A question about Projection Transformation for Go Checkerboard

Hello everyone. I am trying to detect Go checkerboard and the chess manual from normal input image.

Input: Image contain one checkerboard. And the four corners of the checkerboard.

Output: Chess manual.

I want to warp the normal image first. Make the camera focus the center of checkerboard vertically. But I am not sure whether the input has enough information to get the homography matrix.

what is “chess manual”?

a “go” board looks like this:


please clarify.

Thanks for your replay! Sorry for my poor English. I just want to get the place of each chess piece. And the input images just are like the example provided by you (without the two bowls).

I find a blog that using cv::findHomography() to solve the problem and I am going to take a try.