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。 Aiming at the problem of vehicle license plate recognition in high-speed tunneling in tunnel environment, this paper proposes the use of infrared camera to collect surveillance video and background reconstruction method to detect vehicle information. The Canny edge location operator and morphology are combined to determine the license plate area, projection method and fixed boundary The method of character segmentation is introduced, and feature extraction is combined with BP neural network to recognize characters and extract license plate information. The learning speed of character recognition is improved by improving BP neural network learning method. The project uses Matlab to do a lot of sample experiments of license plate images to verify the speed and accuracy of this algorithm.