PCA based feature vector for handwritten Kannada characters recognition

An approach for selection of features using principal component analysis technique to classify segmented (isolated) Kannada characters is presented in this paper. Artificial neural network is used as classifier. The ability of neural networks to learn by ordinary experience, as we do, and to take sensitive decisions give them the power to solve problems found intractable or difficult for traditional computation. Handwritten characters are scan converted to binary images and normalized to a size of 50 × 50 pixels. The features are extracted using spatial co ordinates. Prominent features are then selected by principal component analysis using these spatial features, and are given to neural network for classification. With the implementation of this approach on a comprehensive database, higher degree of accuracy in results has been obtained.