SRL Archive - Suggestions

So I've been thinking… We should gather a lot of .bmp's of uptext, and save them somewhere, along with the text that *should* be detected. This way we can check how well the algorithm works, and perform some nice benchmarking, as well as comparisons with several algorithms.

I added a test "Suite". Projects/OCRBench. It features a simplistic GUI, which you give a BMP and a path to the UpChars. It will then show the four steps taken to filter, and it will also show the output of the BETA OCR. ('t' is still seen as 'l')