一种单摄像机视域的车辆检测与跟踪方法  

Vehicle detection and tracking method on sight of single-camera

作　　者：黄永亮[1] 蔡坚勇[1] 晏嫚[1] 

机构地区：[1]福建师范大学光电与信息工程学院,福州350007

出　　处：《计算机应用》2012年第A02期171-173,共3页journal of Computer Applications

基　　金：福建省高校产学合作科技重大项目(2011H61010034)

摘　　要：在单摄像机视域中,针对复杂环境下运动车辆的检测与跟踪问题,提出一种基于高斯混合模型和灰色预测模型GM(1,1)的检测与跟踪方法,该方法通过不断更新高斯分布参数和GM(1,1),从而减小环境对车辆检测与跟踪的影响。实验结果表明,这两个模型结合起来能够快速、准确地实现运动车辆的位置预测和跟踪,提高了视频监控系统的鲁棒性。According to vehicle detection and tracking in complex environment, this paper proposed a method of vehicle detection and tracking on the sight of single-camera based on Gaussian Mixture Model （GMM） and GM （1, I）. The GMM and GM （1, 1） were updated and the environmental impact was reduced on the vehicle detection and tracking according to the updated models. The experiment result shows that the combination of these two models can realize the forecasting position and the tracking of vehicle quickly and accurately, improving the robusmess of video surveillance system.

关 键 词：高斯混合模型 灰色预测模型GM(1 1) 位置预测 车辆跟踪 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]