Handwritten Letter Recognition Based On BP Neural Network Optimized By Genetic Algorithm

With the development of the current society and the improvement of life,more and more characters are closely related to people,for example the English characters.A lot of English information needs to be arranged and counted in daily life,which is a time-consuming work.In particular,it takes more time to sort out the written English documents,thus the recognition of the English language becomes especially important.Therefore,this paper designs a set of recognition system for handwritten letters.The main research contents are as follows:Image preprocessing and feature extraction are firstly introduced in this paper.The main work of image preprocessing includes the using of the global threshold method to process binarization,image edge extraction by using Canny operator,target image segmentation by using the improved vertical projection method,image normalization and thinning.This paper adopts the structural feature method to extract the feature vectors of the target image.The structural feature method divides the target image into equant network and then calculates the proportion of melanin in each network to get all the eigenvalues.This method not only has stable accurate effect,but also guarantees the stability in the process of neural network training.Image recognition is secondly introduced in this paper.The specific template used to compare can not be found because of the irregular handwritten letters.So the BP neural network optimized by genetic algorithm is treated as a handwritten character recognition classifier.This paper uses these features of genetic algorithm to find the optimal weight and threshold of BP neural network which can be substituted into the BP neural network,so that the slow convergence speed and the difficulty of getting rid of local minimum can be sloved.Finally the BP neural network can be optimized.The handwritten letter recognition experiment is designed by MATLAB.After the training of the network,the recognition rate of the handwritten letters can reach 85.898%by using the improved algorithm.The experimental results show that the improved BP neural network based on genetic algorithm is effective and feasible for the handwritten letter recognition,which can meet the requirements of the handwriting recognition.