基于ARM平台的手写体数字快速识别方法  

A Fast Recognition Method of Handwritten Digits Based on ARM Platform

作　　者：李王辉[1] 肖文珂 LI Wanghui;XIAO Wenke(Hebi Polytechnic,Hebi Henan 458030,China)

机构地区：[1]鹤壁职业技术学院,河南鹤壁458030

出　　处：《信息与电脑》2022年第7期201-203,共3页Information & Computer

基　　金：鹤壁职业技术学院科技类重点课题“基于FPGA实现手写体识别的教学实验设备开发”(项目编号:2021-KJZD-004)。

摘　　要：手写体数字对我国各项领域的发展起着重要的促进作用。然而,手写体数字包含的特征信息较少,字符具有一定的差异性,导致传统识别方法的识别精度较低。基于此,文章提出了将ARM平台与手写体数字快速识别相结合的方法设计。首先,采用分类器,提取手写体数字特征,生成对应的特征矢量;其次,通过自定义分类器的内置参数与拓扑结构,计算手写体数字识别梯度;最后,获取手写体数字图像的灰度值,对图像进行二值化处理,基于ARM平台对二值化图像进行切割,实现快速识别的目标。实验证明,此种快速识别方法的识别率均保持在95%以上,识别结果的准确性与精度均存在明显优势。Handwritten numerals play an important role in promoting the development of various fields in our country. However,because handwritten numerals contain less feature information and characters have certain differences, the recognition accuracy of traditional recognition methods is low. Based on this, a method design combining the ARM platform with the fast recognition of handwritten digits is proposed. First, a classifier is used to extract the features of handwritten digits and generate corresponding feature vectors;the gradient of handwritten digits recognition is calculated by customizing the built-in parameters and topology of the classifier;finally, the gray value of the handwritten digits image is obtained, and the image is subjected to two Value processing, based on the ARM platform, the binarized image is cut to achieve the goal of rapid identification. Experiments show that the recognition rate of this fast recognition method is kept above 95%, and the accuracy and precision of the recognition results have obvious advantages.

关 键 词：ARM平台 数字识别 识别率 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]