一种基于分割经验搜索矩形和AdaBoost算法的车标定位方法  

A Location Method of Vehicle Logos Based on the Ada Boost and the Segmentation Experience Search Rectangle

作　　者：李若愚 文志强 Li Ruoyu;Wen Zhiqiang(School of Computer Science, Hunan University of Technology, Zhuzhou Hunan 412007, China)

机构地区：[1]湖南工业大学计算机学院,湖南株洲412007

出　　处：《信息与电脑》2018年第9期56-59,共4页Information & Computer

摘　　要：国内外目前关于车标定位的研究大体上使用的方法有"四阶段法"^([1])、基于模板匹配的车标定位方法^([2])、基于车脸有关区域纹理一致性的车标定位方法^([3])、基于形态学滤波的车标定位方法^([4-6])以及基于差分和对称性检测的车标定位方法^([7-10])等。这些方法或者因为针对性太强而导致适用面较窄,或者为了提高准确率而导致定位效率大幅度降低。笔者针对此提出一种实现起来相对简单并能迅速准确地利用道路监控所采集到的图像定位车标的方法。此法加入了车辆的倾斜校正,并利用车标相关先验知识改进了传统经验搜索矩形法,提出分割经验搜索矩形的概念,然后将其与Ada Boost分类器结合进行分类并加以适当的投影处理,最终达到准确定位车标的目的。与传统方法相比,不仅准确率高,而且适用范围广,具有一定的实用价值。At present, there are four stages of research on vehicle mark positioning at home and abroad, such as ＂phase method＂, template matching based location method, vehicle label positioning method based on regional texture consistency of car face, vehicle label positioning method based on morphological filtering, and localization method based on difference and symmetry detection.. These methods, or because of too strong pertinence, result in a narrower application area, or a significant reduction in positioning efficiency in order to improve accuracy. In this paper, we propose a relatively simple method which can be used to locate vehicle logo quickly and accurately by road monitoring. This method adds the correction of the vehicle＇s tilt, and improves the traditional experience search rectangle by using the prior knowledge of the vehicle label, and puts forward the concept of dividing the experience to search the rectangle, then classifies it with the Ada Boost classifier and properly projection processing, and finally achieves the purpose of accurate positioning of the vehicle mark. Compared with traditional methods, it has high accuracy and wide application scope, and has certain practical value.

关 键 词：车标定位 分割经验搜索矩形 ADABOOST 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]