A novel handwritten character recognition system using gradient based features and run length count

Abstract

In this paper, we propose a novel hand written character recognition system using a combination of gradient-based features and run length count (GBF–RLC). The performance of the proposed method has been tested on Malayalam script, a South Indian language. The gradient of image is the intensity at each point, giving the direction of the largest possible increase from light to dark and the rate of change in that direction. RLC is the count of contiguous group of 1’s encountered in a left to right/top to bottom scan of a character image or block of an image. Classification was carried out with a Simplified Quadratic Classifier (SQDF) and Multi Layer Perceptron (MLP). A database containing 19,800 isolated handwritten characters pertaining to 44 classes was used for the study. The feature vector is augmented by including aspect ratio, position of centroid and ratio of pixels on the vertical halves of a character image. The recognition accuracy of 99.78% was achieved with minimum computational and storage requirement.