ABSTRACT
Handwritten digit recognition plays a vital role in the automation of various aspect of our life, which starts from office automation to postal automation and many more. Handwritten digit recognition is basically a 10-class problem, but the writing styles of different individuals impose the challenge for the recognition process. In this paper, a method is devised for the recognition of handwritten Devanagari digits. For that purpose, a grid based Hausdroff distance feature descriptor is developed to represent the handwritten digits at the feature space. An appropriate classifier is then selected for the classification by comparing the performance of five well known classifiers. The proposed method is evaluated using a dataset of 6,000 images of Devanagari digits and on an average it has achieved 93.03% recognition accuracy.
