This week has been pretty good! I’ve been playing around with a bunch of different ways of thresholding a puzzle piece using a combination of methods and combinations of methods. These each seem to have their strengths but combining most of these methods with OTSU seems to generate helpful results. These should be solid enough to be able to extract which blob is the actual piece and the next step is to try to match up the piece to where it should fall on the image itself. Furthermore I’ve been playing around with the CLAHE (Contrast Limited AHE) method in order to normalize lighting conditions in the images, as we foresee this potentially being a problem as the user is moving around and depending on how glossy the puzzle ends up being. (we may have to use some light diffusing material like tissue paper to physically create a more even lighting condition on the pieces).
