Abstract: This paper presents the recognition of basic Kannada handwritten Characters based on shape features such as normalized chain code and transform domain based features as wavelet filters. In Kannada character set some characters are similar in shape; Images are made noise free by median filter and images are normalized into 64x64 dimensions. A normalized chain code and two-dimensional discrete wavelet transforms are proposed to extract as a feature vector of size 22 from the normalized binary images of size 64x64. Experiments are performed on handwritten Kannada characters consisting of 9600 images with 200 samples for each character per class. The total number of classes is 48. The SVM classifier with 2 fold cross validation is applied for classification of handwritten basic Kannada characters .These features are fed to the SVM classifier for classification of character images. To test the performance of the proposed algorithm two fold cross validation is used. The average recognition accuracy of 90.09% is obtained for Kannada vowels and consonants respectively. The experiment has given encouraging results with less number of feature set. The proposed algorithm is independent of thinning.
Keywords: Basic Kannada character Recognition; Normalized chain code, wavelet transform features; SVM classifier.