Chessboard black square is NOT detected via contour?

edit

Hi, I'm testing some chessboard images using cv2.findContours(), using cnts = cv2.findContours(otsu, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE) .

I obtained the binary image via otsu binarization, and did some cv2.HoughLinesP experiments. So far, what I can obtain are as follows:

My question is: why ONLY the white squares/rectangles have been successfully detected, but those BLACK ones failed to be detected ?

Any possibilities?

Thank you ...