极端复杂背景下的牌照快速定位新方法  

New method for fast vehicle plate location in very complicated background image

作　　者：管庶安[1] 邓红耀[1] 

机构地区：[1]武汉工业学院计算机与信息工程系,湖北武汉430023

出　　处：《计算机工程与设计》2006年第7期1170-1172,共3页Computer Engineering and Design

基　　金：湖北省自然科学基金项目(2004ABA033);湖北省教育厅重大项目(2002D01001)

摘　　要：研究了从背景极为复杂、拍摄环境较差的图像中定位汽车牌照的方法。从限制色调变化范围着手,导出了一种新的RGB分割阈值计算式。结合牌照的两钟颜色搭配特征,运用该计算式实现牌照定位。对416幅随机拍摄图像进行实验,结果牌照无一遗漏,仅有一个不含车牌的侯选区。此方法能快速排除极端复杂背景的干扰,对拍摄光照的强弱适应性强。Study on a new method of fast car license plate location in image with very complicated background and bad shooting condition. A new RGB threshold formula by limiting the hue change range is educed. Combining with the arrange characteristics of two kinds of colors, the formula is used to perform location. 416 images are tested, none of the results manifested missed. Only 1 candidate-absented plate is got. The method could exclude very complex background fast and has a wide adaptability to illumination condition.

关 键 词：牌照 随机拍摄 RGB阈值 颜色样本集 快速定位 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]