Performance analysis of character segmentation approach for cursive script recognition on benchmark database

This paper analyzes the improved performance of our proposed character segmentation algorithm in comparison to others presented in the literature from accuracy and computational complexity points of view. The training set is taken from IAM and test set is from CEDAR benchmark databases. Segmentation is achieved by analyzing characterʼs geometric features and ligatures which are strong points for segmentation in cursive handwritten words. Following pre-processing, a new heuristic technique is employed to over-segment each word at potential segmentation points. Subsequently, a simple criterion is adopted to come out with fine segmentation points based on character shape analysis. Finally, the fine segmentation points are fed to train neural network for validating segment points to enhance segmentation accuracy. Based on detailed analysis and comparison, it is observed that proposed approach enhances the cursive script segmentation accuracy with minimum computational complexity.