基于数学形态学和Hough变换的车牌定位算法

针对复杂环境下的车牌定位率较低的问题,提出了一种基于数学形态学和Hough变换检测车牌区域的方法。首先,对车牌图像进行图像预处理,然后,利用数学形态学的高帽变换突出车牌字符区域,并对图像进行边缘检测和连通区域分析;最后,结合Hough变换和车牌的先验知识实现车牌的精确定位。实验结果表明,针对不同复杂背景下采集到的车辆图像,该算法具有很强的鲁棒性,准确率达97.3%,能够满足现代智能交通系统对车牌定位准确性和实时性的要求。

License plate location algorithm based on mathematical morphology and Hough transform

To solve the problem of low locating rate with complex environment, a precise license plate locating algorithm is proposed based on mathematical morphology and Hough transforms. Firstly, color plate image is preprocessed. Seeondly, Top-hat transformation is used to highlight the license plates regions, and edge detection and connected component analysis are implemented on the image. Finally, Hough transform and the prior knowledge of license plate are applied to locate the plate precisely. The experimental results show that the algorithm has high robustness when image is in complicated background, and the accuracy is 97.3%, which meet the requirements of intelligent transportation systems about accuracy and real-time positioning of license plate location.