0

I am playing around with OpenCV. The task I made up for myself to make the process interesting is recognizing the chess board with random set of pieces on it, they can be placed between the squares as well or even on top of each other.

There are complications: the piece's colors are very close to chessboard's. The final goal is to mark intersections (know their coordinates) and fill the cells according to their color (black/white). With perspective support, shadows etc.

example

another example

red squares are created automatically by my current code. But it is not as good as I expected.

Any recommendations? What image processing algorithms?

Rauf Aliev
  • 25
  • 6
  • if your camera lens is undistorted you can compute every chessboard square position with homography, after you detected a single square well. Then you can test how well the rest of the squares are visible in the image. – Micka Jun 23 '18 at 10:47
  • thanks! I will try then! – Rauf Aliev Jun 25 '18 at 13:41

0 Answers0