We have developed a font-based intelligent character segmentation and recognition system. Using font-characteristics of structurally similar looking font-files, this system aims at building a font-model to aid in the recognition of new scripts irrespective of glyph composition. Three feature extraction schemes have been used to demonstrate the importance of appropriate features for classification. The schemes have been tested on Latin as non-syllabic and Khmer as syllabic scripts and results are reported. Experiment results show the recognition accuracy can reach 92 % for Khmer and 96 % for Latin degraded image documents, both at character level. This work is a step towards recognition of scripts of low-density languages which typically do not warrant the

For a collection of planar shapes, (e.g. characters in the given font) a fractal operator is defined which when applied iteratively to a standard set of initial contours, it reconstructs original objects with a high precision. This construction enables fast scalings and rotations of planar individual objects by simple algebraic operations on the fractal representation. Significant compression of data volume was obtained for shape representation. Key words: Fractals, Image Processing, Shape Modelling. 3 On leave from Institute of Computer Science, Polish Academy of Sciences. y The work was supported by the grant no. 8T11C01808 of Polish State Scientific Committee (KBN). 1 1 Introduction Several attempts were made to apply fractal approach to shape modelling. For instance Barnsley and Jacquin [2] defined RIFS (recurrent iterative function system) in which a collection of contours can be reproduced. However, the design process exhibits very high algorithmic complexity. Ip, Wong, an...

A careful comparison of three numeric techniques for estimation of the curvature along spatially quantized contours is reported. Two of the considered techniques are based on the Fourier transform (operating over 1D and 2D signals) and Gaussian regularization required to attenuate the spatial quantization noise. While the 1D approach has been reported before and used in a series of applications, the 2D Fourier transform-based method is reported in this article for the first time. The third approach, based on splines, represents a more traditional alternative. Three classes of parametric curves are investigated: analytical, B-splines, and synthesized in the Fourier domain. Four quantization schemes are considered: grid intersect quantization, square box quantization, a table scanner, and a video camera. The performances of the methods are evaluated in terms of their execution speed, curvature error, and sensitivity to the involved parameters. The third approach resulted the fastest, but implied larger errors; the Fourier methods allowed higher accuracy and were robust to parameter configurations. The 2D Fourier method provides the curvature values along the whole image, but exhibits interference in some situations. Such results are important not only for characterizing the relative performance of the considered methods, but also for providing practical guidelines for those interested in applying those techniques to real problems.

Abstract: We present in this paper an overview of realized works in the field of automatic segmentation of off-line Arabic handwriting. The Arabic writing is cursive in nature even printed or handwritten. The shapes of characters vary considerably according to their positions within the word. The word shapes change depending on whether letters are horizontally or vertically ligatured, i.e. superposed letters. This variability makes word decomposition in letters very delicate and not always assured, what explains the lack of robust commercialized systems. The objective of this paper is to realize a state of the art of the different techniques for off-line Arabic handwriting segmentation proposed in the literature.

Abstract:- This paper presents the details of a comprehensive database of Printed Arabic text for Arabic text recognition research. It consists of scanned images of different forms of Arabic printed text (viz. book chapters, advertisements, magazines, newspapers, and reports) scanned with 200, 300, and 600 dpi resolutions. A total of 6954 pages are scanned. The database may be utilized by Arabic printed text recognition research community. It may be used as a benchmark database where researchers can evaluate their algorithms and results compared with published work of other researchers using the same database. To the best of our knowledge, there is no public comprehensive printed Arabic text database that is freely available. Hence, this database may address this deficiency in Arabic printed text recognition research. This database will be made freely available to interested researchers. In addition, this paper presents a software GUI environment to make the manipulation of the created database easier. Moreover, the software GUI provides a number of imageprocessing functions that can be used in the field of automatic text recognition.

In this paper, we present a multi-level recognizer for online Arabic handwriting. In Arabic script (handwritten and printed), cursive writing – is not a style – it is an inherent part of the script. In addition, the connection between letters is done with almost no ligatures, which complicates segmenting a word into individual letters. In this work, we have adopted the holistic approach and avoided segmenting words into individual letters. To reduce the search space, we apply a series of filters in a hierarchical manner. The earlier filters perform light processing on a large number of candidates, and the later filters perform heavy processing on a small number of candidates. In the first filter, global features and delayed strokes patterns are used to reduce candidate word-part models. In the second filter, local features are used to guide a dynamic time warping (DTW) classification. The resulting k top ranked candidates are sent for shape-context based classifier, which determines the recognized word-part. In this work, we have modified the classic DTW to enable different costs for the different operations and control their behavior. We have performed several experimental tests and have received encouraging results. 1

In this paper, an efficient approach for the recognition of online Arabic handwritten characters is presented. The method employed involves three phases: First, pre-processing in which the original image is transformed into a binary image.Second, training neural networks with feed-forward back propagation algorithm.Finally, the recognition of the character through the use of Neural Network techniques. The proposed approach is tested on 1400 different characters written by ten users. Each user wrote 28 Arabic characters five times in order to get different writing variations. Experiment results showed the effectiveness of our approach for recognizing handwritten Arabic characters.