Consider an example as shown below:
I ran imageProcess and I fail to get C (I also get a false positive "J", but this is okay). In some other images there are more true negatives. In general, is there any easy way to deal with this kind of image? If not, can I make the OCR more lenient so that I increase recall rate (sacrificing precision)?