#5 Trigrams / Dictionary / Interactive

[Alexander Th. Kloss]

Hi, there!

Are there any plans to improve the recognition by using Trigrams (like T9, trying to guess syllabes, could be done with statistics) or even Dictionaries (like a|i|pspell)? If it is possible to calculate the probability for every single letter, one could try to guess the more undecided ones.

Furthermore, it would be interesting to see an interactive mode of gocr (like launching the user's favourite image display with the snippet gocr is unable to read while asking what letter is meant).

My current test results show that gocr is almost unusable: less than 80% on the average. Since there is no real good free ocr package on linux, that's a sad picture. ocrad may show promise, but I believe gocr to have had the better start. Please show a lifesign :-)

Greetings, atk