A Technique for Offline Handwritten Character Recognition

Offline handwritten character recognition has been one of the most engrossing and challenging research areas in the field of pattern recognition in the recent years. Offline handwritten character recognition is a very problematic research area because writing styles may vary from one user to another. In this paper a proposed technique for offline handwritten Gurmukhi character recognition has been presented. The success rate of our proposed scheme depends upon the feature extraction technique which has been applied in this work. We have proposed a feature extraction technique named as Neighborhood Foreground Pixels Density technique. As there could be some insignificant feature values so to reject those we have used a dimensionality reduction technique namely Principal component analysis (PCA). Maximum recognition accuracy of 91.95% has been achieved with SVM (Radial Basis function kernel) classifier by using 10 fold cross validation test method.