Combined zemike moments, binary pixel and histogram of oriented gradients feature extraction technique for recognizing hand written Bangla characters

Feature selection is an essential step of optical character recognition. Accurate and distinguishable feature plays an important role to improve the performance of a classifier. There are several methods for feature extraction techniques for recognizing Bangla hand written characters; however the constraints of high recognition rate should be taken into consideration. The target of this paper is to provide an improvement in recognizing hand written Bangla Characters using combined Histogram of Oriented Gradients, Zernike Moments and Binary Pixel for feature extraction technique than each individual feature extraction techniques. A total of 30000 characters have been used, 24,000 for training dataset and 6,000 for testing dataset for this proposed system and this system shows 46.98% for Zernike Moments, 66.60% for Binary Pixels and 87.62% for Histogram of Oriented Gradients where overall combined features achieve an accuracy of 94.88% in recognizing characters.