Persian Handwritten Digits Recognition by Using Zoning and Histogram Projection

In this paper, Persian handwritten digits reorganization by using zoning features and projection histogram for extracting feature vectors with 69-dimensions is presented. In classification stage, support vector machines (SVM) with three linear kernels, polynomial kernel and Gaussian kernel have been used as classifier. We tested our algorithm on the dataset that contained 8600 samples of Persian handwritten digits for performance analysis. Using 8000 samples in learning stage and another 600 samples in testing stage. The results got with use of every three kernels of support vector machine and achieved maximum accuracy by using Gaussian kernel with gamma equal to 0.16. In pre-processing stage only image binarization is used and all the images of this dataset had been normalized at center with size 40x40. The recognition rate of this method, on the test dataset 97.83% and on all samples of dataset 100% was earned.