measure of disparity
hello!
I'm studying in stereo vision using 2 webcams.
Now, I have some questions.
I want real depth(from webcams to object). So, I calibrated my webcams, then, I got disparity maps.
I use a formula that is depth = baseline * focal length / disparity
However, I got wrong depths.
So, here are some questions.
First, I think depth(cm) = baseline(cm)*focal length(pixel) / disparity(pixel) is right. I'm not quite sure that measure of disparity is pixel. I calibrate my webcams using by 1-inch chessboard(point to point = 1-inch). Then, the result of disparity is pixel or inch??
Next, I set a min disparity(4). It means that originally min disparity was increased. right? I think it can be affected to real depth. If so I would do not touch min disparity ...?
I think it is better you should the same metric for both calibration and depth estimation. One is centimeters and the other one is inch can cause problems. It can be a problem. However, I also try for getting 3D maps and exact(almost) locations. So I can only give you this to try. Also since you changed min disparity, it will effect what you see. You actually set the limit to see. So I don't believe it will have an effect on your calculations. And after stereo calibrations your focal length same for both cameras? If not, you can get the average. It may give you better and closer results I believe.