车牌定位基本流程包括预处理、粗定位和精定位.利用数学形态学处理图像,关键是结构元素的选取.针对使用同一结构元素确定车牌位置时所带来的缺陷,改进了将行列自适应的结构元素选取算法应用于车牌粗定位部分.针对不同光线环境对采集的车牌图像可能造成影响,提出将自适应对比度增强算法应用于预处理阶段,对车牌图像进行对比度增强处理.粗定位部分,改进了行列自适应数学形态学结构元素的方法对车牌进行定位,提高了只基于行的自适应结构元素选取算法获取候选车牌的效率;精定位依据不同条件,提出采用区域标记法和投影法结合来提取精确车牌位置.采集不同时间地点的600多张小型车辆图片进行仿真实验,实验结果表明,该算法可有效处理不同光线下的车牌,提高车牌定位精精确度.%The basic process of vehicle license plate location includes pretreatment, the rough location and the accurately location of license plate. In the step of the rough location, the traditional license plate locating algorithm based on mathematical morphology, and the key to its success is the selection of the structural element. Generally, we use the same structural elements to determine the location of license plate. But there are some faults. An algorithm based on adaptive row and columns structure element of mathematical morphology is proposed. Under natural conditions, the location of license plate in vehicle image may be effected by the brightness or darkness. Based on this, the adaptive contrast enhancement algorithm is proposed in this paper which is applied in the image pretreatment stage. The results show that the algorithm is effective. In the step of precise positioning, this paper used method of region-labeling and the method of projection to locate the license plate accurately. We collect more than six hundred images in different time or place and experiment results show that the method is effective.
展开▼
