Text-Line Segmentation Techniques for Arabic-Handwritten Documents: A Review

Abstract

In this paper, a review of text-line segmentation is conducted, more specifically the focus was on Arabic-handwritten documents. Arabic-handwritten has a lot of difficulties when it comes to segmentation since the handwriting styles vary from one person to another, not to mention the skewness or inclined text-lines that might appear in the document that makes it difficult to implement text-line segmentation on it, along with diacritics marks that can also pose intricacies for the segmentation process. Therefore, in this research a general review about the related work that has been done in this field is first conducted, and the miscellaneous text-line segmentation algorithms used in recent years are reviewed. Moreover, comprehensive techniques for Arabic-handwritten text-lines segmentation are presented, elaborating the results of each method used along with datasets type. Finally, this review paper underscores the continuous need for innovative solutions that can handle the complexities of text-line segmentation in Arabic-handwritten documents, emphasizing the importance of pattern recognition and computer vision techniques. This will pave the way for researchers to build upon this work and invent robust new techniques to solve this problem in the future.