الفهرس | Only 14 pages are availabe for public view |
Abstract Document layout analysis is important in converting document images into text. Arabic script cursive nature and different writing styles cause challenges. In this work, we introduce an approach for segmenting image into zones. Text zones are segmented into lines and then words. System accuracy achieved is 93.2% for zone classification and 98.3% for line segmentation. Also, a posteriori, word based and font-size invariant approach for font recognition using textural features based on cosine transform is proposed. Results show that the average recognition rate is 93.2% |