复杂道路环境下的车辆牌照检测与识别  

Vehicle license plate detection and recognition in complex road environments

作　　者：万雨昊 WAN Yuhao(Schools of Information and Communication Engineering,Shanghai University,Shanghai 200444,China)

机构地区：[1]上海大学通信与信息工程学院,上海200444

出　　处：《计算机辅助工程》2024年第2期31-37,共7页Computer Aided Engineering

摘　　要：针对复杂道路环境中车牌因倾斜、模糊、遮挡导致图像定位检测效果不佳和识别精度低等问题,提出一种基于全局阈值的灰度二值化图像预处理方法,采用YoloV5l算法对后处理阶段的数据集进行定位检测和检测结果评估,并通过R-CNN模型识别定位检测后的车牌图像字符。结果表明:当训练过程持续到100轮次时,相比于Faster R-CNN算法,该模型检测的平均精度均值(mAP)提升9.2%,识别准确率提升17.33%,验证该方法检测和识别车牌的有效性与优越性。A gray binarization image preprocessing method based on a global threshold is proposed to aim at the issues of poor positioning detection effect and low recognition accuracy caused by tilt,blurring and occlusion of license plates in complex road.YoloV5l algorithm is adopted to conduct positioning detection and evaluate detection results on data sets in the post-processing stage.R-CNN model is used to recognize the license plate image characters after location detection.The results show that when the training process continues to 100 rounds,compared with the Faster R-CNN algorithm,the mean average precision(mAP)of the model detection is improved by 9.2%,and the recognition accuracy is improved by 17.33%,which verifies the effectiveness and superiority of this method in detecting and recognizing license plates.

关 键 词：灰度二值化 图像去噪 深度学习 YoloV5l 车牌定位 R-CNN 字符识别 目标检测 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] U491.116[自动化与计算机技术—计算机科学与技术]