stereolabs/zed-open-capture

[Question] Is the result of algorithm in OpenCV stereo the same as the result in ZED SDK?

violet17 opened this issue · 3 comments

violet17 commented

I want to use ZED camera to compute depth map without discrete GPU.
According to the issue, I want to get more detail about computing depth map.

Thanks.

Myzhar commented

Hi @violet17
the starting point for you is the rectify example.
Once you have the rectified images right_rect and left_rect you can directly use them with one of the OpenCV stereo matcher to obtain the disparity map:
https://docs.opencv.org/trunk/dd/d86/group__stereo.html

This is an example:
https://resources.oreilly.com/examples/9781786469717/blob/master/OpenCV3ComputerVisionApplicationProgrammingCookbookThirdEdition_Code/Chapter11/stereoMatcher.cpp#L144-173

violet17 commented

@Myzhar Thanks for the solution. I will have a try.

violet17 commented

Hi @violet17
the starting point for you is the rectify example.
Once you have the rectified images right_rect and left_rect you can directly use them with one of the OpenCV stereo matcher to obtain the disparity map:
https://docs.opencv.org/trunk/dd/d86/group__stereo.html

This is an example:
https://resources.oreilly.com/examples/9781786469717/blob/master/OpenCV3ComputerVisionApplicationProgrammingCookbookThirdEdition_Code/Chapter11/stereoMatcher.cpp#L144-173

Hi @Myzhar , Can I directly compute the depth map using this formula： Z = B*f/disparity? Do I need to add some scales?
How to get the value of distance of two cameras B?