I'm training characters, then verifying my program is able to read the same characters. Each item I'm reading has 8 of the same characters in different parts of the roi. Due to it being a curved surface, each position is being trained independently. In 3 of the posistions, I have trained the character '5' and it will recognize it with no problems in the ocr training program. Using the exact same roi, it will not read the characters in my program. All settings, ie Read resolution, read strategy, erosions, etc are the same in both programs. It finds the character, just doesn't know what it is. I'm using non linear thresholding. On the character report indicator, I took the threshold values it uses in my program, and entered those as a fixed range threshold on the ocr training. This turned the character blue just as expected and read with no problems. Unless there is a setting I'm missing, it seems the OCR Training Application is not using the same OCR Read Text 3 I'm using in my program. Any suggestions? I know this is complicated, I'll add the image and a screenshot showing what I mean
