大视场目标的高精度快速自动检测系统  

High-accuracy and Fast-speed Automatic Detection Systemunder Wide Field Goals

作　　者：谢发忠[1] 邹华东[1] 吴年祥[1] 

机构地区：[1]安徽国防科技职业学院,安徽六安237011

出　　处：《重庆科技学院学报（自然科学版）》2014年第5期141-145,共5页Journal of Chongqing University of Science and Technology：Natural Sciences Edition

基　　金：安徽省教育厅自然科学资助项目(KJ2013B045)

摘　　要：针对传统接触式测量存在的附带损伤、精度低以及速度慢等问题,设计了一种适合大视场目标的高精度快速自动检测系统。首先,利用双目相机标定技术,精确计算出相机的内参数和外参数,完成目标和相机的空间定位。其次,对背景进行实时建模,通过重构背景的影像快速提取出目标的轮廓,完成目标的稳定检测,对光照和噪声鲁棒性好。最后,用Kalman滤波器对轮廓进行精化,从而保证了测量的高精度。实验和分析证明本系统的精度高、速度快,能够稳定的检测出大视场中存在的目标。There exist many problems in traditional contact measurement, such as collateral damage, tow accuracy and slow speed. This paper designs a high - accuracy and quick measurement for objects in the wide field. First, we use the binocular calibration technique to accurately calculate the intrinsic and extrinsic parameters. Second, we model the background in real time and extract the object by re - constructing the background. Our system can de- tect the objects robustly, which is invariant to tile illumination and noise. Finally, we refine the contours by using the Kalman filter, so that the high accuracy can be guaranteed. Experimental results verify the high accuracy and fast speed of our system, while detecting objects in the large field.

关 键 词：目标测量 大视场 背景建模 KALMAN滤波 相机标定 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]