摘要： 选票信息的自动处理能提高录入速度、保证选举的公正性。绝大多数的选票信息自动处理系统都是采用光学标记识别的方法，该方法对于选票纸张和填写要求较高，而且误识率较高。该文提出了一种基于神经网络和学习子空间光学字符识别(OCR)方法。该识别方法具有对纸张和填写的要求低，精度高、通用性好的特点。

关键词: 神经网络, 学习子空间, 选票信息处理, 字符识别

Abstract: Vote automatic processing system can assure process speed and justness. Most of the vote processing system is base on the optical mark recognition, it may make mistake in recognition. In this paper, we put forward optical character recognition techniques for vote information processing based on learning subspace method and neural network. This method has lower requirement to paper and higher accuracy in recognition and universal property in vote processing.

中图分类号: 

• TP391.4