Feature selection in recognition of handwritten Chinese characters

Recognition of handwritten Chinese characters is a large-scale pattern recognition task, which means it is difficult and time consuming to build the corresponding classifiers. In this paper, two feature selection methods are tried to reduce the complexity and speed up the handwritten Chinese recognition. One is ReliefF-Wrapper, which evaluates the original features with ReliefF method, and then uses the wrapper method to decide the number of features to be selected. The other is GA-Wrapper that uses genetic algorithm to search optimal subset of features with high train accuracy. Experiments are performed on 800 most frequently used Chinese characters, with 80,000 handwritten samples. Results show that ReliefF-Wrapper method has good interpretation and high speed and GA-Wrapper gains higher accuracy. Limitation of both methods and future work are also discussed.