Local Language Handwritten Character Recognition

The idea of the proposed methodology is to develop an efficient system that can recognize handwritten Kannada characters, a language whose script is quite complex. It makes use of feature extraction by using CNN in conjunction with a hybrid SVM-CNN approach in classification to overcome handwriting variation. The SVM-CNN hybrid takes the strength of CNN pattern recognition and complements it with SVM classification; Also, the preprocessing would include the binarization of images and noise removal to have more qualitative data, hence significantly improving the recognition process as a whole. This holistic approach provides greater accuracy and robustness in recognition of regional languages in further digitization efforts focused on document digitization applications, automated systems, and educational tool sets for an inclusive digital practice.