Hand-written digit recognition using Fourier descriptors and contour information

Abstract

We propose a method for the recognition of handwritten digits. This recognition method is based on contour information and Fourier descriptors. The proposed method is divided into three steps. First, in the preprocessing step, we extract the contours of the input digit image and separate the outer and inner contour from the contour image. Second, we extract features from the outer contour and use them to build standard models. In the last step, we recognize the digits by comparing the features of the input digits with those of the models. We use 500 data for each digit. So a total of 5000 data are used in this paper. The overall recognition rate is 99.04%.