基于先验知识和标记分水岭的车牌定位算法

License Plate Location Method Based on Prior Knowledge and Marker-Watershed Algorithm

车牌定位是智能交通系统中的重要组成部分之一,是实现车牌自动识别系统的关键。针对复杂环境下的车辆图像,提出一种基于车牌先验知识和标记分水岭的车牌定位算法来对图像进行定位和分割。该算法首先对灰度图像进行形态学预处理,然后使用标记分水岭算法得到车牌候选区域,最后利用车牌的先验知识来定位车牌区域。实验结果表明,该方法对干扰字符和复杂背景具有很好的鲁棒性,是一种简单、快速、有效的车牌定位算法。

License plate location is an important part of Intelligent Transportation System（ITS）.It is considered to be the most crucial step in the Automatic License Plate Recognition（ALPR）.A license plate location method based on prior knowledge and Marker-Watershed algorithm is proposed to locate and split the plate region of vehicle images from complicated environment.In this method,firstly,mathematical morphology technique is used to preprocess the gray images,and then Marker-Watershed algorithm is used to segment the preprocessed images to get candidate regions,lastly,the prior knowledge is used to locate the accurate location of license plate.Experimental results show that this method is simple,quick,effective,and robust to interference characters and complicated background.