A Two-Pass Approach to Pattern Classification

Abstract

A two-pass approach to pattern recognition has been described here. In this approach, an input pattern is classified by refining possible classification decisions obtained through coarse classification of the same. Coarse classification here is performed to produce a group of possible candidate classes by considering the entire input pattern, whereas the finer classification is performed to select the most appropriate one from the group by considering features only from certain group specific regions of the same. This makes search for the true pattern class in the decision space more focused or guided towards the goal by restricting the finer classification decision within a smaller group of possible candidate classes in the second pass. The technique has been successfully applied for optical character recognition (OCR) of handwritten Bengali digits. It has improved the classification rate to 93.5% in the second pass from 90.5% obtained in the first pass.