Off-line Handwritten Verification Model for Processing Bank Checks Based on Truncated-SVD and Support Vector Machine (SVM)

Handwritten is widely used for personal identification and confirmation. It is utilized in numerous fields such as banking, e-business, access control, etc. Nonetheless, it is easily subjected to forgery. Studies have advanced in the field of handwritten verification and recognition. However, a handwritten verification system based on the Arabic language has received minimal attention. Therefore, an Arabic Off-line Verification Model (AOVM) specifically for bank check processing is suggested in this paper. The AOVM model has three major stages: preliminary processing, feature extraction using Truncated- Singular Value Decomposition (TSVD), verification using Polynomial, Linear, and Gaussian Support Vector Machine (SVM) classifiers. In the proposed model, the images of the text are normalized into uniform size at first. Consequently, the skeleton of the text is extracted for feature extraction of the handwritten using TSVD. The accuracy of the proposed model is evaluated on version 2 of the IFN/ENIT dataset of handwritten Arabic text using Polynomial, Linear, and Gaussian SVM classifiers. Moreover, the accuracy of the proposed system was compared with the result produced by a benchmark OHATRS, which is based on Principle Component Analysis (PCA) and SVM classifiers using several normalization sizes of Arabic text images. The experiential results show the effectiveness of the proposed model compared to the OHATRS model.