基于特征多样性的车牌定位方法

摘要

Location of vehicle license plate is difficult in vehicle license plate recognition. A new approach of vehicle license plate locating based on structure and texture features is introduced. The accurate region of the plate is obtained by using Bayesian classifier. The algorithm is not confined by the size or position of plate. The performance of the method is investigated on many car images under complex conditions. The experimental results show that the locating method has high accuracy, speed and robustness.%本文提出了一种基于结构特征和纹理特征的车牌定位方法：对车辆图像进行预处理，检测垂直方向的边缘，执行数学形态学操作，并利用车牌的结构特征粗定位车牌区域；提取车牌的纹理特征并构建特征向量，通过贝叶斯分类器来精确定位车牌区域。该方法不受车牌大小、位置等因素限制。对采集到的各种复杂背景、环境下的车辆图像进行了大量实验。实验证明：该方法定位率高，速度快，具有较好的鲁棒性。