Overview

ALOCR-C Ver1.0 is a Japanese character recognition library released in May 2000.
Ver.1.0 provides text block extraction functionality for both vertical and horizontal writing, as well as Japanese character OCR functionality.
The API is based on C language invocation and is used by dynamically linking it from C++, C, VB, etc. as a dynamic link library (DLL) for 32-bit Windows.

Ver.2.0

ALOCR-C Ver2.0, released in December 2005, enhances the performance and functionality of ALOCR-C Ver.1.0. We also started offering it through in-house server licenses.
  1. Significant improvement in recognition speed and accuracy.
  2. Support for PDF output of recognition results.
  3. With batch processing commands, you can convert image files to image + transparent text PDF by simply calling one API.
    Regarding batch processing commands, customization options such as image support for book scanners, XML output, and Unicode output have been provided based on user needs.

Ver.3.0

ALOCR-C Ver3.0, released in December 2008, further enhances the functionality of ALOCR-C Ver.2.0. We have expanded the server license to include external service licensing.
  1. In addition to the conventional TIF/BMP formats, support for PDF/JPEG/GIF/PNG/JPEG2000 formats as input image formats.
  2. Support for specifying text through manual operation
    1. Rectangle Hook (inside rectangle + image overlapping with the rectangle)
    2. Rectangle Inner (only inside rectangle, images overlapping with the rectangle are not included in the recognition target)
    3. Line Hook (line segment + images overlapping with the line segment + images inside the rectangle that encloses the entire line)
  3. Added API with editing functionality for recognition results
    1. Delete and insert recognition results
    2. Replace candidate characters
    3. Register recognition character images with any character code in the user dictionary
    4. Undo/Redo support for a specified number of times

Recognized character types

The recognized character types include all kanji characters from JIS level 1, some symbols, alphabets, numbers, katakana, hiragana, and 58 personal name kanji characters from JIS level 2. However, by adding or replacing pattern dictionaries, the recognized character types can be extended freely.

Target fonts for recognition

The target fonts for recognition include Mincho, Gothic, Maru Gothic, and Kyokasho, with support for multiple styles and weights. A low-quality image dictionary and a pattern dictionary for Kaisho are provided as standard.
For Kaisho, please include it for recognition in special documents such as business cards, and normally exclude it.

Target character sizes for recognition

The target character size for recognition is 6 points or larger at 400 dpi. By incorporating a low-quality image dictionary, it is possible to improve the recognition rate for small characters of 4 to 6 points.
リファレンスマニュアル

パラメータ調整ツール操作マニュアル