Recognition of Printed and Handwritten Kannada Characters using SVM Classifier

Abstract

The optical character recognition is the process of converting textual scanned image into a computer editable format but one of the major challenges faced is the recognition of character from the image. The proposed system is application software for Recognition of Kannada Printed and Handwritten Characters from an image. The input image is subjected for pre-processing to make the image noise free by using median filter and then it is converted to binary image. Segmentation process is carried out to extract one character from the image by performing horizontal segmentation followed by vertical segmentation. Co-relation coefficient is used for extracting the features from the image then the character is classified using SVM classifier finally the classified character is post-processed using its Unicode values to display the recognized character. We have obtained perfectness of 100% and 99% in recognition of Kannada Printed and Handwritten characters respectively.