基于双目视觉的运动目标检测跟踪与定位  

Detection,Tracking and Positioning of Moving Target Based on Binocular Vision

作　　者：王婷婷[1] 李戈[1] 赵杰[1] 张学贺[1] 

机构地区：[1]哈尔滨工业大学机器人技术与系统国家重点实验室,黑龙江哈尔滨150001

出　　处：《机械与电子》2015年第6期73-76,共4页Machinery & Electronics

基　　金：国家自然科学基金(61473104)

摘　　要：为克服传统目标检测跟踪方法无法对目标准确定位,以及在复杂环境下容易受光照、阴影等因素干扰的问题,提出了基于双目视觉的目标检测跟踪与定位方法。首先使用Matlab标定工具箱和OpenCV进行摄像机标定和图像校正,然后利用基于双目视觉的背景差分法实现目标检测,结合基于Kalman滤波的改进Camshift算法实现目标跟踪,最后利用视差图得到目标三维信息,从而实现目标定位。实验结果表明,所提方法能够实时跟踪运动目标并实现目标的准确定位。In order to solve problems such as tradi- tional target detecting and tracking methods failing to locate accurately and method being susceptible to inter- ference from light and shadow, target detecting and positioning methods based on binocular vision are put forward. Firstly, the Matlab Calibration Toolbox and OpenCV were used for camera calibration as well as for image correction. Next, the background difference method based on binocular vision was used to achieve target detection, while the improved Camshift algo- rithm based on Kalman filtering was employed to real- ize target tracking. Finally, a disparity map was used to obtain the three - dimensional target information, thereby achieving target location. Experimental results show that the proposed method can track the moving target in real time and can achieve an accurate positio- ning.

关 键 词：双目视觉 目标跟踪 立体匹配 三维 测距 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]