Tesseract is probably the most accurate open source OCR engine available. Combined with the Leptonica Image Processing Library it can read a wide variety of image formats and convert them to text in over 60 languages. It was one of the top 3 engines in the 1995 UNLV Accuracy test. Between 1995 and 2006 it had little work done on it, but since then it has been improved extensively by Google. It is released under the Apache License 2.0.
Tesseract works on Linux, Windows (with VC++ Express or CygWin) and Mac OSX. See the ReadMe for more details and install instructions. It can also be compiled for other platforms, including Android and the iPhone, though these are not as well tested platforms. See also the AddOns page for other projects using Tesseract on various platforms.
If you're interested in supporting other platforms or languages, please get in touch with Ray Smith or the Developers.
With the discontinuation of downloads at code.google.com, new source downloads will be posted to GoogleDrive. Other download folders will be setup as new files are uploaded, and the original Downloads page will go away. During the transition, other downloads can still be found at the Old Downloads page.
Version 3.03 release candidate is now available (source only so far) for download and contains many new features. (See the ReleaseNotes for a full list.) Please check out the ReadMe before going to Downloads as you need more than one file. Even the windows executables tarball is incomplete as language files are required. Most notable new features:
The core developer on the project is Ray Smith (theraysmith).
In related work, Thomas Breuel (tmbdev) and Ilya Mezhirov (mezhirov) work on the OCRopus project, which also provides layout analysis and statistical language modeling.
Most of the work on Tesseract is sponsored by Google.