Feature Optimization to Recognize Telugu Handwritten Characters by Implementing DE and PSO Techniques

Recognizing Indian handwritten text is relatively complex compared to recognized foreign language such as English. In this work optimization techniques are presented to recognize Telugu handwritten characters. By extracting cell-based directional features from these characters, optimum features are selected by implementing optimization algorithms such as differential evolution and particle swarm optimization. An improvement of 3.5 % recognition accuracy is achieved using differential evolution algorithm. The optimization techniques are compared with the existing hybrid approach of Telugu script.