Shape-Based Analysis for Automatic Segmentation of Arabic Handwritten Text

摘要

Text segmentation is an essential pre-processing step for many methods of recognition and for spotting systems as well. There are some characteristics in Arabic that differentiates it from Latin-based scripts. In this thesis proposal, we address the challenges of segmenting offline Arabic handwritten text. Our proposed approach of text segmentaion utilizes the knowledge of Arabic writing. Furthermore, a method for touching segmentation is proposed. To facilitate touching segmentation, a new learning-based baseline estimation method is introduced.