief Finds an initial camera matrix from 3D-2D point correspondences. 00616 00617 @param objectPoints Vector of vectors of the calibration pattern points in the calibration pattern 00618 coordinate space. In the old interface all the per-view vectors are concatenated. See 00619 calibrateCamera for details. 00620 @param imagePoints Vector of vectors of the projections of the calibration pattern points. In the 00621 old interface all the per-view vectors are concatenated. 00622 @param imageSize Image size in pixels used to initialize the principal point. 00623 @param aspectRatio If it is zero or negative, both \f$f_x\f$ and \f$f_y\f$ are estimated independently. 00624 Otherwise, \f$f_x = f_y * \texttt{aspectRatio}\f$ . 00625 00626 The function estimates and returns an initial camera matrix for the camera calibration process. 00627 Currently, the function only supports planar calibration patterns, which are patterns where each 00628 object point has z-coordinate =0. 00629 */ 00630 CV_EXPORTS_W Mat initCameraMatrix2D( InputArrayOfArrays objectPoints, 00631 InputArrayOfArrays imagePoints, 00632 Size imageSize, double aspectRatio = 1.0 ); 00633 00634 /** @brief Finds the positions of internal corners of the chessboard. 00635 00636 @param image Source chessboard view. It must be an 8-bit grayscale or color image. 00637 @param patternSize Number of inner corners per a chessboard row and column 00638 ( patternSize = cvSize(points_per_row,points_per_colum) = cvSize(columns,rows) ). 00639 @param corners Output array of detected corners. 00640 @param flags Various operation flags that can be zero or a combination of the following values: 00641 - **CV_CALIB_CB_ADAPTIVE_THRESH** Use adaptive thresholding to convert the image to black 00642 and white, rather than a fixed threshold level (computed from the average image brightness). 00643 - **CV_CALIB_CB_NORMALIZE_IMAGE** Normalize the image gamma with equalizeHist before 00644 applying fixed or adaptive thresholding. 00645 - **CV_CALIB_CB_FILTER_QUADS** Use additional criteria (like contour area, perimeter, 00646 square-like shape) to filter out false quads extracted at the contour retrieval stage. 00647 - **CALIB_CB_FAST_CHECK** Run a fast check on the image that looks for chessboard corners, 00648 and shortcut the call if none is found. This can drastically speed up the call in the 00649 degenerate condition when no chessboard is observed. 00650 00651 The function attempts to determine whether the input image is a view of the chessboard pattern and 00652 locate the internal chessboard corners. The function returns a non-zero value if all of the corners 00653 are found and they are placed in a certain order (row by row, left to right in every row). 00654 Otherwise, if the function fails to find all the corners or reorder them, it returns 0. For example, 00655 a regular chessboard has 8 x 8 squares and 7 x 7 internal corners, that is, points where the black 00656 squares touch each other. The detected coordinates are approximate, and to determine their positions 00657 more accurately, the function calls cornerSubPix. You also may use the function cornerSubPix with 00658 different parameters if returned coordinates are not accurate enough. 00659 00660 Sample usage of detecting and drawing chessboard corners: : 00661 @code 00662 Size patternsize(8,6); //interior number of corners 00663 Mat gray = ....; //source image 00664 vector<Point2f> corners; //this will be filled by the detected corners 00665 00666 //CALIB_CB_FAST_CHECK saves a lot of time on images 00667 //that do not contain any chessboard corners 00668 bool patternfound = findChessboardCorners(gray, patternsize, corners, 00669 CALIB_CB_ADAPTIVE_THRESH + CALIB_CB_NORMALIZE_IMAGE 00670 + CALIB_CB_FAST_CHECK); 00671 00672 if(patternfound) 00673 cornerSubPix(gray, corners, Size(11, 11), Size(-1, -1), 00674 TermCriteria(CV_TERMCRIT_EPS + CV_TERMCRIT_ITER, 30, 0.1)); 00675 00676 drawChessboardCorners(img, patternsize, Mat(corners), patternfound); 00677 @endcode 00678 @note The function requires white space (like a square-thick border, the wider the better) around 00679 the board to make the detection more robust in various environments. Otherwise, if there is no 00680 border and the background is dark, the outer black squares cannot be segmented properly and so the 00681 square grouping and ordering algorithm fails. 00682 */ 00683 CV_EXPORTS_W bool findChessboardCorners( InputArray image, Size patternSize, OutputArray corners, 00684 int flags = CALIB_CB_ADAPTIVE_THRESH + CALIB_CB_NORMALIZE_IMAGE ); 00685 00686 //! finds subpixel-accurate positions of the chessboard corners 00687 CV_EXPORTS bool find4QuadCornerSubpix( InputArray img, InputOutputArray corners, Size region_size ); 00688 00689 /** @brief Renders the detected chessboard corners. 00690 00691 @param image Destination image. It must be an 8-bit color image. 00692 @param patternSize Number of inner corners per a chessboard row and column 00693 (patternSize = cv::Size(points_per_row,points_per_column)). 00694 @param corners Array of detected corners, the output of findChessboardCorners. 00695 @param patternWasFound Parameter indicating whether the complete board was found or not. The 00696 return value of findChessboardCorners should be passed here. 00697 00698 The function draws individual chessboard corners detected either as red circles if the board was not 00699 found, or as colored corners connected with lines if the board was found. 00700 */ 00701 CV_EXPORTS_W void drawChessboardCorners( InputOutputArray image, Size patternSize, 00702 InputArray corners, bool patternWasFound ); 00703 00704 /** @brief Finds centers in the grid of circles. 00705 00706 @param image grid view of input circles; it must be an 8-bit grayscale or color image. 00707 @param patternSize number of circles per row and column 00708 ( patternSize = Size(points_per_row, points_per_colum) ). 00709 @param centers output array of detected centers. 00710 @param flags various operation flags that can be one of the following values: 00711 - **CALIB_CB_SYMMETRIC_GRID** uses symmetric pattern of circles. 00712 - **CALIB_CB_ASYMMETRIC_GRID** uses asymmetric pattern of circles. 00713 - **CALIB_CB_CLUSTERING** uses a special algorithm for grid detection. It is more robust to 00714 perspective distortions but much more sensitive to background clutter. 00715 @param blobDetector feature detector that finds blobs like dark circles on light background. 00716 00717 The function attempts to determine whether the input image contains a grid of circles. If it is, the 00718 function locates centers of the circles. The function returns a non-zero value if all of the centers 00719 have been found and they have been placed in a certain order (row by row, left to right in every 00720 row). Otherwise, if the function fails to find all the corners or reorder them, it returns 0. 00721 00722 Sample usage of detecting and drawing the centers of circles: : 00723 @code 00724 Size patternsize(7,7); //number of centers 00725 Mat gray = ....; //source image 00726 vector<Point2f> centers; //this will be filled by the detected centers 00727 00728 bool patternfound = findCirclesGrid(gray, patternsize, centers); 00729 00730 drawChessboardCorners(img, patternsize, Mat(centers), patternfound); 00731 @endcode 00732 @note The function requires white space (like a square-thick border, the wider the better) around 00733 the board to make the detection more robust in various environments. 00734 */ 00735 CV_EXPORTS_W bool findCirclesGrid( InputArray image, Size patternSize, 00736 OutputArray centers, int flags = CALIB_CB_SYMMETRIC_GRID, 00737 const Ptr<FeatureDetector> &blobDetector = SimpleBlobDetector::create()); 00738 00739 /** @brief Finds the camera intrinsic and extrinsic parameters from several views of a calibration pattern. 00740 00741 @param objectPoints In the new interface it is a vector of vectors of calibration pattern points in 00742 the calibration pattern coordinate space (e.g. std::vector<std::vector<cv::Vec3f>>). The outer 00743 vector contains as many elements as the number of the pattern views. If the same calibration pattern 00744 is shown in each view and it is fully visible, all the vectors will be the same. Although, it is 00745 possible to use partially occluded patterns, or even different patterns in different views. Then, 00746 the vectors will be different. The points are 3D, but since they are in a pattern coordinate system, 00747 then, if the rig is planar, it may make sense to put the model to a XY coordinate plane so that 00748 Z-coordinate of each input object point is 0. 00749 In the old interface all the vectors of object points from different views are concatenated 00750 together. 00751 @param imagePoints In the new interface it is a vector of vectors of the projections of calibration 00752 pattern points (e.g. std::vector<std::vector<cv::Vec2f>>). imagePoints.size() and 00753 objectPoints.size() and imagePoints[i].size() must be equal to objectPoints[i].size() for each i. 00754 In the old interface all the vectors of object points from different views are concatenated 00755 together. 00756 @param imageSize Size of the image used only to initialize the intrinsic camera matrix. 00757 @param cameraMatrix Output 3x3 floating-point camera matrix 00758 \f$A = \vecthreethree{f_x}{0}{c_x}{0}{f_y}{c_y}{0}{0}{1}\f$ . If CV\_CALIB\_USE\_INTRINSIC\_GUESS 00759 and/or CV_CALIB_FIX_ASPECT_RATIO are specified, some or all of fx, fy, cx, cy must be 00760 initialized before calling the function. 00761 @param distCoeffs Output vector of distortion coefficients 00762 \f$(k_1, k_2, p_1, p_2[, k_3[, k_4, k_5, k_6 [, s_1, s_2, s_3, s_4[, \tau_x, \tau_y]]]])\f$ of 00763 4, 5, 8, 12 or 14 elements. 00764 @param rvecs Output vector of rotation vectors (see Rodrigues ) estimated for each pattern view 00765 (e.g. std::vector<cv::Mat>>). That is, each k-th rotation vector together with the corresponding 00766 k-th translation vector (see the next output parameter description) brings the calibration pattern 00767 from the model coordinate space (in which object points are specified) to the world coordinate 00768 space, that is, a real position of the calibration pattern in the k-th pattern view (k=0.. *M* -1). 00769 @param tvecs Output vector of translation vectors estimated for each pattern view. 00770 @param flags Different flags that may be zero or a combination of the following values: 00771 - **CV_CALIB_USE_INTRINSIC_GUESS** cameraMatrix contains valid initial values of 00772 fx, fy, cx, cy that are optimized further. Otherwise, (cx, cy) is initially set to the image 00773 center ( imageSize is used), and focal distances are computed in a least-squares fashion. 00774 Note, that if intrinsic parameters are known, there is no need to use this function just to 00775 estimate extrinsic parameters. Use solvePnP instead. 00776 - **CV_CALIB_FIX_PRINCIPAL_POINT** The principal point is not changed during the global 00777 optimization. It stays at the center or at a different location specified when 00778 CV_CALIB_USE_INTRINSIC_GUESS is set too. 00779 - **CV_CALIB_FIX_ASPECT_RATIO** The functions considers only fy as a free parameter. The 00780 ratio fx/fy stays the same as in the input cameraMatrix . When 00781 CV_CALIB_USE_INTRINSIC_GUESS is not set, the actual input values of fx and fy are 00782 ignored, only their ratio is computed and used further. 00783 - **CV_CALIB_ZERO_TANGENT_DIST** Tangential distortion coefficients \f$(p_1, p_2)\f$ are set 00784 to zeros and stay zero. 00785 - **CV_CALIB_FIX_K1,...,CV_CALIB_FIX_K6** The corresponding radial distortion 00786 coefficient is not changed during the optimization. If CV_CALIB_USE_INTRINSIC_GUESS is 00787 set, the coefficient from the supplied distCoeffs matrix is used. Otherwise, it is set to 0. 00788 - **CV_CALIB_RATIONAL_MODEL** Coefficients k4, k5, and k6 are enabled. To provide the 00789 backward compatibility, this extra flag should be explicitly specified to make the 00790 calibration function use the rational model and return 8 coefficients. If the flag is not 00791 set, the function computes and returns only 5 distortion coefficients. 00792 - **CALIB_THIN_PRISM_MODEL** Coefficients s1, s2, s3 and s4 are enabled. To provide the 00793 backward compatibility, this extra flag should be explicitly specified to make the 00794 calibration function use the thin prism model and return 12 coefficients. If the flag is not 00795 set, the function computes and returns only 5 distortion coefficients. 00796 - **CALIB_FIX_S1_S2_S3_S4** The thin prism distortion coefficients are not changed during 00797 the optimization. If CV_CALIB_USE_INTRINSIC_GUESS is set, the coefficient from the 00798 supplied distCoeffs matrix is used. Otherwise, it is set to 0. 00799 - **CALIB_TILTED_MODEL** Coefficients tauX and tauY are enabled. To provide the 00800 backward compatibility, this extra flag should be explicitly specified to make the 00801 calibration function use the tilted sensor model and return 14 coefficients. If the flag is not 00802 set, the function computes and returns only 5 distortion coefficients. 00803 - **CALIB_FIX_TAUX_TAUY** The coefficients of the tilted sensor model are not changed during 00804 the optimization. If CV_CALIB_USE_INTRINSIC_GUESS is set, the coefficient from the 00805 supplied distCoeffs matrix is used. Otherwise, it is set to 0. 00806 @param criteria Termination criteria for the iterative optimization algorithm. 00807 00808 The function estimates the intrinsic camera parameters and extrinsic parameters for each of the 00809 views. The algorithm is based on @cite Zhang2000 and @cite BouguetMCT . The coordinates of 3D object 00810 points and their corresponding 2D projections in each view must be specified. That may be achieved 00811 by using an object with a known geometry and easily detectable feature points. Such an object is 00812 called a calibration rig or calibration pattern, and OpenCV has built-in support for a chessboard as 00813 a calibration rig (see findChessboardCorners ). Currently, initialization of intrinsic parameters 00814 (when CV_CALIB_USE_INTRINSIC_GUESS is not set) is only implemented for planar calibration 00815 patterns (where Z-coordinates of the object points must be all zeros). 3D calibration rigs can also 00816 be used as long as initial cameraMatrix is provided. 00817 00818 The algorithm performs the following steps: 00819 00820 - Compute the initial intrinsic parameters (the option only available for planar calibration 00821 patterns) or read them from the input parameters. The distortion coefficients are all set to 00822 zeros initially unless some of CV_CALIB_FIX_K? are specified. 00823 00824 - Estimate the initial camera pose as if the intrinsic parameters have been already known. This is 00825 done using solvePnP . 00826 00827 - Run the global Levenberg-Marquardt optimization algorithm to minimize the reprojection error, 00828 that is, the total sum of squared distances between the observed feature points imagePoints and 00829 the projected (using the current estimates for camera parameters and the poses) object points 00830 objectPoints. See projectPoints for details. 00831 00832 The function returns the final re-projection error. 00833 00834 @note 00835 If you use a non-square (=non-NxN) grid and findChessboardCorners for calibration, and 00836 calibrateCamera returns bad values (zero distortion coefficients, an image center very far from 00837 (w/2-0.5,h/2-0.5), and/or large differences between \f$f_x\f$ and \f$f_y\f$ (ratios of 10:1 or more)), 00838 then you have probably used patternSize=cvSize(rows,cols) instead of using 00839 patternSize=cvSize(cols,rows) in findChessboardCorners . 00840 00841 @sa 00842 findChessboardCorners, solvePnP, initCameraMatrix2D, stereoCalibrate, undistort 00843 */ 00844 CV_EXPORTS_W double calibrateCamera( InputArrayOfArrays objectPoints, 00845 InputArrayOfArrays imagePoints, Size imageSize, 00846 InputOutputArray cameraMatrix, InputOutputArray distCoeffs, 00847 OutputArrayOfArrays rvecs, OutputArrayOfArrays tvecs, 00848 int flags = 0, TermCriteria criteria = TermCriteria( 00849 TermCriteria::COUNT + TermCriteria::EPS, 30, DBL_EPSILON) ); 00850 00851 /** @brief Computes useful camera characteristics from the camera matrix. 00852 00853 @param cameraMatrix Input camera matrix that can be estimated by calibrateCamera or 00854 stereoCalibrate . 00855 @param imageSize Input image size in pixels. 00856 @param apertureWidth Physical width in mm of the sensor. 00857 @param apertureHeight Physical height in mm of the sensor. 00858 @param fovx Output field of view in degrees along the horizontal sensor axis. 00859 @param fovy Output field of view in degrees along the vertical sensor axis. 00860 @param focalLength Focal length of the lens in mm. 00861 @param principalPoint Principal point in mm. 00862 @param aspectRatio \f$f_y/f_x\f$ 00863 00864 The function computes various useful camera characteristics from the previously estimated camera 00865 matrix. 00866 00867 @note 00868 Do keep in mind that the unity measure 'mm' stands for whatever unit of measure one chooses for 00869 the chessboard pitch (it can thus be any value). 00870 */ 00871 CV_EXPORTS_W void calibrationMatrixValues( InputArray cameraMatrix, Size imageSize, 00872 double apertureWidth, double apertureHeight, 00873 CV_OUT double& fovx, CV_OUT double& fovy, 00874 CV_OUT double& focalLength, CV_OUT Point2d& principalPoint, 00875 CV_OUT double& aspectRatio ); 00876 00877 /** @brief Calibrates the stereo camera. 00878 00879 @param objectPoints Vector of vectors of the calibration pattern points. 00880 @param imagePoints1 Vector of vectors of the projections of the calibration pattern points, 00881 observed by the first camera. 00882 @param imagePoints2 Vector of vectors of the projections of the calibration pattern points, 00883 observed by the second camera. 00884 @param cameraMatrix1 Input/output first camera matrix: 00885 \f$\vecthreethree{f_x^{(j)}}{0}{c_x^{(j)}}{0}{f_y^{(j)}}{c_y^{(j)}}{0}{0}{1}\f$ , \f$j = 0,\, 1\f$ . If 00886 any of CV_CALIB_USE_INTRINSIC_GUESS , CV_CALIB_FIX_ASPECT_RATIO , 00887 CV_CALIB_FIX_INTRINSIC , or CV_CALIB_FIX_FOCAL_LENGTH are specified, some or all of the 00888 matrix components must be initialized. See the flags description for details. 00889 @param distCoeffs1 Input/output vector of distortion coefficients 00890 \f$(k_1, k_2, p_1, p_2[, k_3[, k_4, k_5, k_6 [, s_1, s_2, s_3, s_4[, \tau_x, \tau_y]]]])\f$ of 00891 4, 5, 8, 12 or 14 elements. The output vector length depends on the flags. 00892 @param cameraMatrix2 Input/output second camera matrix. The parameter is similar to cameraMatrix1 00893 @param distCoeffs2 Input/output lens distortion coefficients for the second camera. The parameter 00894 is similar to distCoeffs1 . 00895 @param imageSize Size of the image used only to initialize intrinsic camera matrix. 00896 @param R Output rotation matrix between the 1st and the 2nd camera coordinate systems. 00897 @param T Output translation vector between the coordinate systems of the cameras. 00898 @param E Output essential matrix. 00899 @param F Output fundamental matrix. 00900 @param flags Different flags that may be zero or a combination of the following values: 00901 - **CV_CALIB_FIX_INTRINSIC** Fix cameraMatrix? and distCoeffs? so that only R, T, E , and F 00902 matrices are estimated. 00903 - **CV_CALIB_USE_INTRINSIC_GUESS** Optimize some or all of the intrinsic parameters 00904 according to the specified flags. Initial values are provided by the user. 00905 - **CV_CALIB_FIX_PRINCIPAL_POINT** Fix the principal points during the optimization. 00906 - **CV_CALIB_FIX_FOCAL_LENGTH** Fix \f$f^{(j)}_x\f$ and \f$f^{(j)}_y\f$ . 00907 - **CV_CALIB_FIX_ASPECT_RATIO** Optimize \f$f^{(j)}_y\f$ . Fix the ratio \f$f^{(j)}_x/f^{(j)}_y\f$ 00908 . 00909 - **CV_CALIB_SAME_FOCAL_LENGTH** Enforce \f$f^{(0)}_x=f^{(1)}_x\f$ and \f$f^{(0)}_y=f^{(1)}_y\f$ . 00910 - **CV_CALIB_ZERO_TANGENT_DIST** Set tangential distortion coefficients for each camera to 00911 zeros and fix there. 00912 - **CV_CALIB_FIX_K1,...,CV_CALIB_FIX_K6** Do not change the corresponding radial 00913 distortion coefficient during the optimization. If CV_CALIB_USE_INTRINSIC_GUESS is set, 00914 the coefficient from the supplied distCoeffs matrix is used. Otherwise, it is set to 0. 00915 - **CV_CALIB_RATIONAL_MODEL** Enable coefficients k4, k5, and k6. To provide the backward 00916 compatibility, this extra flag should be explicitly specified to make the calibration 00917 function use the rational model and return 8 coefficients. If the flag is not set, the 00918 function computes and returns only 5 distortion coefficients. 00919 - **CALIB_THIN_PRISM_MODEL** Coefficients s1, s2, s3 and s4 are enabled. To provide the 00920 backward compatibility, this extra flag should be explicitly specified to make the 00921 calibration function use the thin prism model and return 12 coefficients. If the flag is not 00922 set, the function computes and returns only 5 distortion coefficients. 00923 - **CALIB_FIX_S1_S2_S3_S4** The thin prism distortion coefficients are not changed during 00924 the optimization. If CV_CALIB_USE_INTRINSIC_GUESS is set, the coefficient from the 00925 supplied distCoeffs matrix is used. Otherwise, it is set to 0. 00926 - **CALIB_TILTED_MODEL** Coefficients tauX and tauY are enabled. To provide the 00927 backward compatibility, this extra flag should be explicitly specified to make the 00928 calibration function use the tilted sensor model and return 14 coefficients. If the flag is not 00929 set, the function computes and returns only 5 distortion coefficients. 00930 - **CALIB_FIX_TAUX_TAUY** The coefficients of the tilted sensor model are not changed during 00931 the optimization. If CV_CALIB_USE_INTRINSIC_GUESS is set, the coefficient from the 00932 supplied distCoeffs matrix is used. Otherwise, it is set to 0. 00933 @param criteria Termination criteria for the iterative optimization algorithm. 00934 00935 The function estimates transformation between two cameras making a stereo pair. If you have a stereo 00936 camera where the relative position and orientation of two cameras is fixed, and if you computed 00937 poses of an object relative to the first camera and to the second camera, (R1, T1) and (R2, T2), 00938 respectively (this can be done with solvePnP ), then those poses definitely relate to each other. 00939 This means that, given ( \f$R_1\f$,\f$T_1\f$ ), it should be possible to compute ( \f$R_2\f$,\f$T_2\f$ ). You only 00940 need to know the position and orientation of the second camera relative to the first camera. This is 00941 what the described function does. It computes ( \f$R\f$,\f$T\f$ ) so that: 00942 00943 \f[R_2=R*R_1 00944 T_2=R*T_1 + T,\f] 00945 00946 Optionally, it computes the essential matrix E: 00947 00948 \f[E= \vecthreethree{0}{-T_2}{T_1}{T_2}{0}{-T_0}{-T_1}{T_0}{0} *R\f] 00949 00950 where \f$T_i\f$ are components of the translation vector \f$T\f$ : \f$T=[T_0, T_1, T_2]^T\f$ . And the function 00951 can also compute the fundamental matrix F: 00952 00953 \f[F = cameraMatrix2^{-T} E cameraMatrix1^{-1}\f] 00954 00955 Besides the stereo-related information, the function can also perform a full calibration of each of 00956 two cameras. However, due to the high dimensionality of the parameter space and noise in the input 00957 data, the function can diverge from the correct solution. If the intrinsic parameters can be 00958 estimated with high accuracy for each of the cameras individually (for example, using 00959 calibrateCamera ), you are recommended to do so and then pass CV_CALIB_FIX_INTRINSIC flag to the 00960 function along with the computed intrinsic parameters. Otherwise, if all the parameters are 00961 estimated at once, it makes sense to restrict some parameters, for example, pass 00962 CV_CALIB_SAME_FOCAL_LENGTH and CV_CALIB_ZERO_TANGENT_DIST flags, which is usually a 00963 reasonable assumption. 00964 00965 Similarly to calibrateCamera , the function minimizes the total re-projection error for all the 00966 points in all the available views from both cameras. The function returns the final value of the 00967 re-projection error. 00968 */ 00969 CV_EXPORTS_W double stereoCalibrate( InputArrayOfArrays objectPoints, 00970 InputArrayOfArrays imagePoints1, InputArrayOfArrays imagePoints2, 00971 InputOutputArray cameraMatrix1, InputOutputArray distCoeffs1, 00972 InputOutputArray cameraMatrix2, InputOutputArray distCoeffs2, 00973 Size imageSize, OutputArray R,OutputArray T, OutputArray E, OutputArray F, 00974 int flags = CALIB_FIX_INTRINSIC, 00975 TermCriteria criteria = TermCriteria(TermCriteria::COUNT+TermCriteria::EPS, 30, 1e-6) ); 00976 00977 00978 /** @brief Computes rectification transforms for each head of a calibrated stereo camera. 00979 00980 @param cameraMatrix1 First camera matrix. 00981 @param distCoeffs1 First camera distortion parameters. 00982 @param cameraMatrix2 Second camera matrix. 00983 @param distCoeffs2 Second camera distortion parameters. 00984 @param imageSize Size of the image used for stereo calibration. 00985 @param R Rotation matrix between the coordinate systems of the first and the second cameras. 00986 @param T Translation vector between coordinate systems of the cameras. 00987 @param R1 Output 3x3 rectification transform (rotation matrix) for the first camera. 00988 @param R2 Output 3x3 rectification transform (rotation matrix) for the second camera. 00989 @param P1 Output 3x4 projection matrix in the new (rectified) coordinate systems for the first 00990 camera. 00991 @param P2 Output 3x4 projection matrix in the new (rectified) coordinate systems for the second 00992 camera. 00993 @param Q Output \f$4 \times 4\f$ disparity-to-depth mapping matrix (see reprojectImageTo3D ). 00994 @param flags Operation flags that may be zero or CV_CALIB_ZERO_DISPARITY . If the flag is set, 00995 the function makes the principal points of each camera have the same pixel coordinates in the 00996 rectified views. And if the flag is not set, the function may still shift the images in the 00997 horizontal or vertical direction (depending on the orientation of epipolar lines) to maximize the 00998 useful image area. 00999 @param alpha Free scaling parameter. If it is -1 or absent, the function performs the default 01000 scaling. Otherwise, the parameter should be between 0 and 1. alpha=0 means that the rectified 01001 images are zoomed and shifted so that only valid pixels are visible (no black areas after 01002 rectification). alpha=1 means that the rectified image is decimated and shifted so that all the 01003 pixels from the original images from the cameras are retained in the rectified images (no source 01004 image pixels are lost). Obviously, any intermediate value yields an intermediate result between 01005 those two extreme cases. 01006 @param newImageSize New image resolution after rectification. The same size should be passed to 01007 initUndistortRectifyMap (see the stereo_calib.cpp sample in OpenCV samples directory). When (0,0) 01008 is passed (default), it is set to the original imageSize . Setting it to larger value can help you 01009 preserve details in the original image, especially when there is a big radial distortion. 01010 @param validPixROI1 Optional output rectangles inside the rectified images where all the pixels 01011 are valid. If alpha=0 , the ROIs cover the whole images. Otherwise, they are likely to be smaller 01012 (see the picture below). 01013 @param validPixROI2 Optional output rectangles inside the rectified images where all the pixels 01014 are valid. If alpha=0 , the ROIs cover the whole images. Otherwise, they are likely to be smaller 01015 (see the picture below). 01016 01017 The function computes the rotation matrices for each camera that (virtually) make both camera image 01018 planes the same plane. Consequently, this makes all the epipolar lines parallel and thus simplifies 01019 the dense stereo correspondence problem. The function takes the matrices computed by stereoCalibrate 01020 as input. As output, it provides two rotation matrices and also two projection matrices in the new 01021 coordinates. The function distinguishes the following two cases: 01022 01023 - **Horizontal stereo**: the first and the second camera views are shifted relative to each other 01024 mainly along the x axis (with possible small vertical shift). In the rectified images, the 01025 corresponding epipolar lines in the left and right cameras are horizontal and have the same 01026 y-coordinate. P1 and P2 look like: 01027 01028 \f[\texttt{P1} = \begin{bmatrix} f & 0 & cx_1 & 0 \\ 0 & f & cy & 0 \\ 0 & 0 & 1 & 0 \end{bmatrix}\f] 01029 01030 \f[\texttt{P2} = \begin{bmatrix} f & 0 & cx_2 & T_x*f \\ 0 & f & cy & 0 \\ 0 & 0 & 1 & 0 \end{bmatrix} ,\f] 01031 01032 where \f$T_x\f$ is a horizontal shift between the cameras and \f$cx_1=cx_2\f$ if 01033 CV_CALIB_ZERO_DISPARITY is set. 01034 01035 - **Vertical stereo**: the first and the second camera views are shifted relative to each other 01036 mainly in vertical direction (and probably a bit in the horizontal direction too). The epipolar 01037 lines in the rectified images are vertical and have the same x-coordinate. P1 and P2 look like: 01038 01039 \f[\texttt{P1} = \begin{bmatrix} f & 0 & cx & 0 \\ 0 & f & cy_1 & 0 \\ 0 & 0 & 1 & 0 \end{bmatrix}\f] 01040 01041 \f[\texttt{P2} = \begin{bmatrix} f & 0 & cx & 0 \\ 0 & f & cy_2 & T_y*f \\ 0 & 0 & 1 & 0 \end{bmatrix} ,\f] 01042 01043 where \f$T_y\f$ is a vertical shift between the cameras and \f$cy_1=cy_2\f$ if CALIB_ZERO_DISPARITY is 01044 set. 01045 01046 As you can see, the first three columns of P1 and P2 will effectively be the new "rectified" camera 01047 matrices. The matrices, together with R1 and R2 , can then be passed to initUndistortRectifyMap to 01048 initialize the rectification map for each camera. 01049 01050 See below the screenshot from the stereo_calib.cpp sample. Some red horizontal lines pass through 01051 the corresponding image regions. This means that the images are well rectified, which is what most 01052 stereo correspondence algorithms rely on. The green rectangles are roi1 and roi2 . You see that 01053 their interiors are all valid pixels. 01054 01055  01056 */ 01057 CV_EXPORTS_W void stereoRectify( InputArray cameraMatrix1, InputArray distCoeffs1, 01058 InputArray cameraMatrix2, InputArray distCoeffs2, 01059 Size imageSize, InputArray R, InputArray T, 01060 OutputArray R1, OutputArray R2, 01061 OutputArray P1, OutputArray P2, 01062 OutputArray Q, int flags = CALIB_ZERO_DISPARITY, 01063 double alpha = -1, Size newImageSize = Size(), 01064 CV_OUT Rect* validPixROI1 = 0, CV_OUT Rect* validPixROI2 = 0 ); 01065 01066 /** @brief Computes a rectification transform for an uncalibrated stereo camera. 01067 01068 @param points1 Array of feature points in the first image. 01069 @param points2 The corresponding points in the second image. The same formats as in 01070 findFundamentalMat are supported. 01071 @param F Input fundamental matrix. It can be computed from the same set of point pairs using 01072 findFundamentalMat . 01073 @param imgSize Size of the image. 01074 @param H1 Output rectification homography matrix for the first image. 01075 @param H2 Output rectification homography matrix for the second image. 01076 @param threshold Optional threshold used to filter out the outliers. If the parameter is greater 01077 than zero, all the point pairs that do not comply with the epipolar geometry (that is, the points 01078 for which \f$|\texttt{points2[i]}^T*\texttt{F}*\texttt{points1[i]}|>\texttt{threshold}\f$ ) are 01079 rejected prior to computing the homographies. Otherwise,all the points are considered inliers. 01080 01081 The function computes the rectification transformations without knowing intrinsic parameters of the 01082 cameras and their relative position in the space, which explains the suffix "uncalibrated". Another 01083 related difference from stereoRectify is that the function outputs not the rectification 01084 transformations in the object (3D) space, but the planar perspective transformations encoded by the 01085 homography matrices H1 and H2 . The function implements the algorithm @cite Hartley99 . 01086 01087 @note 01088 While the algorithm does not need to know the intrinsic parameters of the cameras, it heavily 01089 depends on the epipolar geometry. Therefore, if the camera lenses have a significant distortion, 01090 it would be better to correct it before computing the fundamental matrix and calling this 01091 function. For example, distortion coefficients can be estimated for each head of stereo camera 01092 separately by using calibrateCamera . Then, the images can be corrected using undistort , or 01093 just the point coordinates can be corrected with undistortPoints . 01094 */ 01095 CV_EXPORTS_W bool stereoRectifyUncalibrated( InputArray points1, InputArray points2, 01096 InputArray F, Size imgSize, 01097 OutputArray H1, OutputArray H2, 01098 double threshold = 5 ); 01099 01100 //! computes the rectification transformations for 3-head camera, where all the heads are on the same line. 01101 CV_EXPORTS_W float rectify3Collinear( InputArray cameraMatrix1, InputArray distCoeffs1, 01102 InputArray cameraMatrix2, InputArray distCoeffs2, 01103 InputArray cameraMatrix3, InputArray distCoeffs3, 01104 InputArrayOfArrays imgpt1, InputArrayOfArrays imgpt3, 01105 Size imageSize, InputArray R12, InputArray T12, 01106 InputArray R13, InputArray T13, 01107 OutputArray R1, OutputArray R2, OutputArray R3, 01108 OutputArray P1, OutputArray P2, OutputArray P3, 01109 OutputArray Q, double alpha, Size newImgSize, 01110 CV_OUT Rect* roi1, CV_OUT Rect* roi2, int flags ); 01111 01112 /** @brief Returns the new camera matrix based on the free scaling parameter. 01113 01114 @param cameraMatrix Input camera matrix. 01115 @param distCoeffs Input vector of distortion coefficients 01116 \f$(k_1, k_2, p_1, p_2[, k_3[, k_4, k_5, k_6 [, s_1, s_2, s_3, s_4[, \tau_x, \tau_y]]]])\f$ of 01117 4, 5, 8, 12 or 14 elements. If the vector is NULL/empty, the zero distortion coefficients are 01118 assumed. 01119 @param imageSize Original image size. 01120 @param alpha Free scaling parameter between 0 (when all the pixels in the undistorted image are 01121 valid) and 1 (when all the source image pixels are retained in the undistorted image). See 01122 stereoRectify for details. 01123 @param newImgSize Image size after rectification. By default,it is set to imageSize . 01124 @param validPixROI Optional output rectangle that outlines all-good-pixels region in the 01125 undistorted image. See roi1, roi2 description in stereoRectify . 01126 @param centerPrincipalPoint Optional flag that indicates whether in the new camera matrix the 01127 principal point should be at the image center or not. By default, the principal point is chosen to 01128 best fit a subset of the source image (determined by alpha) to the corrected image. 01129 @return new_camera_matrix Output new camera matrix. 01130 01131 The function computes and returns the optimal new camera matrix based on the free scaling parameter. 01132 By varying this parameter, you may retrieve only sensible pixels alpha=0 , keep all the original 01133 image pixels if there is valuable information in the corners alpha=1 , or get something in between. 01134 When alpha>0 , the undistortion result is likely to have some black pixels corresponding to 01135 "virtual" pixels outside of the captured distorted image. The original camera matrix, distortion 01136 coefficients, the computed new camera matrix, and newImageSize should be passed to 01137 initUndistortRectifyMap to produce the maps for remap . 01138 */ 01139 CV_EXPORTS_W Mat getOptimalNewCameraMatrix( InputArray cameraMatrix, InputArray distCoeffs, 01140 Size imageSize, double alpha, Size newImgSize = Size(), 01141 CV_OUT Rect* validPixROI = 0, 01142 bool centerPrincipalPoint = false); 01143 01144 /** @brief Converts points from Euclidean to homogeneous space. 01145 01146 @param src Input vector of N-dimensional points. 01147 @param dst Output vector of N+1-dimensional points. 01148 01149 The function converts points from Euclidean to homogeneous space by appending 1's to the tuple of 01150 point coordinates. That is, each point (x1, x2, ..., xn) is converted to (x1, x2, ..., xn, 1). 01151 */ 01152 CV_EXPORTS_W void convertPointsToHomogeneous( InputArray src, OutputArray dst ); 01153 01154 /** @brief Converts points from homogeneous to Euclidean space. 01155 01156 @param src Input vector of N-dimensional points. 01157 @param dst Output vector of N-1-dimensional points. 01158 01159 The function converts points homogeneous to Euclidean space using perspective projection. That is, 01160 each point (x1, x2, ... x(n-1), xn) is converted to (x1/xn, x2/xn, ..., x(n-1)/xn). When xn=0, the 01161 output point coordinates will be (0,0,0,...). 01162 */ 01163 CV_EXPORTS_W void convertPointsFromHomogeneous( InputArray src, OutputArray dst ); 01164 01165 /** @brief Converts points to/from homogeneous coordinates. 01166 01167 @param src Input array or vector of 2D, 3D, or 4D points. 01168 @param dst Output vector of 2D, 3D, or 4D points. 01169 01170 The function converts 2D or 3D points from/to homogeneous coordinates by calling either 01171 convertPointsToHomogeneous or convertPointsFromHomogeneous. 01172 01173 @note The function is obsolete. Use one of the previous two functions instead. 01174 */ 01175 CV_EXPORTS void convertPointsHomogeneous( InputArray src, OutputArray dst ); 01176 01177 /** @brief Calculates a fundamental matrix from the corresponding points in two images. 01178 01179 @param points1 Array of N points from the first image. The point coordinates should be 01180 floating-point (single or double precision). 01181 @param points2 Array of the second image points of the same size and format as points1 . 01182 @param method Method for computing a fundamental matrix. 01183 - **CV_FM_7POINT** for a 7-point algorithm. \f$N = 7\f$ 01184 - **CV_FM_8POINT** for an 8-point algorithm. \f$N \ge 8\f$ 01185 - **CV_FM_RANSAC** for the RANSAC algorithm. \f$N \ge 8\f$ 01186 - **CV_FM_LMEDS** for the LMedS algorithm. \f$N \ge 8\f$ 01187 @param param1 Parameter used for RANSAC. It is the maximum distance from a point to an epipolar 01188 line in pixels, beyond which the point is considered an outlier and is not used for computing the 01189 final fundamental matrix. It can be set to something like 1-3, depending on the accuracy of the 01190 point localization, image resolution, and the image noise. 01191 @param param2 Parameter used for the RANSAC or LMedS methods only. It specifies a desirable level 01192 of confidence (probability) that the estimated matrix is correct. 01193 @param mask 01194 01195 The epipolar geometry is described by the following equation: 01196 01197 \f[[p_2; 1]^T F [p_1; 1] = 0\f] 01198 01199 where \f$F\f$ is a fundamental matrix, \f$p_1\f$ and \f$p_2\f$ are corresponding points in the first and the 01200 second images, respectively. 01201 01202 The function calculates the fundamental matrix using one of four methods listed above and returns 01203 the found fundamental matrix. Normally just one matrix is found. But in case of the 7-point 01204 algorithm, the function may return up to 3 solutions ( \f$9 \times 3\f$ matrix that stores all 3 01205 matrices sequentially). 01206 01207 The calculated fundamental matrix may be passed further to computeCorrespondEpilines that finds the 01208 epipolar lines corresponding to the specified points. It can also be passed to 01209 stereoRectifyUncalibrated to compute the rectification transformation. : 01210 @code 01211 // Example. Estimation of fundamental matrix using the RANSAC algorithm 01212 int point_count = 100; 01213 vector<Point2f> points1(point_count); 01214 vector<Point2f> points2(point_count); 01215 01216 // initialize the points here ... 01217 for( int i = 0; i < point_count; i++ ) 01218 { 01219 points1[i] = ...; 01220 points2[i] = ...; 01221 } 01222 01223 Mat fundamental_matrix = 01224 findFundamentalMat(points1, points2, FM_RANSAC, 3, 0.99); 01225 @endcode 01226 */ 01227 CV_EXPORTS_W Mat findFundamentalMat( InputArray points1, InputArray points2, 01228 int method = FM_RANSAC, 01229 double param1 = 3., double param2 = 0.99, 01230 OutputArray mask = noArray() ); 01231 01232 /** @overload */ 01233 CV_EXPORTS Mat findFundamentalMat( InputArray points1, InputArray points2, 01234 OutputArray mask, int method = FM_RANSAC, 01235 double param1 = 3., double param2 = 0.99 ); 01236 01237 /** @brief Calculates an essential matrix from the corresponding points in two images. 01238 01239 @param points1 Array of N (N >= 5) 2D points from the first image. The point coordinates should 01240 be floating-point (single or double precision). 01241 @param points2 Array of the second image points of the same size and format as points1 . 01242 @param cameraMatrix Camera matrix \f$K = \vecthreethree{f_x}{0}{c_x}{0}{f_y}{c_y}{0}{0}{1}\f$ . 01243 Note that this function assumes that points1 and points2 are feature points from cameras with the 01244 same camera matrix. 01245 @param method Method for computing a fundamental matrix. 01246 - **RANSAC** for the RANSAC algorithm. 01247 - **MEDS** for the LMedS algorithm. 01248 @param threshold Parameter used for RANSAC. It is the maximum distance from a point to an epipolar 01249 line in pixels, beyond which the point is considered an outlier and is not used for computing the 01250 final fundamental matrix. It can be set to something like 1-3, depending on the accuracy of the 01251 point localization, image resolution, and the image noise. 01252 @param prob Parameter used for the RANSAC or LMedS methods only. It specifies a desirable level of 01253 confidence (probability) that the estimated matrix is correct. 01254 @param mask Output array of N elements, every element of which is set to 0 for outliers and to 1 01255 for the other points. The array is computed only in the RANSAC and LMedS methods. 01256 01257 This function estimates essential matrix based on the five-point algorithm solver in @cite Nister03 . 01258 @cite SteweniusCFS is also a related. The epipolar geometry is described by the following equation: 01259 01260 \f[[p_2; 1]^T K^{-T} E K^{-1} [p_1; 1] = 0\f] 01261 01262 where \f$E\f$ is an essential matrix, \f$p_1\f$ and \f$p_2\f$ are corresponding points in the first and the 01263 second images, respectively. The result of this function may be passed further to 01264 decomposeEssentialMat or recoverPose to recover the relative pose between cameras. 01265 */ 01266 CV_EXPORTS_W Mat findEssentialMat( InputArray points1, InputArray points2, 01267 InputArray cameraMatrix, int method = RANSAC, 01268 double prob = 0.999, double threshold = 1.0, 01269 OutputArray mask = noArray() ); 01270 01271 /** @overload 01272 @param points1 Array of N (N >= 5) 2D points from the first image. The point coordinates should 01273 be floating-point (single or double precision). 01274 @param points2 Array of the second image points of the same size and format as points1 . 01275 @param focal focal length of the camera. Note that this function assumes that points1 and points2 01276 are feature points from cameras with same focal length and principle point. 01277 @param pp principle point of the camera. 01278 @param method Method for computing a fundamental matrix. 01279 - **RANSAC** for the RANSAC algorithm. 01280 - **LMEDS** for the LMedS algorithm. 01281 @param threshold Parameter used for RANSAC. It is the maximum distance from a point to an epipolar 01282 line in pixels, beyond which the point is considered an outlier and is not used for computing the 01283 final fundamental matrix. It can be set to something like 1-3, depending on the accuracy of the 01284 point localization, image resolution, and the image noise. 01285 @param prob Parameter used for the RANSAC or LMedS methods only. It specifies a desirable level of 01286 confidence (probability) that the estimated matrix is correct. 01287 @param mask Output array of N elements, every element of which is set to 0 for outliers and to 1 01288 for the other points. The array is computed only in the RANSAC and LMedS methods. 01289 01290 This function differs from the one above that it computes camera matrix from focal length and 01291 principal point: 01292 01293 \f[K = 01294 \begin{bmatrix} 01295 f & 0 & x_{pp} \\ 01296 0 & f & y_{pp} \\ 01297 0 & 0 & 1 01298 \end{bmatrix}\f] 01299 */ 01300 CV_EXPORTS_W Mat findEssentialMat( InputArray points1, InputArray points2, 01301 double focal = 1.0, Point2d pp = Point2d(0, 0), 01302 int method = RANSAC, double prob = 0.999, 01303 double threshold = 1.0, OutputArray mask = noArray() ); 01304 01305 /** @brief Decompose an essential matrix to possible rotations and translation. 01306 01307 @param E The input essential matrix. 01308 @param R1 One possible rotation matrix. 01309 @param R2 Another possible rotation matrix. 01310 @param t One possible translation. 01311 01312 This function decompose an essential matrix E using svd decomposition @cite HartleyZ00 . Generally 4 01313 possible poses exists for a given E. They are \f$[R_1, t]\f$, \f$[R_1, -t]\f$, \f$[R_2, t]\f$, \f$[R_2, -t]\f$. By 01314 decomposing E, you can only get the direction of the translation, so the function returns unit t. 01315 */ 01316 CV_EXPORTS_W void decomposeEssentialMat( InputArray E, OutputArray R1, OutputArray R2, OutputArray t ); 01317 01318 /** @brief Recover relative camera rotation and translation from an estimated essential matrix and the 01319 corresponding points in two images, using cheirality check. Returns the number of inliers which pass 01320 the check. 01321 01322 @param E The input essential matrix. 01323 @param points1 Array of N 2D points from the first image. The point coordinates should be 01324 floating-point (single or double precision). 01325 @param points2 Array of the second image points of the same size and format as points1 . 01326 @param cameraMatrix Camera matrix \f$K = \vecthreethree{f_x}{0}{c_x}{0}{f_y}{c_y}{0}{0}{1}\f$ . 01327 Note that this function assumes that points1 and points2 are feature points from cameras with the 01328 same camera matrix. 01329 @param R Recovered relative rotation. 01330 @param t Recoverd relative translation. 01331 @param mask Input/output mask for inliers in points1 and points2. 01332 : If it is not empty, then it marks inliers in points1 and points2 for then given essential 01333 matrix E. Only these inliers will be used to recover pose. In the output mask only inliers 01334 which pass the cheirality check. 01335 This function decomposes an essential matrix using decomposeEssentialMat and then verifies possible 01336 pose hypotheses by doing cheirality check. The cheirality check basically means that the 01337 triangulated 3D points should have positive depth. Some details can be found in @cite Nister03 . 01338 01339 This function can be used to process output E and mask from findEssentialMat. In this scenario, 01340 points1 and points2 are the same input for findEssentialMat. : 01341 @code 01342 // Example. Estimation of fundamental matrix using the RANSAC algorithm 01343 int point_count = 100; 01344 vector<Point2f> points1(point_count); 01345 vector<Point2f> points2(point_count); 01346 01347 // initialize the points here ... 01348 for( int i = 0; i < point_count; i++ ) 01349 { 01350 points1[i] = ...; 01351 points2[i] = ...; 01352 } 01353 01354 // cametra matrix with both focal lengths = 1, and principal point = (0, 0) 01355 Mat cameraMatrix = Mat::eye(3, 3, CV_64F); 01356 01357 Mat E, R, t, mask; 01358 01359 E = findEssentialMat(points1, points2, cameraMatrix, RANSAC, 0.999, 1.0, mask); 01360 recoverPose(E, points1, points2, cameraMatrix, R, t, mask); 01361 @endcode 01362 */ 01363 CV_EXPORTS_W int recoverPose( InputArray E, InputArray points1, InputArray points2, 01364 InputArray cameraMatrix, OutputArray R, OutputArray t, 01365 InputOutputArray mask = noArray() ); 01366 01367 /** @overload 01368 @param E The input essential matrix. 01369 @param points1 Array of N 2D points from the first image. The point coordinates should be 01370 floating-point (single or double precision). 01371 @param points2 Array of the second image points of the same size and format as points1 . 01372 @param R Recovered relative rotation. 01373 @param t Recoverd relative translation. 01374 @param focal Focal length of the camera. Note that this function assumes that points1 and points2 01375 are feature points from cameras with same focal length and principle point. 01376 @param pp Principle point of the camera. 01377 @param mask Input/output mask for inliers in points1 and points2. 01378 : If it is not empty, then it marks inliers in points1 and points2 for then given essential 01379 matrix E. Only these inliers will be used to recover pose. In the output mask only inliers 01380 which pass the cheirality check. 01381 01382 This function differs from the one above that it computes camera matrix from focal length and 01383 principal point: 01384 01385 \f[K = 01386 \begin{bmatrix} 01387 f & 0 & x_{pp} \\ 01388 0 & f & y_{pp} \\ 01389 0 & 0 & 1 01390 \end{bmatrix}\f] 01391 */ 01392 CV_EXPORTS_W int recoverPose( InputArray E, InputArray points1, InputArray points2, 01393 OutputArray R, OutputArray t, 01394 double focal = 1.0, Point2d pp = Point2d(0, 0), 01395 InputOutputArray mask = noArray() ); 01396 01397 /** @brief For points in an image of a stereo pair, computes the corresponding epilines in the other image. 01398 01399 @param points Input points. \f$N \times 1\f$ or \f$1 \times N\f$ matrix of type CV_32FC2 or 01400 vector<Point2f> . 01401 @param whichImage Index of the image (1 or 2) that contains the points . 01402 @param F Fundamental matrix that can be estimated using findFundamentalMat or stereoRectify . 01403 @param lines Output vector of the epipolar lines corresponding to the points in the other image. 01404 Each line \f$ax + by + c=0\f$ is encoded by 3 numbers \f$(a, b, c)\f$ . 01405 01406 For every point in one of the two images of a stereo pair, the function finds the equation of the 01407 corresponding epipolar line in the other image. 01408 01409 From the fundamental matrix definition (see findFundamentalMat ), line \f$l^{(2)}_i\f$ in the second 01410 image for the point \f$p^{(1)}_i\f$ in the first image (when whichImage=1 ) is computed as: 01411 01412 \f[l^{(2)}_i = F p^{(1)}_i\f] 01413 01414 And vice versa, when whichImage=2, \f$l^{(1)}_i\f$ is computed from \f$p^{(2)}_i\f$ as: 01415 01416 \f[l^{(1)}_i = F^T p^{(2)}_i\f] 01417 01418 Line coefficients are defined up to a scale. They are normalized so that \f$a_i^2+b_i^2=1\f$ . 01419 */ 01420 CV_EXPORTS_W void computeCorrespondEpilines( InputArray points, int whichImage, 01421 InputArray F, OutputArray lines ); 01422 01423 /** @brief Reconstructs points by triangulation. 01424 01425 @param projMatr1 3x4 projection matrix of the first camera. 01426 @param projMatr2 3x4 projection matrix of the second camera. 01427 @param projPoints1 2xN array of feature points in the first image. In case of c++ version it can 01428 be also a vector of feature points or two-channel matrix of size 1xN or Nx1. 01429 @param projPoints2 2xN array of corresponding points in the second image. In case of c++ version 01430 it can be also a vector of feature points or two-channel matrix of size 1xN or Nx1. 01431 @param points4D 4xN array of reconstructed points in homogeneous coordinates. 01432 01433 The function reconstructs 3-dimensional points (in homogeneous coordinates) by using their 01434 observations with a stereo camera. Projections matrices can be obtained from stereoRectify. 01435 01436 @note 01437 Keep in mind that all input data should be of float type in order for this function to work. 01438 01439 @sa 01440 reprojectImageTo3D 01441 */ 01442 CV_EXPORTS_W void triangulatePoints( InputArray projMatr1, InputArray projMatr2, 01443 InputArray projPoints1, InputArray projPoints2, 01444 OutputArray points4D ); 01445 01446 /** @brief Refines coordinates of corresponding points. 01447 01448 @param F 3x3 fundamental matrix. 01449 @param points1 1xN array containing the first set of points. 01450 @param points2 1xN array containing the second set of points. 01451 @param newPoints1 The optimized points1. 01452 @param newPoints2 The optimized points2. 01453 01454 The function implements the Optimal Triangulation Method (see Multiple View Geometry for details). 01455 For each given point correspondence points1[i] <-> points2[i], and a fundamental matrix F, it 01456 computes the corrected correspondences newPoints1[i] <-> newPoints2[i] that minimize the geometric 01457 error \f$d(points1[i], newPoints1[i])^2 + d(points2[i],newPoints2[i])^2\f$ (where \f$d(a,b)\f$ is the 01458 geometric distance between points \f$a\f$ and \f$b\f$ ) subject to the epipolar constraint 01459 \f$newPoints2^T * F * newPoints1 = 0\f$ . 01460 */ 01461 CV_EXPORTS_W void correctMatches( InputArray F, InputArray points1, InputArray points2, 01462 OutputArray newPoints1, OutputArray newPoints2 ); 01463 01464 /** @brief Filters off small noise blobs (speckles) in the disparity map 01465 01466 @param img The input 16-bit signed disparity image 01467 @param newVal The disparity value used to paint-off the speckles 01468 @param maxSpeckleSize The maximum speckle size to consider it a speckle. Larger blobs are not 01469 affected by the algorithm 01470 @param maxDiff Maximum difference between neighbor disparity pixels to put them into the same 01471 blob. Note that since StereoBM, StereoSGBM and may be other algorithms return a fixed-point 01472 disparity map, where disparity values are multiplied by 16, this scale factor should be taken into 01473 account when specifying this parameter value. 01474 @param buf The optional temporary buffer to avoid memory allocation within the function. 01475 */ 01476 CV_EXPORTS_W void filterSpeckles( InputOutputArray img, double newVal, 01477 int maxSpeckleSize, double maxDiff, 01478 InputOutputArray buf = noArray() ); 01479 01480 //! computes valid disparity ROI from the valid ROIs of the rectified images (that are returned by cv::stereoRectify()) 01481 CV_EXPORTS_W Rect getValidDisparityROI( Rect roi1, Rect roi2, 01482 int minDisparity, int numberOfDisparities, 01483 int SADWindowSize ); 01484 01485 //! validates disparity using the left-right check. The matrix "cost" should be computed by the stereo correspondence algorithm 01486 CV_EXPORTS_W void validateDisparity( InputOutputArray disparity, InputArray cost, 01487 int minDisparity, int numberOfDisparities, 01488 int disp12MaxDisp = 1 ); 01489 01490 /** @brief Reprojects a disparity image to 3D space. 01491 01492 @param disparity Input single-channel 8-bit unsigned, 16-bit signed, 32-bit signed or 32-bit 01493 floating-point disparity image. If 16-bit signed format is used, the values are assumed to have no 01494 fractional bits. 01495 @param _3dImage Output 3-channel floating-point image of the same size as disparity . Each 01496 element of _3dImage(x,y) contains 3D coordinates of the point (x,y) computed from the disparity 01497 map. 01498 @param Q \f$4 \times 4\f$ perspective transformation matrix that can be obtained with stereoRectify. 01499 @param handleMissingValues Indicates, whether the function should handle missing values (i.e. 01500 points where the disparity was not computed). If handleMissingValues=true, then pixels with the 01501 minimal disparity that corresponds to the outliers (see StereoMatcher::compute ) are transformed 01502 to 3D points with a very large Z value (currently set to 10000). 01503 @param ddepth The optional output array depth. If it is -1, the output image will have CV_32F 01504 depth. ddepth can also be set to CV_16S, CV_32S or CV_32F. 01505 01506 The function transforms a single-channel disparity map to a 3-channel image representing a 3D 01507 surface. That is, for each pixel (x,y) andthe corresponding disparity d=disparity(x,y) , it 01508 computes: 01509 01510 \f[\begin{array}{l} [X \; Y \; Z \; W]^T = \texttt{Q} *[x \; y \; \texttt{disparity} (x,y) \; 1]^T \\ \texttt{\_3dImage} (x,y) = (X/W, \; Y/W, \; Z/W) \end{array}\f] 01511 01512 The matrix Q can be an arbitrary \f$4 \times 4\f$ matrix (for example, the one computed by 01513 stereoRectify). To reproject a sparse set of points {(x,y,d),...} to 3D space, use 01514 perspectiveTransform . 01515 */ 01516 CV_EXPORTS_W void reprojectImageTo3D( InputArray disparity, 01517 OutputArray _3dImage, InputArray Q, 01518 bool handleMissingValues = false, 01519 int ddepth = -1 ); 01520 01521 /** @brief Calculates the Sampson Distance between two points. 01522 01523 The function sampsonDistance calculates and returns the first order approximation of the geometric error as: 01524 \f[sd( \texttt{pt1} , \texttt{pt2} )= \frac{(\texttt{pt2}^t \cdot \texttt{F} \cdot \texttt{pt1})^2}{(\texttt{F} \cdot \texttt{pt1})(0) + (\texttt{F} \cdot \texttt{pt1})(1) + (\texttt{F}^t \cdot \texttt{pt2})(0) + (\texttt{F}^t \cdot \texttt{pt2})(1)}\f] 01525 The fundamental matrix may be calculated using the cv::findFundamentalMat function. See HZ 11.4.3 for details. 01526 @param pt1 first homogeneous 2d point 01527 @param pt2 second homogeneous 2d point 01528 @param F fundamental matrix 01529 */ 01530 CV_EXPORTS_W double sampsonDistance(InputArray pt1, InputArray pt2, InputArray F); 01531 01532 /** @brief Computes an optimal affine transformation between two 3D point sets. 01533 01534 @param src First input 3D point set. 01535 @param dst Second input 3D point set. 01536 @param out Output 3D affine transformation matrix \f$3 \times 4\f$ . 01537 @param inliers Output vector indicating which points are inliers. 01538 @param ransacThreshold Maximum reprojection error in the RANSAC algorithm to consider a point as 01539 an inlier. 01540 @param confidence Confidence level, between 0 and 1, for the estimated transformation. Anything 01541 between 0.95 and 0.99 is usually good enough. Values too close to 1 can slow down the estimation 01542 significantly. Values lower than 0.8-0.9 can result in an incorrectly estimated transformation. 01543 01544 The function estimates an optimal 3D affine transformation between two 3D point sets using the 01545 RANSAC algorithm. 01546 */ 01547 CV_EXPORTS_W int estimateAffine3D(InputArray src, InputArray dst, 01548 OutputArray out, OutputArray inliers, 01549 double ransacThreshold = 3, double confidence = 0.99); 01550 01551 /** @brief Decompose a homography matrix to rotation(s), translation(s) and plane normal(s). 01552 01553 @param H The input homography matrix between two images. 01554 @param K The input intrinsic camera calibration matrix. 01555 @param rotations Array of rotation matrices. 01556 @param translations Array of translation matrices. 01557 @param normals Array of plane normal matrices. 01558 01559 This function extracts relative camera motion between two views observing a planar object from the 01560 homography H induced by the plane. The intrinsic camera matrix K must also be provided. The function 01561 may return up to four mathematical solution sets. At least two of the solutions may further be 01562 invalidated if point correspondences are available by applying positive depth constraint (all points 01563 must be in front of the camera). The decomposition method is described in detail in @cite Malis . 01564 */ 01565 CV_EXPORTS_W int decomposeHomographyMat(InputArray H, 01566 InputArray K, 01567 OutputArrayOfArrays rotations, 01568 OutputArrayOfArrays translations, 01569 OutputArrayOfArrays normals); 01570 01571 /** @brief The base class for stereo correspondence algorithms. 01572 */ 01573 class CV_EXPORTS_W StereoMatcher : public Algorithm 01574 { 01575 public: 01576 enum { DISP_SHIFT = 4, 01577 DISP_SCALE = (1 << DISP_SHIFT) 01578 }; 01579 01580 /** @brief Computes disparity map for the specified stereo pair 01581 01582 @param left Left 8-bit single-channel image. 01583 @param right Right image of the same size and the same type as the left one. 01584 @param disparity Output disparity map. It has the same size as the input images. Some algorithms, 01585 like StereoBM or StereoSGBM compute 16-bit fixed-point disparity map (where each disparity value 01586 has 4 fractional bits), whereas other algorithms output 32-bit floating-point disparity map. 01587 */ 01588 CV_WRAP virtual void compute( InputArray left, InputArray right, 01589 OutputArray disparity ) = 0; 01590 01591 CV_WRAP virtual int getMinDisparity() const = 0; 01592 CV_WRAP virtual void setMinDisparity(int minDisparity) = 0; 01593 01594 CV_WRAP virtual int getNumDisparities() const = 0; 01595 CV_WRAP virtual void setNumDisparities(int numDisparities) = 0; 01596 01597 CV_WRAP virtual int getBlockSize() const = 0; 01598 CV_WRAP virtual void setBlockSize(int blockSize) = 0; 01599 01600 CV_WRAP virtual int getSpeckleWindowSize() const = 0; 01601 CV_WRAP virtual void setSpeckleWindowSize(int speckleWindowSize) = 0; 01602 01603 CV_WRAP virtual int getSpeckleRange() const = 0; 01604 CV_WRAP virtual void setSpeckleRange(int speckleRange) = 0; 01605 01606 CV_WRAP virtual int getDisp12MaxDiff() const = 0; 01607 CV_WRAP virtual void setDisp12MaxDiff(int disp12MaxDiff) = 0; 01608 }; 01609 01610 01611 /** @brief Class for computing stereo correspondence using the block matching algorithm, introduced and 01612 contributed to OpenCV by K. Konolige. 01613 */ 01614 class CV_EXPORTS_W StereoBM : public StereoMatcher 01615 { 01616 public: 01617 enum { PREFILTER_NORMALIZED_RESPONSE = 0, 01618 PREFILTER_XSOBEL = 1 01619 }; 01620 01621 CV_WRAP virtual int getPreFilterType() const = 0; 01622 CV_WRAP virtual void setPreFilterType(int preFilterType) = 0; 01623 01624 CV_WRAP virtual int getPreFilterSize() const = 0; 01625 CV_WRAP virtual void setPreFilterSize(int preFilterSize) = 0; 01626 01627 CV_WRAP virtual int getPreFilterCap() const = 0; 01628 CV_WRAP virtual void setPreFilterCap(int preFilterCap) = 0; 01629 01630 CV_WRAP virtual int getTextureThreshold() const = 0; 01631 CV_WRAP virtual void setTextureThreshold(int textureThreshold) = 0; 01632 01633 CV_WRAP virtual int getUniquenessRatio() const = 0; 01634 CV_WRAP virtual void setUniquenessRatio(int uniquenessRatio) = 0; 01635 01636 CV_WRAP virtual int getSmallerBlockSize() const = 0; 01637 CV_WRAP virtual void setSmallerBlockSize(int blockSize) = 0; 01638 01639 CV_WRAP virtual Rect getROI1() const = 0; 01640 CV_WRAP virtual void setROI1(Rect roi1) = 0; 01641 01642 CV_WRAP virtual Rect getROI2() const = 0; 01643 CV_WRAP virtual void setROI2(Rect roi2) = 0; 01644 01645 /** @brief Creates StereoBM object 01646 01647 @param numDisparities the disparity search range. For each pixel algorithm will find the best 01648 disparity from 0 (default minimum disparity) to numDisparities. The search range can then be 01649 shifted by changing the minimum disparity. 01650 @param blockSize the linear size of the blocks compared by the algorithm. The size should be odd 01651 (as the block is centered at the current pixel). Larger block size implies smoother, though less 01652 accurate disparity map. Smaller block size gives more detailed disparity map, but there is higher 01653 chance for algorithm to find a wrong correspondence. 01654 01655 The function create StereoBM object. You can then call StereoBM::compute() to compute disparity for 01656 a specific stereo pair. 01657 */ 01658 CV_WRAP static Ptr<StereoBM> create(int numDisparities = 0, int blockSize = 21); 01659 }; 01660 01661 /** @brief The class implements the modified H. Hirschmuller algorithm @cite HH08 that differs from the original 01662 one as follows: 01663 01664 - By default, the algorithm is single-pass, which means that you consider only 5 directions 01665 instead of 8. Set mode=StereoSGBM::MODE_HH in createStereoSGBM to run the full variant of the 01666 algorithm but beware that it may consume a lot of memory. 01667 - The algorithm matches blocks, not individual pixels. Though, setting blockSize=1 reduces the 01668 blocks to single pixels. 01669 - Mutual information cost function is not implemented. Instead, a simpler Birchfield-Tomasi 01670 sub-pixel metric from @cite BT98 is used. Though, the color images are supported as well. 01671 - Some pre- and post- processing steps from K. Konolige algorithm StereoBM are included, for 01672 example: pre-filtering (StereoBM::PREFILTER_XSOBEL type) and post-filtering (uniqueness 01673 check, quadratic interpolation and speckle filtering). 01674 01675 @note 01676 - (Python) An example illustrating the use of the StereoSGBM matching algorithm can be found 01677 at opencv_source_code/samples/python/stereo_match.py 01678 */ 01679 class CV_EXPORTS_W StereoSGBM : public StereoMatcher 01680 { 01681 public: 01682 enum 01683 { 01684 MODE_SGBM = 0, 01685 MODE_HH = 1, 01686 MODE_SGBM_3WAY = 2 01687 }; 01688 01689 CV_WRAP virtual int getPreFilterCap() const = 0; 01690 CV_WRAP virtual void setPreFilterCap(int preFilterCap) = 0; 01691 01692 CV_WRAP virtual int getUniquenessRatio() const = 0; 01693 CV_WRAP virtual void setUniquenessRatio(int uniquenessRatio) = 0; 01694 01695 CV_WRAP virtual int getP1() const = 0; 01696 CV_WRAP virtual void setP1(int P1) = 0; 01697 01698 CV_WRAP virtual int getP2() const = 0; 01699 CV_WRAP virtual void setP2(int P2) = 0; 01700 01701 CV_WRAP virtual int getMode() const = 0; 01702 CV_WRAP virtual void setMode(int mode) = 0; 01703 01704 /** @brief Creates StereoSGBM object 01705 01706 @param minDisparity Minimum possible disparity value. Normally, it is zero but sometimes 01707 rectification algorithms can shift images, so this parameter needs to be adjusted accordingly. 01708 @param numDisparities Maximum disparity minus minimum disparity. The value is always greater than 01709 zero. In the current implementation, this parameter must be divisible by 16. 01710 @param blockSize Matched block size. It must be an odd number >=1 . Normally, it should be 01711 somewhere in the 3..11 range. 01712 @param P1 The first parameter controlling the disparity smoothness. See below. 01713 @param P2 The second parameter controlling the disparity smoothness. The larger the values are, 01714 the smoother the disparity is. P1 is the penalty on the disparity change by plus or minus 1 01715 between neighbor pixels. P2 is the penalty on the disparity change by more than 1 between neighbor 01716 pixels. The algorithm requires P2 > P1 . See stereo_match.cpp sample where some reasonably good 01717 P1 and P2 values are shown (like 8\*number_of_image_channels\*SADWindowSize\*SADWindowSize and 01718 32\*number_of_image_channels\*SADWindowSize\*SADWindowSize , respectively). 01719 @param disp12MaxDiff Maximum allowed difference (in integer pixel units) in the left-right 01720 disparity check. Set it to a non-positive value to disable the check. 01721 @param preFilterCap Truncation value for the prefiltered image pixels. The algorithm first 01722 computes x-derivative at each pixel and clips its value by [-preFilterCap, preFilterCap] interval. 01723 The result values are passed to the Birchfield-Tomasi pixel cost function. 01724 @param uniquenessRatio Margin in percentage by which the best (minimum) computed cost function 01725 value should "win" the second best value to consider the found match correct. Normally, a value 01726 within the 5-15 range is good enough. 01727 @param speckleWindowSize Maximum size of smooth disparity regions to consider their noise speckles 01728 and invalidate. Set it to 0 to disable speckle filtering. Otherwise, set it somewhere in the 01729 50-200 range. 01730 @param speckleRange Maximum disparity variation within each connected component. If you do speckle 01731 filtering, set the parameter to a positive value, it will be implicitly multiplied by 16. 01732 Normally, 1 or 2 is good enough. 01733 @param mode Set it to StereoSGBM::MODE_HH to run the full-scale two-pass dynamic programming 01734 algorithm. It will consume O(W\*H\*numDisparities) bytes, which is large for 640x480 stereo and 01735 huge for HD-size pictures. By default, it is set to false . 01736 01737 The first constructor initializes StereoSGBM with all the default parameters. So, you only have to 01738 set StereoSGBM::numDisparities at minimum. The second constructor enables you to set each parameter 01739 to a custom value. 01740 */ 01741 CV_WRAP static Ptr<StereoSGBM> create(int minDisparity, int numDisparities, int blockSize, 01742 int P1 = 0, int P2 = 0, int disp12MaxDiff = 0, 01743 int preFilterCap = 0, int uniquenessRatio = 0, 01744 int speckleWindowSize = 0, int speckleRange = 0, 01745 int mode = StereoSGBM::MODE_SGBM); 01746 }; 01747 01748 //! @} calib3d 01749 01750 /** @brief The methods in this namespace use a so-called fisheye camera model. 01751 @ingroup calib3d_fisheye 01752 */ 01753 namespace fisheye 01754 { 01755 //! @addtogroup calib3d_fisheye 01756 //! @{ 01757 01758 enum{ 01759 CALIB_USE_INTRINSIC_GUESS = 1, 01760 CALIB_RECOMPUTE_EXTRINSIC = 2, 01761 CALIB_CHECK_COND = 4, 01762 CALIB_FIX_SKEW = 8, 01763 CALIB_FIX_K1 = 16, 01764 CALIB_FIX_K2 = 32, 01765 CALIB_FIX_K3 = 64, 01766 CALIB_FIX_K4 = 128, 01767 CALIB_FIX_INTRINSIC = 256 01768 }; 01769 01770 /** @brief Projects points using fisheye model 01771 01772 @param objectPoints Array of object points, 1xN/Nx1 3-channel (or vector<Point3f> ), where N is 01773 the number of points in the view. 01774 @param imagePoints Output array of image points, 2xN/Nx2 1-channel or 1xN/Nx1 2-channel, or 01775 vector<Point2f>. 01776 @param affine 01777 @param K Camera matrix \f$K = \vecthreethree{f_x}{0}{c_x}{0}{f_y}{c_y}{0}{0}{_1}\f$. 01778 @param D Input vector of distortion coefficients \f$(k_1, k_2, k_3, k_4)\f$. 01779 @param alpha The skew coefficient. 01780 @param jacobian Optional output 2Nx15 jacobian matrix of derivatives of image points with respect 01781 to components of the focal lengths, coordinates of the principal point, distortion coefficients, 01782 rotation vector, translation vector, and the skew. In the old interface different components of 01783 the jacobian are returned via different output parameters. 01784 01785 The function computes projections of 3D points to the image plane given intrinsic and extrinsic 01786 camera parameters. Optionally, the function computes Jacobians - matrices of partial derivatives of 01787 image points coordinates (as functions of all the input parameters) with respect to the particular 01788 parameters, intrinsic and/or extrinsic. 01789 */ 01790 CV_EXPORTS void projectPoints(InputArray objectPoints, OutputArray imagePoints, const Affine3d& affine, 01791 InputArray K, InputArray D, double alpha = 0, OutputArray jacobian = noArray()); 01792 01793 /** @overload */ 01794 CV_EXPORTS_W void projectPoints(InputArray objectPoints, OutputArray imagePoints, InputArray rvec, InputArray tvec, 01795 InputArray K, InputArray D, double alpha = 0, OutputArray jacobian = noArray()); 01796 01797 /** @brief Distorts 2D points using fisheye model. 01798 01799 @param undistorted Array of object points, 1xN/Nx1 2-channel (or vector<Point2f> ), where N is 01800 the number of points in the view. 01801 @param K Camera matrix \f$K = \vecthreethree{f_x}{0}{c_x}{0}{f_y}{c_y}{0}{0}{_1}\f$. 01802 @param D Input vector of distortion coefficients \f$(k_1, k_2, k_3, k_4)\f$. 01803 @param alpha The skew coefficient. 01804 @param distorted Output array of image points, 1xN/Nx1 2-channel, or vector<Point2f> . 01805 */ 01806 CV_EXPORTS_W void distortPoints(InputArray undistorted, OutputArray distorted, InputArray K, InputArray D, double alpha = 0); 01807 01808 /** @brief Undistorts 2D points using fisheye model 01809 01810 @param distorted Array of object points, 1xN/Nx1 2-channel (or vector<Point2f> ), where N is the 01811 number of points in the view. 01812 @param K Camera matrix \f$K = \vecthreethree{f_x}{0}{c_x}{0}{f_y}{c_y}{0}{0}{_1}\f$. 01813 @param D Input vector of distortion coefficients \f$(k_1, k_2, k_3, k_4)\f$. 01814 @param R Rectification transformation in the object space: 3x3 1-channel, or vector: 3x1/1x3 01815 1-channel or 1x1 3-channel 01816 @param P New camera matrix (3x3) or new projection matrix (3x4) 01817 @param undistorted Output array of image points, 1xN/Nx1 2-channel, or vector<Point2f> . 01818 */ 01819 CV_EXPORTS_W void undistortPoints(InputArray distorted, OutputArray undistorted, 01820 InputArray K, InputArray D, InputArray R = noArray(), InputArray P = noArray()); 01821 01822 /** @brief Computes undistortion and rectification maps for image transform by cv::remap(). If D is empty zero 01823 distortion is used, if R or P is empty identity matrixes are used. 01824 01825 @param K Camera matrix \f$K = \vecthreethree{f_x}{0}{c_x}{0}{f_y}{c_y}{0}{0}{_1}\f$. 01826 @param D Input vector of distortion coefficients \f$(k_1, k_2, k_3, k_4)\f$. 01827 @param R Rectification transformation in the object space: 3x3 1-channel, or vector: 3x1/1x3 01828 1-channel or 1x1 3-channel 01829 @param P New camera matrix (3x3) or new projection matrix (3x4) 01830 @param size Undistorted image size. 01831 @param m1type Type of the first output map that can be CV_32FC1 or CV_16SC2 . See convertMaps() 01832 for details. 01833 @param map1 The first output map. 01834 @param map2 The second output map. 01835 */ 01836 CV_EXPORTS_W void initUndistortRectifyMap(InputArray K, InputArray D, InputArray R, InputArray P, 01837 const cv::Size& size, int m1type, OutputArray map1, OutputArray map2); 01838 01839 /** @brief Transforms an image to compensate for fisheye lens distortion. 01840 01841 @param distorted image with fisheye lens distortion. 01842 @param undistorted Output image with compensated fisheye lens distortion. 01843 @param K Camera matrix \f$K = \vecthreethree{f_x}{0}{c_x}{0}{f_y}{c_y}{0}{0}{_1}\f$. 01844 @param D Input vector of distortion coefficients \f$(k_1, k_2, k_3, k_4)\f$. 01845 @param Knew Camera matrix of the distorted image. By default, it is the identity matrix but you 01846 may additionally scale and shift the result by using a different matrix. 01847 @param new_size 01848 01849 The function transforms an image to compensate radial and tangential lens distortion. 01850 01851 The function is simply a combination of fisheye::initUndistortRectifyMap (with unity R ) and remap 01852 (with bilinear interpolation). See the former function for details of the transformation being 01853 performed. 01854 01855 See below the results of undistortImage. 01856 - a\) result of undistort of perspective camera model (all possible coefficients (k_1, k_2, k_3, 01857 k_4, k_5, k_6) of distortion were optimized under calibration) 01858 - b\) result of fisheye::undistortImage of fisheye camera model (all possible coefficients (k_1, k_2, 01859 k_3, k_4) of fisheye distortion were optimized under calibration) 01860 - c\) original image was captured with fisheye lens 01861 01862 Pictures a) and b) almost the same. But if we consider points of image located far from the center 01863 of image, we can notice that on image a) these points are distorted. 01864 01865  01866 */ 01867 CV_EXPORTS_W void undistortImage(InputArray distorted, OutputArray undistorted, 01868 InputArray K, InputArray D, InputArray Knew = cv::noArray(), const Size& new_size = Size()); 01869 01870 /** @brief Estimates new camera matrix for undistortion or rectification. 01871 01872 @param K Camera matrix \f$K = \vecthreethree{f_x}{0}{c_x}{0}{f_y}{c_y}{0}{0}{_1}\f$. 01873 @param image_size 01874 @param D Input vector of distortion coefficients \f$(k_1, k_2, k_3, k_4)\f$. 01875 @param R Rectification transformation in the object space: 3x3 1-channel, or vector: 3x1/1x3 01876 1-channel or 1x1 3-channel 01877 @param P New camera matrix (3x3) or new projection matrix (3x4) 01878 @param balance Sets the new focal length in range between the min focal length and the max focal 01879 length. Balance is in range of [0, 1]. 01880 @param new_size 01881 @param fov_scale Divisor for new focal length. 01882 */ 01883 CV_EXPORTS_W void estimateNewCameraMatrixForUndistortRectify(InputArray K, InputArray D, const Size &image_size, InputArray R, 01884 OutputArray P, double balance = 0.0, const Size& new_size = Size(), double fov_scale = 1.0); 01885 01886 /** @brief Performs camera calibaration 01887 01888 @param objectPoints vector of vectors of calibration pattern points in the calibration pattern 01889 coordinate space. 01890 @param imagePoints vector of vectors of the projections of calibration pattern points. 01891 imagePoints.size() and objectPoints.size() and imagePoints[i].size() must be equal to 01892 objectPoints[i].size() for each i. 01893 @param image_size Size of the image used only to initialize the intrinsic camera matrix. 01894 @param K Output 3x3 floating-point camera matrix 01895 \f$A = \vecthreethree{f_x}{0}{c_x}{0}{f_y}{c_y}{0}{0}{1}\f$ . If 01896 fisheye::CALIB_USE_INTRINSIC_GUESS/ is specified, some or all of fx, fy, cx, cy must be 01897 initialized before calling the function. 01898 @param D Output vector of distortion coefficients \f$(k_1, k_2, k_3, k_4)\f$. 01899 @param rvecs Output vector of rotation vectors (see Rodrigues ) estimated for each pattern view. 01900 That is, each k-th rotation vector together with the corresponding k-th translation vector (see 01901 the next output parameter description) brings the calibration pattern from the model coordinate 01902 space (in which object points are specified) to the world coordinate space, that is, a real 01903 position of the calibration pattern in the k-th pattern view (k=0.. *M* -1). 01904 @param tvecs Output vector of translation vectors estimated for each pattern view. 01905 @param flags Different flags that may be zero or a combination of the following values: 01906 - **fisheye::CALIB_USE_INTRINSIC_GUESS** cameraMatrix contains valid initial values of 01907 fx, fy, cx, cy that are optimized further. Otherwise, (cx, cy) is initially set to the image 01908 center ( imageSize is used), and focal distances are computed in a least-squares fashion. 01909 - **fisheye::CALIB_RECOMPUTE_EXTRINSIC** Extrinsic will be recomputed after each iteration 01910 of intrinsic optimization. 01911 - **fisheye::CALIB_CHECK_COND** The functions will check validity of condition number. 01912 - **fisheye::CALIB_FIX_SKEW** Skew coefficient (alpha) is set to zero and stay zero. 01913 - **fisheye::CALIB_FIX_K1..4** Selected distortion coefficients are set to zeros and stay 01914 zero. 01915 @param criteria Termination criteria for the iterative optimization algorithm. 01916 */ 01917 CV_EXPORTS_W double calibrate(InputArrayOfArrays objectPoints, InputArrayOfArrays imagePoints, const Size& image_size, 01918 InputOutputArray K, InputOutputArray D, OutputArrayOfArrays rvecs, OutputArrayOfArrays tvecs, int flags = 0, 01919 TermCriteria criteria = TermCriteria(TermCriteria::COUNT + TermCriteria::EPS, 100, DBL_EPSILON)); 01920 01921 /** @brief Stereo rectification for fisheye camera model 01922 01923 @param K1 First camera matrix. 01924 @param D1 First camera distortion parameters. 01925 @param K2 Second camera matrix. 01926 @param D2 Second camera distortion parameters. 01927 @param imageSize Size of the image used for stereo calibration. 01928 @param R Rotation matrix between the coordinate systems of the first and the second 01929 cameras. 01930 @param tvec Translation vector between coordinate systems of the cameras. 01931 @param R1 Output 3x3 rectification transform (rotation matrix) for the first camera. 01932 @param R2 Output 3x3 rectification transform (rotation matrix) for the second camera. 01933 @param P1 Output 3x4 projection matrix in the new (rectified) coordinate systems for the first 01934 camera. 01935 @param P2 Output 3x4 projection matrix in the new (rectified) coordinate systems for the second 01936 camera. 01937 @param Q Output \f$4 \times 4\f$ disparity-to-depth mapping matrix (see reprojectImageTo3D ). 01938 @param flags Operation flags that may be zero or CV_CALIB_ZERO_DISPARITY . If the flag is set, 01939 the function makes the principal points of each camera have the same pixel coordinates in the 01940 rectified views. And if the flag is not set, the function may still shift the images in the 01941 horizontal or vertical direction (depending on the orientation of epipolar lines) to maximize the 01942 useful image area. 01943 @param newImageSize New image resolution after rectification. The same size should be passed to 01944 initUndistortRectifyMap (see the stereo_calib.cpp sample in OpenCV samples directory). When (0,0) 01945 is passed (default), it is set to the original imageSize . Setting it to larger value can help you 01946 preserve details in the original image, especially when there is a big radial distortion. 01947 @param balance Sets the new focal length in range between the min focal length and the max focal 01948 length. Balance is in range of [0, 1]. 01949 @param fov_scale Divisor for new focal length. 01950 */ 01951 CV_EXPORTS_W void stereoRectify(InputArray K1, InputArray D1, InputArray K2, InputArray D2, const Size &imageSize, InputArray R, InputArray tvec, 01952 OutputArray R1, OutputArray R2, OutputArray P1, OutputArray P2, OutputArray Q, int flags, const Size &newImageSize = Size(), 01953 double balance = 0.0, double fov_scale = 1.0); 01954 01955 /** @brief Performs stereo calibration 01956 01957 @param objectPoints Vector of vectors of the calibration pattern points. 01958 @param imagePoints1 Vector of vectors of the projections of the calibration pattern points, 01959 observed by the first camera. 01960 @param imagePoints2 Vector of vectors of the projections of the calibration pattern points, 01961 observed by the second camera. 01962 @param K1 Input/output first camera matrix: 01963 \f$\vecthreethree{f_x^{(j)}}{0}{c_x^{(j)}}{0}{f_y^{(j)}}{c_y^{(j)}}{0}{0}{1}\f$ , \f$j = 0,\, 1\f$ . If 01964 any of fisheye::CALIB_USE_INTRINSIC_GUESS , fisheye::CV_CALIB_FIX_INTRINSIC are specified, 01965 some or all of the matrix components must be initialized. 01966 @param D1 Input/output vector of distortion coefficients \f$(k_1, k_2, k_3, k_4)\f$ of 4 elements. 01967 @param K2 Input/output second camera matrix. The parameter is similar to K1 . 01968 @param D2 Input/output lens distortion coefficients for the second camera. The parameter is 01969 similar to D1 . 01970 @param imageSize Size of the image used only to initialize intrinsic camera matrix. 01971 @param R Output rotation matrix between the 1st and the 2nd camera coordinate systems. 01972 @param T Output translation vector between the coordinate systems of the cameras. 01973 @param flags Different flags that may be zero or a combination of the following values: 01974 - **fisheye::CV_CALIB_FIX_INTRINSIC** Fix K1, K2? and D1, D2? so that only R, T matrices 01975 are estimated. 01976 - **fisheye::CALIB_USE_INTRINSIC_GUESS** K1, K2 contains valid initial values of 01977 fx, fy, cx, cy that are optimized further. Otherwise, (cx, cy) is initially set to the image 01978 center (imageSize is used), and focal distances are computed in a least-squares fashion. 01979 - **fisheye::CALIB_RECOMPUTE_EXTRINSIC** Extrinsic will be recomputed after each iteration 01980 of intrinsic optimization. 01981 - **fisheye::CALIB_CHECK_COND** The functions will check validity of condition number. 01982 - **fisheye::CALIB_FIX_SKEW** Skew coefficient (alpha) is set to zero and stay zero. 01983 - **fisheye::CALIB_FIX_K1..4** Selected distortion coefficients are set to zeros and stay 01984 zero. 01985 @param criteria Termination criteria for the iterative optimization algorithm. 01986 */ 01987 CV_EXPORTS_W double stereoCalibrate(InputArrayOfArrays objectPoints, InputArrayOfArrays imagePoints1, InputArrayOfArrays imagePoints2, 01988 InputOutputArray K1, InputOutputArray D1, InputOutputArray K2, InputOutputArray D2, Size imageSize, 01989 OutputArray R, OutputArray T, int flags = fisheye::CALIB_FIX_INTRINSIC, 01990 TermCriteria criteria = TermCriteria(TermCriteria::COUNT + TermCriteria::EPS, 100, DBL_EPSILON)); 01991 01992 //! @} calib3d_fisheye 01993 } 01994 01995 } // cv 01996 01997 #ifndef DISABLE_OPENCV_24_COMPATIBILITY 01998 #include "opencv2/calib3d/calib3d_c.h" 01999 #endif 02000 02001 #endif 02002
Generated on Tue Jul 12 2022 16:42:37 by
1.7.2