A feature extraction technique in conjunction with neural network to classify cursive segmented handwritten characters

We propose a feature extraction technique in conjunction with a neural network to classify segmented cursive handwritten characters. A heuristic and neural network based algorithm is used to segment the characters. After segmentation, the proposed technique is applied to segmented and preprocessed characters. The technique extracts global features from segmented characters and feeds them into the neural network for classification. It is able to recognise characters even if the character is rotated 90 degrees and is a little bit distorted. The proposed approach has been implemented in C++ on the SP2 supercomputer and tested on many sets of difficult cursive handwritten characters. The experimental results have demonstrated that the proposed approach performs successfully on real-world handwriting.