Recognition Method for Handwritten Digits Based on Improved Chain Code Histogram Feature

Abstract

The chain code histogram feature is a simple and effective feature extraction technology. This paper proposes improvements based on Chain Code His-togram (CCH) and its first differential characteristics. According to the first Differential Chain Code Histogram (DCCH), the turning points in the direction are extracted，and the judging method of Direction Turning Point (DTP) is given. We combine CCH and DTP into a new feature, then handwritten digits of MNIST database are recognized and classified by Support Vector Machine (SVM) classifier. The experimental results proved that the recognition rate of the improved method is not only higher than CCH and first differential CCH, but also closely to the recognition rate of their combination. Obviously, the new combination reduces the feature dimension, improves the speed of training and recognition.