基于边缘特征的智能车辆字符识别  

Intelligent Vehicle character recognition based on edge feature

作　　者：张帆[1] 王晓东[1] 郝贤鹏 ZHANG Fan;WANG Xiaodong;HAO Xianpeng(Chinese Academy of Sciences,Changchun Guartgji Institute,Changchun Jilin 130033,China)

机构地区：[1]中国科学院长春光机所,长春吉林130033

出　　处：《自动化与仪器仪表》2020年第6期11-14,20,共5页Automation & Instrumentation

基　　金：中国科学院战略性先导科技专项(A类)资助(No.XDA17010205)。

摘　　要：为了提高智能车辆字符识别算法实时性和准确度,提出了一种基于字符边缘梯度特征的识别算法。通过计算车牌图像的梯度信息对字符曲线进行分类,获取图像的边缘特征,然后根据K最邻近分类算法(KNN)对待检测字符分类实现字符识别。测试结果表明,在车牌识别中边缘梯度特征的算法相比于模板匹配算法对车牌识别率提高了5.23%,识别时间仅为模板匹配算法时间的21.11%。边缘梯度特征的识别算法能够高效准确地实现智能车辆车牌字符识别。In order to improve real-time performance and accuracy of intelligent vehicle character recognition algorithm,a recognition algorithm based on edge gradient feature is proposed.The character curve is classified by calculating the gradient information of the license plate image,and the edge features of the image are obtained,and then the character recognition is realized according to the K nearest neighbor classification algorithm(KNN).Experimental results indicate that the recognition rate of image edge gradient feature algorithm is 5.23%higher than template matching algorithm in license plate recognition.Meanwhile,the recognition time is 21.11%of the template matching algorithm time.The recognition algorithm based on image edge gradient feature can effectively and accurately recognize the license plate characters.

关 键 词：边缘梯度特征 K近邻分类算法 车牌识别 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]