In this paper novel feature selection approach is used for the recognition of Devanagri handwritten numerals. The numeral images used for the experiments in the study are obtained from standard benchmarking data-set created by CVPR (ISI)Kolkata. The recognition algorithm consists of four basic steps; pre-processing, feature generation, feature subset selection and classification. Features are generated from the boundary of characters, utilizing the direction based histogram of segmented compartment of the character image. The feature selection algorithm is utilizing the concept of information theory and is based on maximum relevance minimum redundancy based objective function. The classification results are obtained for a single neural network based classifier as well as for the committee of Neural Network based classifiers. The paper reports an improvement in recognition result when decision combiner based committee is used along with class related feature selection approach.

A reliability evaluation system for the recognition of Devanagri Numerals is proposed in this paper. Reliability of classification is very important in applications of optical character recognition. As we know that the outliers and ambiguity may affect the performance of recognition system, a rejection measure must be there for the reliable recognition of the pattern. For each character image pre-processing steps like normalization, binarization, noise removal and boundary extraction is performed. After calculating the bounding box features are extracted for each partition of the numeral image. Features are calculated on three different zoning methods. Directional feature is considered which is obtained using chain code and gradient direction quantization of the orientations. The Zoning firstly, is considered made up of uniform partitions and secondly of non-uniform compartments based on the density of the pixels. For classification 1-nearest neighbor based classifier, quadratic bayes classifier and linear bayes classifier are chosen as base classifier. The base classifiers are combined using four decision combination rules namely maximum, Median, Average and Majority Voting. The framework is used to test the reliability of recognition system against ambiguity.