基于形态学和轮廓识别的车牌定位
License Plate Location Based on Morphology and Contour Recognition

车牌定位作为车牌识别系统的关键步骤，其定位效果与后续车牌字符分割和识别密切相关，本文对传统方法进行了改进，提出了一种新的车牌定位方法，将数学形态学和轮廓识别相结合，解决了传统定位方法形式单一、易受拍摄环境的影响且识别率不高等问题。通过形态学进行车牌的粗定位；在此基础上查找轮廓，并以宽高比、矩形度作为特征进行轮廓筛选，实现精确定位；最后通过倾斜校正、截取优化，即可输出只包含完整车牌区域的图像。实验结果表明该方法大大提高了车牌的定位准确率，达到了预期效果。
License plate location is a key step of the license plate recognition system, and its positioning effect is closely related to the subsequent segmentation and recognition of license plate characters. This article improves the traditional method and proposes a new license plate location method that combines mathematical morphology and contour recognition. It solves the problems of single form of traditional positioning method, easy to be affected by the shooting environment and low recog-nition rate. Carry out the rough positioning of the license plate through morphology; on this basis, find the contour, and use the aspect ratio and rectangularity as features to filter the contour to achieve precise positioning; finally, through tilt correction and interception optimization, the out-put contains only the complete license plate area image. Experimental results show that compared with the traditional method and this method greatly improves the positioning accuracy of the li-cense plate and achieves the expected effect.