给定形成网格的白线图像，如何使用opencv获取内部形状的蒙版？

Question

I am trying to determine if there are pills in different cells of a pillbox. I have a raspberry pi camera that takes pictures of a pillbox at regular intervals. I would like to use computer vision using opencv and python to determine which of the 14 cells have pills in them.

I have gotten as far as isolating the borders of the cells. How do I create opencv masks for the interiors of the cells?

I tried running hough transform on the processed image, but it does not accurately find lines corresponding to each wall

Original image:

After Processing with cell walls isolated:

Answer 1

You can try Hough Line detection one the image to detect line segments, and use the information to either close the gaps in your current result or simply detect the cells using only the line segments.

The advantage of using line detections is that once you have the parameters of the line, you can extend it below what actually appears in the image. So, even if a cell has a gap, you should be able to close it.