As OCR technology develops, it expands to include more and more types of text. Initially limited to a few special fonts, it later grew to encompass all standard computer fonts, and is now used regularly to process even handwritten text. Thanks to these developments, many processes are being automated that could previously be done only by hand. However, due to differences in handwriting and the preference of some people to use cursive letters, character recognition for handwritten text remains far from perfect. While ICR has helped bring character recognition technology into the realm of handwriting, difficulties begin when the characters stray too far from standard printed text. New research in IWR (Intelligent Word Recognition) is part of an effort to make all text computer-recognizable. IWR is intended to read cursive text as well as printed text, so that all handwriting can be processed.
True to its name, IWR recognizes entire words, rather than breaking text into individual characters. Instead of trying to match each character to a pre-existing one in its database, IWR software searches for matching words, thus it relies on a dictionary containing all of the possible necessary words. This method of identifying text reduces errors associated with misreading individual letters, but is limited by the number of words in its database.
In other ways, the technology behind IWR is very similar to that behind ICR. The software also breaks words into a series of curves, shapes and lines, looking at them in groupings. It then assigns a confidence value to each word, depending on the clarity of the text and the likelihood of other matching possibilities.
IWR is not yet widely used because it lacks the accuracy needed in a consistent, reliable tool. More development is needed before this technology can be relied upon in day-to-day operations, but there is still very interesting potential. Combined into one software, OCR, ICR and IWR could automatically process any text document. This would further speed processing in many offices and eliminate even more manual data entry. For now, however, it remains an experimental area, requiring more work before it can become highly useful.
Intelligent Word Recognition Accuracy
Intelligent Word Recognition Software