Handwritten Arabic Characters Recognition Using Methods Based on Racah, Gegenbauer, Hahn, Tchebychev and Orthogonal Fourier-Mellin Moments

This paper presents for isolated handwritten Arabic characters recognition a comparison between the performances in terms of precision and speediness of six hybrid methods of features extraction based on a combination between profile technique and some efficient moments. More precisely these methods are profile with Racah moment, profile with Gegenbauer moment, profile with Hahn moment, profile with Tchybechev moment and profile with orthogonal Fourier-Mellin moment, finally profile combined in the same time with all these moments. For this purpose we have used for pre-processing the character images the median filter, the thresholding, the centering and the edge detection techniques, while in order to recognize each unknown character we have employed the support vectors machine. The simulation result demonstrates that the most precise method is the profile combined with all moments but in the same time it is the less fast.