I'm trying to design a script that can test functionality of LCDs by checking the individual digits that are displayed (I can cycle through several test displays, example image attached). I am new to vision so not sure of the best techniques to use.
I've had success doing this using NI Vision Assistant, in two methods, but both can be very unreliable.
1) create ROIs around each digit's expected location, and a template image of the digit. I used this originally but changed method as I thought that having individual ROI and template per digit would be increasing the likelihood of a missed match. The VA script included edge detection to determine the location of the LCD, coordinate systems as reference for the ROI locations, and then pattern matching against each individual template.
2) create single large ROI around the entire LCD, and a template of the entire LCD. Again using edge detection/coord system as reference points, and pattern match to determine a good LCD.
Both techniques can be unreliable and cause high numbers of false failures. Is there a better way to do this task, or any tips anyone has to imrpove reliability of the tests?
