手写数字特征的提取与分析.docVIP

毕 业 设 计 (论 文) 题 目 手写数字特征提取与分析 专 业 电子信息工程　 班 级 084班 姓 名 梁 杰 指导教师 周 扬 （讲师） 所在学院 信息学院　 完成时间：2012年 5月  承 诺 书 我谨此郑重承诺： 本毕业设计（论文）是本人在指导老师指导下独立撰写完成的。凡涉及他人观点和材料，均依据著作规范作了注释。如有抄袭或其它违反知识产权的情况，本人愿接受学校处分。 承诺人（签名）： 年 月 日  手写数字特征提取与分析 信息科技学院电子信息工程专业 梁杰 摘 要：关键词： Handwritten digital feature extraction and analysis 　　　　　　　　　　　　　　　　　　　　 L，Electronic and information engineering, College of Information Science and Technology Abstract: At present, the field of pattern recognition in everyday life has been more and more widely used, such as the face, fingerprint recognition, character recognition, vehicle license plate recognition. Therefore, the digital identification of learning and research is very necessary. The topic for the digital character recognition simulation demo system. Mainly using normal distribution under the minimum error rate of Bayes method and Bayes method to achieve the minimum risk, handwritten digits from 0 to9 of the identification. The system first is to realize the simulation of handwritten numeral; then using contour feature will be 5* 5 templates extracted sample characteristics, using the template in the same shape, different sizes of samples to be normalized feature extraction, so the ability of the same shape, different sizes of samples as similar; finally combined with the Bayes decision discriminant. Minimum error rate using the Bayes method, the discrimination process can make the error rate reaches a minimum, even wrong classification and the possibility of the minimum, while minimizing risks Bayes method in judging process, can make the risk minimum, harm reduction in fault classification. This design is the use of Matlab to achieve, experiments show that, the system for the simulation of handwritten digital basically correct identification, but for handwriting irr