Detection of Text Lines of Handwritten Arabic Manuscripts using Markov Decision Processes

Abstract

-In a character recognition systems, the segmentation phase is critical since the accuracy of the recognition depend strongly on it. In this paper we present an approach based on Markov Decision Processes to extract text lines from binary images of Arabic handwritten documents. The proposed approach detects the connected components belonging to the same line by making use of knowledge about features and arrangement of those components. The initial results show that the system is promising for extracting Arabic handwritten lines.

Segmentation of Arabic Handwritten Documents into Text Lines using Watershed Transform

“…The F1-mesure score explained in [13] is used here too. Table I compares the score of the proposed method with previously proposed methods tested on the same samples.…”

“…9, other problems occur where text parts may be located in a place which makes it participate at the same time in more than on field of view thus allowing to link component in adjacent lines which result in merging the two lines together. Thus to resolve this, the component will be linked to the component in which it participate the most based on the percentage of participation described in [13]. Fig.…”

“…Other methods represent the problem as a graph and search the orientations of text components and uses Breadth First Search algorithm and affinity propagation clustering method to assign the components to text lines [11,12]. The work in [13] adapts the problem of text line extraction from binary Arabic handwritten documents as a Markov Decision Processes using knowledge about the features and arrangement of the components belonging to the same line.…”

Multi-agent Systems for Arabic Handwriting Recognition

“…All of them were present at the presentation. The main publications associated with the PhD thesis are [2][3][4][5].…”

IJIMAI Editor's Note - Vol. 4 Issue 1

“…Et al. [6] This article provides a comprehensive survey of recent developments in Arabic handwriting recognition.They propose a novel algorithm for extracting text lines of handwritten Arabic manuscripts using Markov Decision Processes, and they also utilize a neighborhood connected component analysis, which provide information about the location of potential characters in the processed documents. The method requires binarization.…”