ZONAL FEATURE VECTORS FOR KANNADA HAND-WRITTEN CHARACTER RECOGNITION
Keywords:
Image pre- processing, Binarization, Segmentation, feature extraction, SVM classificationAbstract
An offline handwritten Kannada word recognition system using Support Vector Machine (SVM) as classifier is described in this paper. The character recognition system generally involve three major steps namely, preprocessing, feature extraction and classification. In our work in the preprocessing section some of the image processing techniques such as RGB to gray conversion, Binerization, Line segmentation and character segmentation of scanned document are implemented. In the feature extraction section zonal feature extraction method is used to extract the feature vectors, Later these features (Zonal) given as inputs to Support Vector Machine (SVM) classifier individually. There by we obtained results.
In order to evaluate the performance of our proposed Optical Character Recognition (OCR) system, 1050 samples of Kannada alphabets written by various people in various styles are made used. Part of this data set is used to train the SVM and remaining part is used to test the performance of SVM. We achieved satisfactory recognition rate of around 75%.
