I created a new image that that I feel is more representative of what my 6-state / 6 symbol State Transition Table will look like:
I revisited my idea of having a registration mark for each "Tile" when I realized that I could better identify the tile bounding boxes with contours alone. So here is what the binary image for the above looks like:
Now in order for this to work I needed to add a "dot" as a placeholder for the empty cells. I'm thinking these will be part of the "mat" that the tiles when placed will cover. The edge image:
So now when I use findContours I end up with the following bounding boxes around the tiles:
When I pass these sub images to Tesseract, based on the binary image above, I get the following result:
So the first image above will be printed to scale (when I can get to the color printer at kwartzlab), imaged with the Raspberry Pi camera, and I'll try this process with the photo. I expect that I will have to do considerably more working cleaning up the photo. We'll see.