Sakhr Arabic and Islamic Software AramediA Group, technical translation, Multimedia, Localization of software into Arabic. Graphic Web page Design

Creation of character Blocks:
In Arabic, dots and other what we call external objects are used heavily. This may make lines overlap. In such cases linking external objects to main objects to create what is called block of characters (BC) becomes a very difficult process. Sakhr OCR uses an algorithm which contains a big number of intelligent heuristics to create BCs. This algorithm gives almost 100% accuracy.

Segmentation:
We believe that the segmentation algorithm use in Sakhr OCR is almost the most powerful algorithm available now to solve this problem. It gives over 98% success and even in cases the algorithm fails, we use what we call User-defined ligatures to allow users to create his own ligatures.

Line separation:
As I mentioned in the above point and in section 2, Arabic uses a
lot of external objects such as dots, 'hamza', 'madda', and diacritic marks. This makes the line separation much more difficult problem than in Latin languages. However, we have a strong line separation algorithm that even handles to good extent the problem of lines which may be sticked in part of it.

Framing:
There is a module in Sakhr OCR for what we call 'automatic framing'. This function is to divide the input page into columns if any, to exclude what are images or graphics from the recognition process in 'exclusion frames' and to include the paragraphs if any if several inclusion frames. Also the table and image detection are processed in this phase.

Recognition:
The recognition process in Sakhr OCR uses the contour and opological features in addition to the bitmap features using eural networks in OMNI recognition and matching in learning. The accuracy in the best quality can reach 99.8%. Also linguistic models are used to increase accuracy in bad quality.

Post-recognition:
One of the major critical processes in OCR systems in general is the post-recognition operations. For instance the correction process. Sakhr OCR relies on a very powerful linking between the input image and the output text which facilitate the correction process. Also Sakhr OCR uses a very powerful Spelling checker which is built on an accumulative research work in Sakhr for more than five years.

Output:
Sakhr OCR can export the recognized text to different word processors and in different formats. In addition to the text format, it can generate SWF format which can be read by all Sakhr Word processors, RTF to be read by most other word processors such as MS Arabic Word retaining the
original document layout (colums, tables, images, indentation, etc…). Also Sakhr OCR can pass the recognized text directly to different word processors: Al-moharer and MS Arabic Word.

Integration:
Sakhr OCR can integrate with other applications though
DDE communication protocol and OLE Automation.

Sakhr Islamic Software, Sakhr Arabic software, Learn Arabic, Arabic for beginners
Arabic language, software localization, software localisation, translation, Arabic
translation, multimedia, educational programs, Arabic Islam, Moslem, Islamic, Hebrew,
Farsi, Persian Persia, Iran, Iranian