计算机与数字工程2019，Vol.47Issue(10)：2573-2578,6.DOI:10.3969/j.issn.1672-9722.2019.10.038
基于优化BP神经网络的手写体识别系统∗
Handwriting Recognition System Based on Optimized BP Neural Network
摘要
Abstract
Handwritten characters are everywhere in our daily life. With the continuous development and progress of science and technology,it has become a trend to recognize handwritten characters by computer,which provides convenience for subsequent data processing. In this paper,the feature extraction of captured handwritten images and the classification and recognition of the pro?cessed images are studied. Gabor algorithm is used to process the pre-processed images by Sobel operator,and PCA is used to re?duce the dimension and reduce the computational complexity. In this paper,the basic BP neural network is introduced,and the elas?tic BP algorithm is proposed and optimized according to its shortcomings. The evaluation criterion of global error is introduced, which makes the weight updating have the advantages of adaptive learning and elasticity,and greatly improves the training speed and accuracy of the algorithm under the condition of global convergence. Aiming at the specific handwriting recognition problem, the design of handwriting recognition system is realized. The two algorithms are compared and validated,and the data are counted. The improved elastic BP algorithm has better performance,shorter training time and higher accuracy.关键词
神经网络/弹性BP算法/手写体识别Key words
neural network/resilient back-propagation/handwriting recognition分类
信息技术与安全科学引用本文复制引用
张圣杰,郭必广,陈屏国..基于优化BP神经网络的手写体识别系统∗[J].计算机与数字工程,2019,47(10):2573-2578,6.基金项目
国家自然科学基金面上项目"区块链下的跨国供应链合作风险评价与预警机制研究"(编号:71872131)资助. （编号:71872131）
