Optical Character Recognition is a method of recognizing printed text and digitizing it so that it can be read by a computer. In the assistive technology landscape, OCR is primarily used by those who are blind, and thus cannot read printed text. However, it has many uses outside of assistive technology, including recognizing text on cheques, automatic license plate identification, and making large legal and other documents available to search engines.
The first optical character recognition machine that was widely available to the blind was created by Ray Kurzweil in the late 1970’s. Combining a newly invented flatbed scanner, optical recognition software, and the latest in text-to-speech hardware, it allowed any blind user to scan a printed document and have it read to them for the first time. Unfortunately, it was expensive, the recognition accuracy was relatively low, and the early text-to-speech voices were unpleasant to listen to due to their robotic and monotone sound.
But as personal computers became available, and consumer scanning technology improved, the state of OCR would quickly advance. The most popular OCR software for many years was developed by Kurzweil Educational Systems for Windows. It included a choice of two OCR recognition systems (OmniPage or FineReader), multiple text to speech voices to read out the recognized text, a suite of preprocessing techniques to improve the clarity of the scanned image, and many postprocessing algorithms to increase the accuracy of the recognized text.
As smartphones became available, the National Federation for the Blind (NFB) and Kurzweil would partner to develop KNFB, the first ever mobile text recognition software that could run on a smartphone. Originally released for the Symbian operating system, it was later ported to IOS. Today, systems that can recognize text in real-time (without the need to preprocess images) are common and can achieve accuracy of up to 99 percent.