Right now I’m working on a program to turn a regular image, and turn it into a template for a “paint by number”. I managed to get to the point that I can simplify the image into 20 different colors and change the RGB of each pixel in the original image to the closest match from a list of prominent colors, getting a great result. However, there are very small areas of pixels that would be unrealistic to paint by a human, so I was wondering what the best way to go about finding them was. Here’s an example with some of the problem colonies circled

I’ve looked at options such as OpenCV, but its too inaccurate. I haven’t tried doing it pixel by pixel yet, but I’m guessing that would be rather slow and inefficient as well. I’m not sure what the best method would be