Understanding OCR

One of the first application areas for machine vision was optical character recognition (OCR), sometimes confused with optical character verification (OCV.) What’s the difference, I hear you ask. Well, OCV is checking to see if a character is what was expected. The program expects to find an ‘E” in the image, so it compares the shape it found with the ‘E” in memory. If they match, then ‘E’ has been detected.

OCR is rather more complex. The computer has to compare the shape of the character in the image with a library of stored characters and determine which character it has found.

OK, that’s my rather clumsy explanation. To get a better understanding I suggest you read this excellent paper from Bruno Menard of Dalsa, as published in the May ’08 “Vision & Sensors” supplement to Quality Magazine.