基于BP神经网络的规范手写体数字离线识别  

Standard handwritten form numeral off-line recognition based on BP neural network

作　　者：徐鹏[1,2] 

机构地区：[1]东北林业大学机电学院 [2]哈尔滨学院数计学院,讲师150040

出　　处：《木工机床》2008年第4期14-16,9,共4页Woodworking Machinery

摘　　要：手写体数字离线识别一直是模式识别研究的热点问题。手写体数字具有多变性和随意性较强的特点,在实际应用中对识别率要求较高,针对这一问题,设计了基于BP神经网络的手写体数字离线识别分类器,并按照规范手写体标准设计了训练样本和测试样本,通过仿真验证了此方法的有效性,识别率达到了81%。The handwritten form numeral off- line recognition always is hot topic in the pattem recognition research. The handwritten form numeral has variational and capriciousness characteristic, high distinguishes rate is be requested in the practical application. In view of this question, off- line recognition sorter is designed based on the BP neural network handwritten form numeral, and deferred to the standard handwritten form standard to design the training sample and the test sample, confirmed this method validity through the simulation, the recognition rate has achieved 81%.

关 键 词：BP神经网络 手写体数字 识别 

分 类 号：TP391.43[自动化与计算机技术—计算机应用技术]