Abstract

The offline handwritten character recognition is the frontier area of research from last few decades in pattern\nrecognition. It is difficult to recognize handwritten characters as compared to printed characters because of the varying\nwriting styles of individuals. The massive work has been done in languages like Devnagri and Chinese character recognition.\nThe area of Gurmukhi character recognition is even though not new but the problem lies when it comes to look alike and\nunique characters where the system lacks. In this proposed work, 35 different character samples are used for recognition. The\nsamples have been taken on a plain paper in an isolated manner. After the pre-processing of particular character, feature\nextraction technique is applied. The technique used for feature extraction is Gabor filter. Then Artificial Neural Network is\napplied for character recognition and if Artificial Neural Network fails to recognize, then character is recognized with the\nhelp of Particle Swarm Optimized Neural Network. This improves the overall efficiency of the character recognition system.\nBy training the classifier with whole dataset we obtained 100% accuracy for the given samples.