An approach to analysis of arabic text documents into text lines, words, and characters

Text line extraction from a text document image and segmenting it into isolate words these individual characters are considered as one of the most critical processes in OCR systems development turning searchable electronic representation, this paper presents new approach to analyze Arabic documents, proposed contains four steps, preprocessing, segmentation, word character segmentation. The horizontal projection method used detect extract preprocessed documents image, segmentation step space threshold computed determine spaces among connected components within-word or between-words for words, finally thinning applied find skeleton segmented analyses geometric characteristics ligatures characters. was tested evaluated on set 115 images, images KFUPM Handwritten TexT (KHATT) database some produced by authors. experiment results extremely encouraging, with success rate 98.6% lines 96% 87.1%