Optical Recognition of Handwritten Devnagari Numerals with Multifarious Recognition System

摘要

Handwritten Numerals Recognition is the process of automatically recognizing scanned images of handwritten numbers. The problem is difficult to solve because of the huge variety of ways in which people write. In fact each person's handwriting can be regarded as a different font. Ashby's law in Systems Engineering states "Only variety can absorb variety". Therefore, it is not possible to get high recognition accuracies by using only one method. In this work, a method utilizing density features and edge direction histogram with splines (EDHS) along with PCA and post processing using structural features is designed that yields a high recognition rate of 99.45%, The method has been developed in the context of Devnagari numerals recognition. However, it can in principle, be applied to any Optical Character Recognition problem with advantage.