一种新的复杂背景下快速车牌定位方法  

A Fast Vehicle License Plate Locating Algorithm Under Complex Environments

作　　者：张振强[1] 杜树新[1] 

机构地区：[1]浙江大学信息学院控制系智能系统与决策研究所,杭州310027

出　　处：《科技通报》2007年第5期705-710,共6页Bulletin of Science and Technology

摘　　要：针对实际应用中具有复杂背景的高分辨率数码图像下车辆牌照定位问题,提出了一种综合利用了车牌纹理特征、几何特征以及色彩特征的快速车牌定位方法。该方法按顺序分为图像预处理、牌照区域粗定位、牌照精确定位三个步骤对牌照由粗到细准确定位出车辆牌照位置所在。实验结果表明,对于场景大,背景复杂的车辆图像,该方法运行速度快,定位准确率高,具较强抗干扰能力。Under the condition of the complicated backgrounds of high resolution digital vehicle images and the different illumination conditions, a layered algorithm based on multi-features including texture, edge and color features of the vehicle license plate is proposed. This algorithm consists of three steps： color image preprocessing, license plate raw locating, and license plate precise locating. The color image preprocessing turns the colorized image into a gray image. After applying this algorithm to hundreds of images of vehicle taken by a colorized digital acquisition camera device on the real highway, our result indicates that this algorithm has a fast, robust and reliable performance.

关 键 词：车牌定位 纹理特征 边缘检测 RADON变换 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]