Vehicle License Plate Segmentation and Recognition Method Based on Two Times Locations

Number of views： 13

分享到:

摘要：

智慧城市建设中,需对重点街道和路口采集到的交通视频文件进行智能分析.为此,提出一种二次定位车牌分割、识别方法.首先,利用垂直投影区域特征并结合Hough变换得到车牌的粗略定位分割结果;然后,基于该车牌图像的粗略定位分割结果,采用支持向量机的方法,进行车牌的精细定位分割并对车牌号码进行自动提取、识别.通过对多源车流量视频实验数据中的1680帧车牌图像进行自动车牌提取分析,在5°和10°两个倾斜角度,二次定位车牌识别方法的准确率分别达到96.7%和96.2%,优于相关算法.

During the construction of smart city, it is necessary to analyze the traffic video files collected in the main streets and crossroads. Therefore, a vehicle license plate segmentation and recognition method based on two times locations was proposed in this paper. Firstly, the coarse segmentation result of license location was obtained based on the vertical projection regional characteristic and Hough transform. Secondly, the support vector machine （SVM） was adopted to locate, segment and recognize the vehicle license plate number automatically and accurately. After analyzing the 1680 license plate images in multi- source traffic video, the accuracy of the proposed method can reach 96.7% and 96.2% in the two angles of 5 and 10 degrees, which is better than considered algorithms.

作者：

刘尚旺段德全崔艳萌周猛

机构地区：

betway官方app 计算机与信息工程学院 betway官方app 智慧商务与物联网技术河南省工程实验室

出处：

《betway官方app 学报：自然科学版》 CAS 北大核心 2016年第4期151-156,共6页

基金：

国家自然科学基金(U1304607) 河南省高等学校重点科研项目(15A520080) betway官方app 博士科研启动基金(qd12138)

关键词：

车牌识别智慧城市视频文件垂直投影区域特征支持向量机

vehicle license plate recognition） smart city video files vertical projection regional characteristic supportvector machine（SVM）

分类号：