基于立体视觉的监控视频人体目标跟踪

摘要

The target depth information can′t be obtained by existing monocular surveillance video , and the human can′t be detected accurately because the size of human has an obvious change in the video .In order to solve these problems ,an algorithm was presented which included binocular vision to detect and track human in this paper .By using this algorithm ,the human body ,the upper body and the face can be detected continuously according to the size of human in the video .And the distance between the detected region and binocular cameras can be shown .Complete the task which label the detected region and show the distance between the detected region and binocular cameras in the video screen .%针对普通单目监控视频无法得到视频中目标深度信息的问题，以及人体目标在视频画面中因大小变化而不能被准确检测的问题，提出了一种综合双目视觉测距进行视频中人体目标的检测和跟踪的算法。算法可以根据人体目标在视频画面中的大小，按全部人体、人体的肩部以上区域、人脸的顺序进行检测，并显示被检测出的区域相对于摄像机的距离。完成在视频画面中标记检测区域，并显示检测区域和摄像机之间距离的任务。