Final 350 project
Dependencies: uzair Camera_LS_Y201 F7_Ethernet LCD_DISCO_F746NG NetworkAPI SDFileSystem mbed
includes/jfdctflt.c@0:791a779d6220, 2017-07-31 (annotated)
- Committer:
- shoaib_ahmed
- Date:
- Mon Jul 31 09:16:35 2017 +0000
- Revision:
- 0:791a779d6220
final project;
Who changed what in which revision?
User | Revision | Line number | New contents of line |
---|---|---|---|
shoaib_ahmed | 0:791a779d6220 | 1 | /* |
shoaib_ahmed | 0:791a779d6220 | 2 | * jfdctflt.c |
shoaib_ahmed | 0:791a779d6220 | 3 | * |
shoaib_ahmed | 0:791a779d6220 | 4 | * Copyright (C) 1994-1996, Thomas G. Lane. |
shoaib_ahmed | 0:791a779d6220 | 5 | * Modified 2003-2015 by Guido Vollbeding. |
shoaib_ahmed | 0:791a779d6220 | 6 | * This file is part of the Independent JPEG Group's software. |
shoaib_ahmed | 0:791a779d6220 | 7 | * For conditions of distribution and use, see the accompanying README file. |
shoaib_ahmed | 0:791a779d6220 | 8 | * |
shoaib_ahmed | 0:791a779d6220 | 9 | * This file contains a floating-point implementation of the |
shoaib_ahmed | 0:791a779d6220 | 10 | * forward DCT (Discrete Cosine Transform). |
shoaib_ahmed | 0:791a779d6220 | 11 | * |
shoaib_ahmed | 0:791a779d6220 | 12 | * This implementation should be more accurate than either of the integer |
shoaib_ahmed | 0:791a779d6220 | 13 | * DCT implementations. However, it may not give the same results on all |
shoaib_ahmed | 0:791a779d6220 | 14 | * machines because of differences in roundoff behavior. Speed will depend |
shoaib_ahmed | 0:791a779d6220 | 15 | * on the hardware's floating point capacity. |
shoaib_ahmed | 0:791a779d6220 | 16 | * |
shoaib_ahmed | 0:791a779d6220 | 17 | * A 2-D DCT can be done by 1-D DCT on each row followed by 1-D DCT |
shoaib_ahmed | 0:791a779d6220 | 18 | * on each column. Direct algorithms are also available, but they are |
shoaib_ahmed | 0:791a779d6220 | 19 | * much more complex and seem not to be any faster when reduced to code. |
shoaib_ahmed | 0:791a779d6220 | 20 | * |
shoaib_ahmed | 0:791a779d6220 | 21 | * This implementation is based on Arai, Agui, and Nakajima's algorithm for |
shoaib_ahmed | 0:791a779d6220 | 22 | * scaled DCT. Their original paper (Trans. IEICE E-71(11):1095) is in |
shoaib_ahmed | 0:791a779d6220 | 23 | * Japanese, but the algorithm is described in the Pennebaker & Mitchell |
shoaib_ahmed | 0:791a779d6220 | 24 | * JPEG textbook (see REFERENCES section in file README). The following code |
shoaib_ahmed | 0:791a779d6220 | 25 | * is based directly on figure 4-8 in P&M. |
shoaib_ahmed | 0:791a779d6220 | 26 | * While an 8-point DCT cannot be done in less than 11 multiplies, it is |
shoaib_ahmed | 0:791a779d6220 | 27 | * possible to arrange the computation so that many of the multiplies are |
shoaib_ahmed | 0:791a779d6220 | 28 | * simple scalings of the final outputs. These multiplies can then be |
shoaib_ahmed | 0:791a779d6220 | 29 | * folded into the multiplications or divisions by the JPEG quantization |
shoaib_ahmed | 0:791a779d6220 | 30 | * table entries. The AA&N method leaves only 5 multiplies and 29 adds |
shoaib_ahmed | 0:791a779d6220 | 31 | * to be done in the DCT itself. |
shoaib_ahmed | 0:791a779d6220 | 32 | * The primary disadvantage of this method is that with a fixed-point |
shoaib_ahmed | 0:791a779d6220 | 33 | * implementation, accuracy is lost due to imprecise representation of the |
shoaib_ahmed | 0:791a779d6220 | 34 | * scaled quantization values. However, that problem does not arise if |
shoaib_ahmed | 0:791a779d6220 | 35 | * we use floating point arithmetic. |
shoaib_ahmed | 0:791a779d6220 | 36 | */ |
shoaib_ahmed | 0:791a779d6220 | 37 | |
shoaib_ahmed | 0:791a779d6220 | 38 | #define JPEG_INTERNALS |
shoaib_ahmed | 0:791a779d6220 | 39 | #include "jinclude.h" |
shoaib_ahmed | 0:791a779d6220 | 40 | #include "jpeglib.h" |
shoaib_ahmed | 0:791a779d6220 | 41 | #include "jdct.h" /* Private declarations for DCT subsystem */ |
shoaib_ahmed | 0:791a779d6220 | 42 | |
shoaib_ahmed | 0:791a779d6220 | 43 | #ifdef DCT_FLOAT_SUPPORTED |
shoaib_ahmed | 0:791a779d6220 | 44 | |
shoaib_ahmed | 0:791a779d6220 | 45 | |
shoaib_ahmed | 0:791a779d6220 | 46 | /* |
shoaib_ahmed | 0:791a779d6220 | 47 | * This module is specialized to the case DCTSIZE = 8. |
shoaib_ahmed | 0:791a779d6220 | 48 | */ |
shoaib_ahmed | 0:791a779d6220 | 49 | |
shoaib_ahmed | 0:791a779d6220 | 50 | #if DCTSIZE != 8 |
shoaib_ahmed | 0:791a779d6220 | 51 | Sorry, this code only copes with 8x8 DCTs. /* deliberate syntax err */ |
shoaib_ahmed | 0:791a779d6220 | 52 | #endif |
shoaib_ahmed | 0:791a779d6220 | 53 | |
shoaib_ahmed | 0:791a779d6220 | 54 | |
shoaib_ahmed | 0:791a779d6220 | 55 | /* |
shoaib_ahmed | 0:791a779d6220 | 56 | * Perform the forward DCT on one block of samples. |
shoaib_ahmed | 0:791a779d6220 | 57 | * |
shoaib_ahmed | 0:791a779d6220 | 58 | * cK represents cos(K*pi/16). |
shoaib_ahmed | 0:791a779d6220 | 59 | */ |
shoaib_ahmed | 0:791a779d6220 | 60 | |
shoaib_ahmed | 0:791a779d6220 | 61 | GLOBAL(void) |
shoaib_ahmed | 0:791a779d6220 | 62 | jpeg_fdct_float (FAST_FLOAT * data, JSAMPARRAY sample_data, JDIMENSION start_col) |
shoaib_ahmed | 0:791a779d6220 | 63 | { |
shoaib_ahmed | 0:791a779d6220 | 64 | FAST_FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7; |
shoaib_ahmed | 0:791a779d6220 | 65 | FAST_FLOAT tmp10, tmp11, tmp12, tmp13; |
shoaib_ahmed | 0:791a779d6220 | 66 | FAST_FLOAT z1, z2, z3, z4, z5, z11, z13; |
shoaib_ahmed | 0:791a779d6220 | 67 | FAST_FLOAT *dataptr; |
shoaib_ahmed | 0:791a779d6220 | 68 | JSAMPROW elemptr; |
shoaib_ahmed | 0:791a779d6220 | 69 | int ctr; |
shoaib_ahmed | 0:791a779d6220 | 70 | |
shoaib_ahmed | 0:791a779d6220 | 71 | /* Pass 1: process rows. */ |
shoaib_ahmed | 0:791a779d6220 | 72 | |
shoaib_ahmed | 0:791a779d6220 | 73 | dataptr = data; |
shoaib_ahmed | 0:791a779d6220 | 74 | for (ctr = 0; ctr < DCTSIZE; ctr++) { |
shoaib_ahmed | 0:791a779d6220 | 75 | elemptr = sample_data[ctr] + start_col; |
shoaib_ahmed | 0:791a779d6220 | 76 | |
shoaib_ahmed | 0:791a779d6220 | 77 | /* Load data into workspace */ |
shoaib_ahmed | 0:791a779d6220 | 78 | tmp0 = (FAST_FLOAT) (GETJSAMPLE(elemptr[0]) + GETJSAMPLE(elemptr[7])); |
shoaib_ahmed | 0:791a779d6220 | 79 | tmp7 = (FAST_FLOAT) (GETJSAMPLE(elemptr[0]) - GETJSAMPLE(elemptr[7])); |
shoaib_ahmed | 0:791a779d6220 | 80 | tmp1 = (FAST_FLOAT) (GETJSAMPLE(elemptr[1]) + GETJSAMPLE(elemptr[6])); |
shoaib_ahmed | 0:791a779d6220 | 81 | tmp6 = (FAST_FLOAT) (GETJSAMPLE(elemptr[1]) - GETJSAMPLE(elemptr[6])); |
shoaib_ahmed | 0:791a779d6220 | 82 | tmp2 = (FAST_FLOAT) (GETJSAMPLE(elemptr[2]) + GETJSAMPLE(elemptr[5])); |
shoaib_ahmed | 0:791a779d6220 | 83 | tmp5 = (FAST_FLOAT) (GETJSAMPLE(elemptr[2]) - GETJSAMPLE(elemptr[5])); |
shoaib_ahmed | 0:791a779d6220 | 84 | tmp3 = (FAST_FLOAT) (GETJSAMPLE(elemptr[3]) + GETJSAMPLE(elemptr[4])); |
shoaib_ahmed | 0:791a779d6220 | 85 | tmp4 = (FAST_FLOAT) (GETJSAMPLE(elemptr[3]) - GETJSAMPLE(elemptr[4])); |
shoaib_ahmed | 0:791a779d6220 | 86 | |
shoaib_ahmed | 0:791a779d6220 | 87 | /* Even part */ |
shoaib_ahmed | 0:791a779d6220 | 88 | |
shoaib_ahmed | 0:791a779d6220 | 89 | tmp10 = tmp0 + tmp3; /* phase 2 */ |
shoaib_ahmed | 0:791a779d6220 | 90 | tmp13 = tmp0 - tmp3; |
shoaib_ahmed | 0:791a779d6220 | 91 | tmp11 = tmp1 + tmp2; |
shoaib_ahmed | 0:791a779d6220 | 92 | tmp12 = tmp1 - tmp2; |
shoaib_ahmed | 0:791a779d6220 | 93 | |
shoaib_ahmed | 0:791a779d6220 | 94 | /* Apply unsigned->signed conversion. */ |
shoaib_ahmed | 0:791a779d6220 | 95 | dataptr[0] = tmp10 + tmp11 - 8 * CENTERJSAMPLE; /* phase 3 */ |
shoaib_ahmed | 0:791a779d6220 | 96 | dataptr[4] = tmp10 - tmp11; |
shoaib_ahmed | 0:791a779d6220 | 97 | |
shoaib_ahmed | 0:791a779d6220 | 98 | z1 = (tmp12 + tmp13) * ((FAST_FLOAT) 0.707106781); /* c4 */ |
shoaib_ahmed | 0:791a779d6220 | 99 | dataptr[2] = tmp13 + z1; /* phase 5 */ |
shoaib_ahmed | 0:791a779d6220 | 100 | dataptr[6] = tmp13 - z1; |
shoaib_ahmed | 0:791a779d6220 | 101 | |
shoaib_ahmed | 0:791a779d6220 | 102 | /* Odd part */ |
shoaib_ahmed | 0:791a779d6220 | 103 | |
shoaib_ahmed | 0:791a779d6220 | 104 | tmp10 = tmp4 + tmp5; /* phase 2 */ |
shoaib_ahmed | 0:791a779d6220 | 105 | tmp11 = tmp5 + tmp6; |
shoaib_ahmed | 0:791a779d6220 | 106 | tmp12 = tmp6 + tmp7; |
shoaib_ahmed | 0:791a779d6220 | 107 | |
shoaib_ahmed | 0:791a779d6220 | 108 | /* The rotator is modified from fig 4-8 to avoid extra negations. */ |
shoaib_ahmed | 0:791a779d6220 | 109 | z5 = (tmp10 - tmp12) * ((FAST_FLOAT) 0.382683433); /* c6 */ |
shoaib_ahmed | 0:791a779d6220 | 110 | z2 = ((FAST_FLOAT) 0.541196100) * tmp10 + z5; /* c2-c6 */ |
shoaib_ahmed | 0:791a779d6220 | 111 | z4 = ((FAST_FLOAT) 1.306562965) * tmp12 + z5; /* c2+c6 */ |
shoaib_ahmed | 0:791a779d6220 | 112 | z3 = tmp11 * ((FAST_FLOAT) 0.707106781); /* c4 */ |
shoaib_ahmed | 0:791a779d6220 | 113 | |
shoaib_ahmed | 0:791a779d6220 | 114 | z11 = tmp7 + z3; /* phase 5 */ |
shoaib_ahmed | 0:791a779d6220 | 115 | z13 = tmp7 - z3; |
shoaib_ahmed | 0:791a779d6220 | 116 | |
shoaib_ahmed | 0:791a779d6220 | 117 | dataptr[5] = z13 + z2; /* phase 6 */ |
shoaib_ahmed | 0:791a779d6220 | 118 | dataptr[3] = z13 - z2; |
shoaib_ahmed | 0:791a779d6220 | 119 | dataptr[1] = z11 + z4; |
shoaib_ahmed | 0:791a779d6220 | 120 | dataptr[7] = z11 - z4; |
shoaib_ahmed | 0:791a779d6220 | 121 | |
shoaib_ahmed | 0:791a779d6220 | 122 | dataptr += DCTSIZE; /* advance pointer to next row */ |
shoaib_ahmed | 0:791a779d6220 | 123 | } |
shoaib_ahmed | 0:791a779d6220 | 124 | |
shoaib_ahmed | 0:791a779d6220 | 125 | /* Pass 2: process columns. */ |
shoaib_ahmed | 0:791a779d6220 | 126 | |
shoaib_ahmed | 0:791a779d6220 | 127 | dataptr = data; |
shoaib_ahmed | 0:791a779d6220 | 128 | for (ctr = DCTSIZE-1; ctr >= 0; ctr--) { |
shoaib_ahmed | 0:791a779d6220 | 129 | tmp0 = dataptr[DCTSIZE*0] + dataptr[DCTSIZE*7]; |
shoaib_ahmed | 0:791a779d6220 | 130 | tmp7 = dataptr[DCTSIZE*0] - dataptr[DCTSIZE*7]; |
shoaib_ahmed | 0:791a779d6220 | 131 | tmp1 = dataptr[DCTSIZE*1] + dataptr[DCTSIZE*6]; |
shoaib_ahmed | 0:791a779d6220 | 132 | tmp6 = dataptr[DCTSIZE*1] - dataptr[DCTSIZE*6]; |
shoaib_ahmed | 0:791a779d6220 | 133 | tmp2 = dataptr[DCTSIZE*2] + dataptr[DCTSIZE*5]; |
shoaib_ahmed | 0:791a779d6220 | 134 | tmp5 = dataptr[DCTSIZE*2] - dataptr[DCTSIZE*5]; |
shoaib_ahmed | 0:791a779d6220 | 135 | tmp3 = dataptr[DCTSIZE*3] + dataptr[DCTSIZE*4]; |
shoaib_ahmed | 0:791a779d6220 | 136 | tmp4 = dataptr[DCTSIZE*3] - dataptr[DCTSIZE*4]; |
shoaib_ahmed | 0:791a779d6220 | 137 | |
shoaib_ahmed | 0:791a779d6220 | 138 | /* Even part */ |
shoaib_ahmed | 0:791a779d6220 | 139 | |
shoaib_ahmed | 0:791a779d6220 | 140 | tmp10 = tmp0 + tmp3; /* phase 2 */ |
shoaib_ahmed | 0:791a779d6220 | 141 | tmp13 = tmp0 - tmp3; |
shoaib_ahmed | 0:791a779d6220 | 142 | tmp11 = tmp1 + tmp2; |
shoaib_ahmed | 0:791a779d6220 | 143 | tmp12 = tmp1 - tmp2; |
shoaib_ahmed | 0:791a779d6220 | 144 | |
shoaib_ahmed | 0:791a779d6220 | 145 | dataptr[DCTSIZE*0] = tmp10 + tmp11; /* phase 3 */ |
shoaib_ahmed | 0:791a779d6220 | 146 | dataptr[DCTSIZE*4] = tmp10 - tmp11; |
shoaib_ahmed | 0:791a779d6220 | 147 | |
shoaib_ahmed | 0:791a779d6220 | 148 | z1 = (tmp12 + tmp13) * ((FAST_FLOAT) 0.707106781); /* c4 */ |
shoaib_ahmed | 0:791a779d6220 | 149 | dataptr[DCTSIZE*2] = tmp13 + z1; /* phase 5 */ |
shoaib_ahmed | 0:791a779d6220 | 150 | dataptr[DCTSIZE*6] = tmp13 - z1; |
shoaib_ahmed | 0:791a779d6220 | 151 | |
shoaib_ahmed | 0:791a779d6220 | 152 | /* Odd part */ |
shoaib_ahmed | 0:791a779d6220 | 153 | |
shoaib_ahmed | 0:791a779d6220 | 154 | tmp10 = tmp4 + tmp5; /* phase 2 */ |
shoaib_ahmed | 0:791a779d6220 | 155 | tmp11 = tmp5 + tmp6; |
shoaib_ahmed | 0:791a779d6220 | 156 | tmp12 = tmp6 + tmp7; |
shoaib_ahmed | 0:791a779d6220 | 157 | |
shoaib_ahmed | 0:791a779d6220 | 158 | /* The rotator is modified from fig 4-8 to avoid extra negations. */ |
shoaib_ahmed | 0:791a779d6220 | 159 | z5 = (tmp10 - tmp12) * ((FAST_FLOAT) 0.382683433); /* c6 */ |
shoaib_ahmed | 0:791a779d6220 | 160 | z2 = ((FAST_FLOAT) 0.541196100) * tmp10 + z5; /* c2-c6 */ |
shoaib_ahmed | 0:791a779d6220 | 161 | z4 = ((FAST_FLOAT) 1.306562965) * tmp12 + z5; /* c2+c6 */ |
shoaib_ahmed | 0:791a779d6220 | 162 | z3 = tmp11 * ((FAST_FLOAT) 0.707106781); /* c4 */ |
shoaib_ahmed | 0:791a779d6220 | 163 | |
shoaib_ahmed | 0:791a779d6220 | 164 | z11 = tmp7 + z3; /* phase 5 */ |
shoaib_ahmed | 0:791a779d6220 | 165 | z13 = tmp7 - z3; |
shoaib_ahmed | 0:791a779d6220 | 166 | |
shoaib_ahmed | 0:791a779d6220 | 167 | dataptr[DCTSIZE*5] = z13 + z2; /* phase 6 */ |
shoaib_ahmed | 0:791a779d6220 | 168 | dataptr[DCTSIZE*3] = z13 - z2; |
shoaib_ahmed | 0:791a779d6220 | 169 | dataptr[DCTSIZE*1] = z11 + z4; |
shoaib_ahmed | 0:791a779d6220 | 170 | dataptr[DCTSIZE*7] = z11 - z4; |
shoaib_ahmed | 0:791a779d6220 | 171 | |
shoaib_ahmed | 0:791a779d6220 | 172 | dataptr++; /* advance pointer to next column */ |
shoaib_ahmed | 0:791a779d6220 | 173 | } |
shoaib_ahmed | 0:791a779d6220 | 174 | } |
shoaib_ahmed | 0:791a779d6220 | 175 | |
shoaib_ahmed | 0:791a779d6220 | 176 | #endif /* DCT_FLOAT_SUPPORTED */ |