Segmentation of Arabic Handwritten Documents into Text Lines using Watershed Transform

Abstract

A crucial task in character recognition systems is the segmentation of the document into text lines and especially if it is handwritten. When dealing with non-Latin document such as Arabic, the challenge becomes greater since in addition to the variability of writing, the presence of diacritical points and the high number of ascender and descender characters complicates more the process of the segmentation. To remedy with this complexity and even to make this difficulty an advantage since the focus is on the Arabic language which is semi-cursive in nature, a method based on the Watershed Transform technique is proposed. Tested on «Handwritten Arabic Proximity Datasets» [21] a segmentation rate of 93% for a 95% of matching score is achieved.

Handwritten Arabic Documents Segmentation into Text Lines using Seam Carving

“…F1-Score MS2 = 90% MS2 = 95% Souhar et al [2] Not Reported 93.3% Boulid et al [3] 97.4% 94.3% Boulid et al [4] 95.8% 90.5% Kumar et al [12] 95.6% 90.9% Kumar et al [13] 98.8% Not Reported Zhang et al [14] 98.41% Not Reported The proposed method 97.5%…”

“…In the article [2], a watershed approach was used to detect text components locations then trying to find the best arranged parts to form text lines, following a local vision or as more commonly known as "bottom-up" analysis. The segmentation process begins by identifying the multiple text components present on the image, then the use of the watershed technique to split the document image into well-defined areas containing each component.…”

“…Interestingly, in [2] the authors used an hybrid approach to improve the system accuracy. Even if the analysis relies heavily on CC analysis, a projection profile analysis is used as an extension in order to help deciding which components are more likely to be from the same text line following a global vision of the document.…”

“…This can be noticed when the document is presented to the reader from a considerate distance. As a result, we can safely assume that the text line segmentation is based on identifying the interline space where each line is defined by their neighbors as previously stated in the watershed approach [2], where the authors used CC analysis in order to define the locations of high and low information density, recognizing the text areas from the empty interlines. In a similar way, the seam carving is an image processing tool based on the notion of identifying high and low information areas on an image.…”

Diacritic segmentation technique for arabic handwritten using region-based

“…There are different images. Besides, Arabic has variability style of writing [21] and as well as it contains many diacritics. The diacritic here is referring to dots and marks in every word of Arabic language.…”