it is need calibration camera chessboard, before rectified image can be shown like you do with disparity map?? because i have the same project, but the rectified image and disparity map can't show perfectly like yours.. help me please, i have a deadline during my project..
I tried to go further on the Learning Opencv's example of the chapter 12 by using the cvRepreojectImageto3D() function, but well, it seems that it only works when you feed the set of images from the sample file... I tried to feed my own set of image pairs but it doesn't work at all....could anyone please help me???
But why is the ceiling fan and the window appearing near white (very near) in the 3d view, while the calc and the pen are kind of grey (not much close)? Shouldn't it be the opposite?
Good question. My findings are that it has to do with the fact that minDisparity is negative (try to make it positive and you'll get white for closer objects). Couldn't find any decent documentation on BMState but you'll find some comments in cv.h, I've got to try the other disparity function cvFindStereoCorrespondence (now using cvFindStereoCorrespondenceBM that claims to be faster) .
keep in mind that numbers returned by cvFindStereoCorrespondenceBM are signed (CV_16S) , not sure if cvNormalize handle them correctly ... sorry a little busy now but will check when possible . if you find something let me know .
it is need calibration camera chessboard, before rectified image can be shown like you do with disparity map?? because i have the same project, but the rectified image and disparity map can't show perfectly like yours.. help me please, i have a deadline during my project..
domba0912 7 months ago
was this code implemented in windows or linux platform..
for me the code runs way awesomely in windows but doesn't in linux(ubuntu).
does anyone what is the reason for it??
SuperRickesh 7 months ago
Dose it write on Visual Studio?
VanDucTech 9 months ago
I tried to go further on the Learning Opencv's example of the chapter 12 by using the cvRepreojectImageto3D() function, but well, it seems that it only works when you feed the set of images from the sample file... I tried to feed my own set of image pairs but it doesn't work at all....could anyone please help me???
Thanks in advance!!!
diabelliforlife 1 year ago
what kind of frame rate are you getting?
MrTechnovision 2 years ago
Didn't measure it , but I would estimate about 5 fps. Factors that framerate depends on are:
- camera resolution
- camera framerate
- speed of your computer
- algorithm you use to find stereo correspondence
starlino 2 years ago
Congratulations!
But why is the ceiling fan and the window appearing near white (very near) in the 3d view, while the calc and the pen are kind of grey (not much close)? Shouldn't it be the opposite?
Thank you, congrats again!
dannyxyz22 2 years ago
Good question. My findings are that it has to do with the fact that minDisparity is negative (try to make it positive and you'll get white for closer objects). Couldn't find any decent documentation on BMState but you'll find some comments in cv.h, I've got to try the other disparity function cvFindStereoCorrespondence (now using cvFindStereoCorrespondenceBM that claims to be faster) .
starlino 2 years ago
Hum.. good catch... but I don't know... It wouldn't explain why the background is black... but I'll try to check the code.
dannyxyz22 2 years ago
keep in mind that numbers returned by cvFindStereoCorrespondenceBM are signed (CV_16S) , not sure if cvNormalize handle them correctly ... sorry a little busy now but will check when possible . if you find something let me know .
starlino 2 years ago
For all interested , source code and decription available on my site (Google "starlino").
starlino 2 years ago
Hi...starlino. I'm doing my master thesis using OpenCV and I have the same problem with disparity map. Any news about it?
LeoStorm85 2 years ago
PS. cvNormalize works properly...I've analyzed the source code of the library
LeoStorm85 2 years ago