Research on Handwritten Character Recognition Algorithm Based on Serial Classifier in Background Domain

This paper proposed algorithm of handwritten character recognition based on serial classifier domain background. The algorithm can be used for 52 handwritten letters and 10 kinds of numbers and a total of 62 characters. It mainly includes two major modules:the first is to extract the character image background domain features; The seconde is the serial combination of MQDF and SVM two level classifiers. First of all, the MQDF classifier is rough classification at the initial stage of recognition through the concave and convex features of the background domain. Then the SVM classifier is classified by the width to height ratio of the background domain and the cross section feature to identify the fine classification later. The experimental results show that the algorithm in this paper is not only good in generality, and the recognition rate, the rejection rate and the recognition speed are superior to those of the same algorithm.