I have an issue with detecting underscores with OCR. When I have a string of text with an underscore (i.e. ABC_EFG ) and Auto Detect Multiline and Rotation enabled, the OCR algorithms detect the underscore as being on the second line. If I turn off Auto Detect Multiline and Rotation, it detects the underscore properly as part of a single string of chars. Problem is, I need to detect multiple lines of strings like this. I've increased the Bounding Rect Height max size so that the bounding box should incorporate the pixels down to and including the underscore. But that doesn't appear to work.
I have an algorithm that can go through and get each line individually so I'm not dead in the water. But, I'd really like to take advantage of the speed realized by using the vision libraries ability to read multiple lines at a time.
Does anyone know if there is a way to tweak some ABC file settings to do this?
