复杂背景下的车牌自动分割方法  

An Automatic Segmentation Method of Vehicle License Plate in the Complicated Background

作　　者：方敏[1] 刘泓[1] 梁朝军[1] 

机构地区：[1]合肥工业大学电气与自动化工程学院,安徽合肥230009

出　　处：《微电子学与计算机》2004年第7期70-72,76,共4页Microelectronics & Computer

基　　金：安徽省自然科学基金资助项目(01042310)

摘　　要：车牌分割是汽车牌照自动识别的基础。本文针对含有车牌的灰度图像,利用车牌字符具有明显垂直纹理的特征,提出了一种适合字符纹理特征的边缘检测方法,获取垂直边缘图。根据车牌固有的特征,运用改进的投影算法确定车牌区域。文中详细给出了车牌分割步骤。实验结果表明,在复杂的背景下,该方法能够快速有效地实现车牌自动分割。The vehicle license plate segmentation is the basic step of the automatic recognition to the vehicle license plates. Observed that in the gray images the characters of vehicle license plates have the obvious vertical texture, a vertical edge detection method was proposed according to the character texture. In the obtained vertical edgy map, an improved projective algorithm based on the inherent feature of the vehicle license plates was used to locate the plate area. The vehicle license plate segmentation algorithm is given in details. The application examples have shown the good segmentation results in the complicated background.

关 键 词：车牌分割 边缘检测 垂直纹理 高斯叠代 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]