In latter years, character recognition is the most important topic in field of image processing. Character recognition still faces the problem of character matching. In this paper, we have conferred a complete Offline English character recognition system by using a suitable feature extraction algorithm that uses the structural feature primitives of the character image. To overcome this problem we have come up with a suitable feature extraction algorithm. The approach has the states like preprocessing, segmentation, feature extraction and classification and recognition. The offline character recognition system is tested on 26 each of the lowercase and uppercase English characters and 0-9 digits. We performed the experiments by taking different types of benchmark dataset images. The results so obtained are satisfactory. An overall rate of accuracy of 92.5% has been obtained. Performance comparison with other state-of-theart methods is also carried out whereby the proposed method outperformed the others in terms of rate of accuracy.

Keywords : Preprocessing; Segmentation; Feature Extraction; Classification; OCR.