数字模板特征的手写数字识别系统设计  

Design and Implementation of Handwritten Number Recognition System with Digital Template Features

作　　者：李建森 李紫阳 郭文俊 LI Jiansen;LI Ziyang;GUO Wenjun(School of Information Engineering,Huangshan University,Huangshan Anhui 245041)

机构地区：[1]黄山学院信息工程学院,安徽黄山245041

出　　处：《武夷学院学报》2025年第3期43-47,共5页Journal of Wuyi University

基　　金：安徽省教育厅自然科学重点项目(2023AH051371);安徽省创新型省份建设补助资金专项资助项目(2020xzx004);国家级大学生创新创业训练计划项目(202310375055)。

摘　　要：开发一种手写数字识别系统,首先需要准备一个数据集,将其分成训练集和测试集两部分;其次利用训练集去训练数字模板特征,构建数字模板特征库,用于后续的图像处理和识别;最后使用模板匹配方法对测试集进行识别,将测试集中的图像与数字模板特征库进行匹配,从而确定手写数字的类别。通过实验对数字模板特征中的N值进行了测试,并分析了其对识别结果的影响,进而选择合适的N值以提高识别率和识别速度。实验结果表明,手写数字识别系统能准确、快速地识别手写数字,其正确识别率平均达到了98.03%。Developing a handwritten digit recognition system,first of all,prepare a dataset and divide it into two parts:the training set and the testing set;then,use the training set to train digital template features and construct a digital template feature library for subsequent image processing and recognition;finally,the template matching method is used to recognize the test set,matching the images in the test set with the digital template feature library to determine the category of handwritten digits.The N value in digital template features was tested through experiments,and its impact on recognition results was analyzed.An appropriate N value was selected to improve recognition rate and efficiency.The experimental results show that the handwritten digit recognition system can accurately and quickly recognize handwritten digits,with an average recognition rate of 98.03%.

关 键 词：手写数字识别 数据集 数字模板特征 模板匹配 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]