OCR/OCV FAQs

Why is the step not properly recognizing the text inside my region of interest?

Before the OCR/OCV step can read characters in a region of interest, you need to train the software to recognize the characters. If you have not trained for the characters, click Edit Character Set File on the Main tab to launch the NI OCR Training Interface and train for the characters.

Make sure you selected a Character Set Path from the Main tab that describes the characters you want to read.

Also, make sure that each character in the region of interest is tightly bounded by a character bounding rectangle. If each character is not bounded, adjust the parameters in the Threshold and Size tabs. In particular, enable the Ignore Objects Touching Region Borders control in the Threshold tab if artifacts touch the character bounding rectangle. Also, adjust the Max Bounding Rect Width in the Size tab if the characters are slanted and slightly overlap in the vertical direction causing one character bounding rectangle to surround all of the characters. Adjusting this parameter forces the width of the character bounding rectangle.

What is the difference between the Classification Score and Identification Score?

The Classification Score indicates the level of confidence that a character belongs to a specific class. The Identification Score indicates how similar a character is to the reference character specified for the class.