10 years, 2 months ago.

Image processing in mbed using LPC1768

I am new in mbed. i want to capture a image using a camera then i need to process the captured image. The image will contain a laser line. I need to scan the whole image and then i need to tell that the image has a straight line of laser. Normally when i work with image in Android, the image is first converted as bitmap (considered the image as a matrix) then i scanned row and column to each pixel . What about in mbed??

2 Answers

10 years, 2 months ago.

Hello,

there were similar questions asked on mbed, like this one : https://mbed.org/questions/2576/Can-LPC-1768-be-used-for-image-processin/. Use search functionality, to find them and they might provide an answer to your question.

Regards,
0xc0170

10 years, 2 months ago.

Hi,

Yes you can do some basic image processing on mbed - I did it for ERIC here. The problem you will have is memory. I used 160*120 pixel images which worked OK in RAM but any bigger and you will need to use external memory will be much slower. I was looking for circles in an image but I did play around with a Hough transform that would return lines.

Regards
Martin

I cant understand some matters. My image size is 160*120 pixels. When i transfer the image into microcontroller it transfers like ff d8 ff e0 00 10 ...........40 c6 81 90 b9 aa ee 68 19 ff . What does it mean?? how i determine a pixel with these?? how can i check a pixels value whether it match with red color or not??

posted by imon bayazid 01 Sep 2014

Well it depends on what colour depth you have. The camera I used captured pictures in 16-bit RGB565. This means that each pixel is spread over 2 bytes withe the 5 MSBs being the red saturation, the next 6 bring green (the human eye is more sensitive to green) and the 5 LSBs being the blue saturation. I used bitwise operators to strip each colour from the 2 bytes. So for example if the 2 bytes were 0xF8 0x00 then the pixel would be pure red, 0x00 0x00 would be black and 0xFF 0xFF would be white. Does that help?

posted by Martin Smith 02 Sep 2014