基于神经网络的隧道行车车牌识别技术研究  

Research on Vehicle License Plate Recognition in Tunnels Based on Neural Network

作　　者：张西[1] 唐恬[1] 李良荣[1] ZHANG Xi TANG Tian LI Liang-Rong(School of Data and Information Engineering, Guizhou University, Guiyang 550025, Chin)

机构地区：[1]贵州大学大数据与信息工程学院,贵阳市550025

出　　处：《公路》2017年第7期205-210,共6页Highway

基　　金：国家自然科学基金项目"高速公路隧道节能照明关键技术研究";项目编号61361012

摘　　要：针对隧道环境下高速行车的车牌识别问题,提出使用红外摄像机采集监控视频,背景重建法进行车辆信息检测;采用Canny边缘定位算子与形态学结合的方法确定图片的车牌区域、投影法和固定边界法相结合的方法进行字符分割、引入特征提取与BP神经网络相结合进行字符的识别,提取车牌信息;并通过改进BP神经网络的学习方法来提高字符的识别速度。项目研究运用Matlab进行了大量车牌图片的样本实验,以验证此算法车牌识别的速度、准确率。In terms of license plate recognition of high-speed vehicles in the tunnel environment, in this paper the method of using an infrared camera to collect surveillance video, and detecting vehicle information is proposed. The combination of Canny edge location operator and morphology will be used to determine the coarse position of the license plate. It separates the characters by using the projection method and the fixed boundary method. Feature extraction and BP neural network are combined to identify characters and extract license plate information. Also, the paper proposes ways to improve character recognition speed with advancing BP neural network studies. A large number of vehicle license plate samples are tested through Matlab in order to verify the speed and accuracy of this algorithm. In this paper, a processing example video will illustrate its detection methods and application value.

关 键 词：隧道 高速行车 车牌识别 BP神经网络 自适应 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]