车辆识别与样本自动采集方法的研究  

Research on method of vehicle detection and sample acquisition

作　　者：王齐[1] 金小峰[1] 

机构地区：[1]延边大学工学院计算机科学与技术学科智能信息处理研究室,吉林延吉133002

出　　处：《延边大学学报（自然科学版）》2015年第2期164-169,共6页Journal of Yanbian University（Natural Science Edition）

基　　金：吉林省科技厅自然科学基金资助项目(20140101225JC)

摘　　要：为解决车辆样本采集困难的问题,在研究车辆识别与跟踪的基础上提出了样本的自动采集方法.首先,采用Blob分析技术从视频流中检测出车辆;其次,结合Blob和CamShift跟踪算法跟踪运动车辆;再次,通过分析车辆的运动轨迹判定其停驶状态;最后,控制云台变焦摄像机获取车辆的细节图像,以此作为车辆的样本.实验结果表明,本文提出的方法实时性高,对车辆的识别与跟踪、停驶判断具有较高的准确性,获取的车辆样本图像细节丰富,能够满足车辆样本库建设的基本要求.In order to solve the problem of vehicle sample collection difficulty,this paper proposed a method of automatic data collection of vehicle samples on the basis of vehicle detection and tracking.Firstly,the Blob analysis technology was employed to detect vehicles from video.Secondly,moving vehicles were tracked by the Blob method and CamShift tracking algorithm.Again,after analyzing the movement of vehicles,the vehicle states of movement and stop were detected.Finally,the detail image obtained by controlling camera pan was used as the vehicle samples.Experimental results show that the proposed method is of high real-time performance and accuracy in detecting and tracking of vehicles and the judgment of vehicles states.Meanwhile,the obtained vehicle sample images with rich details was able to meet the basic requirements of vehicle sample library construction.

关 键 词：车辆识别与跟踪 BLOB分析 CAMSHIFT 车辆样本采集 车辆样本库 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]