复杂背景下快速车牌定位方法研究  

Plate Localization Method under Complex Background

作　　者：鹿晓亮[1] 陈继荣[1] 

机构地区：[1]中国科学技术大学电子工程与信息科学系,安徽合肥230027

出　　处：《计算机仿真》2006年第7期256-259,共4页Computer Simulation

摘　　要：在YUV颜色空间内,根据车牌图像的纹理特征,充分利用车牌图像的三个分量彩色信息,提出了一种基于彩色边缘的车牌定位方法。该方法利用车牌区域竖直边缘异常丰富的特点,采用一种针对性很强的边缘提取算子,对Y,U和V三个颜色分量分别进行竖直边缘提取;三个分量图像竖直边缘提取之后,依据一定的准则,融合三个竖直边缘图像,得到原彩色图像的边缘。试验结果表明,与一般的单纯灰度图像边缘提取相比,该方法能更准确地定位车牌;并且经边缘提取之后,非车牌区域的边缘所剩很少,使车牌区域搜索部分变得异常简单,从而极大地减少了系统的运算负担和时间开销,满足系统实时性的要求。Making use of the texture characteristics and the color information of plate images, this paper gives out a new method, which is based on the color - edge in the space of YUV, to extract license plate. Firstly, we carry out vertical edge detection on the three components of Y, U and V because the vertical edges in the license plate area are very abundant. At the same time, a new set of arithmetic operators which mainly aim at the vertical edges are applied. Secondly, we combine the three edge maps together under a certain guide line, and thus the edge map of the color image is got. Experimental results show that this method can extract license plate more exactly compared with traditional edge detection which only applies the component of Y. Furthermore, only little of the non - plate edge is left after the color - edge detection, so the following searching for the license plate becomes much easier and the computation load is greatly reduced.

关 键 词：车牌识别 颜色空间 竖直边缘提取 车牌搜索 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]