Efficient Handwritten Digit Classification using User-defined Classification Algorithm

In automatic numeral digit recognition system, feature selection is most important factor for achieving high recognition performance. To achieve this, the present paper proposed system for isolated handwritten numeral recognition using number of contours, skeleton features, Number of watersheds, and ratio between the numbers of foreground pixels in upper half part and lower half-part of the numerical digit image. Based on these features the present paper designed user-defined classification algorithm for handwritten digit recognition. To find the effectiveness of the proposed features, these features are given as an input for standard classification algorithms like k–nearest neighbor classifier, Support Vector Machines and other classification algorithms and evaluate the results. The experimental result proves that the proposed features are well suited for handwritten digit recognition for both user and standard classification algorithms. The novelty of the proposed method is size invariant