After two years of development, the release of the Ocrad 0.29 text recognition system (Optical Character Recognition), developed under the auspices of the GNU project, has been completed. Ocrad can be used both as a library to integrate OCR functions into other applications and as a standalone utility that outputs text in UTF-8 or 8-bit encodings based on the provided input image.
Ocrad uses a feature extraction method for optical recognition. It includes a page layout analyzer that correctly separates columns and blocks of text in printed documents. Recognition is supported only for characters from the 'ascii', 'iso-8859-9', and 'iso-8859-15' encodings (Cyrillic support is not available).
In the new version:
- Improved recognition of the letter 'L' with a slanted right part.
- When using the '-o' ('—output') option, the creation of missing intermediate directories specified in the file path has been ensured.
- The variable MAKEINFO has been added to the configure file and to Makefile.in.
- Diagnostic messages related to file operations have been transformed into the format 'PROGRAM: FILE: MESSAGE'.
- In messages about the use of incorrect command-line arguments, the argument and option name are now provided.
Source: opennet.ru
