基于双目视觉的车辆检测和车速测定

摘要

Real-time motion detection is a key technology in intelligent video monitoring and video traffic flow detection.The widely used motion detection methods based on monocular vision are sensitive to light and have the problems of black hole and shadow.To solve these problems,we propose a motion detection method based on binocular vision.It captures the depth information of the object by the parallex of dual webcams,and detects the moving objects according to the depth difference.After obtaining the moving region,we present a novel speed detection algorithm that adapts to a variety of space shooting angles by using the depth information of the moving object in successive frames.Experiments show that our method can detect accurate contour of the moving object,which solves the problems of sensitive to light,shadow and black hole.these always exist in monocular vision situation,and achieves real-time detection and speed determination on moving object.%实时运动检测是智能视频监控和视频交通流检测中的一项关键技术.目前广泛使用的单目视觉运动检测方法对光线敏感,存在黑洞、阴影等问题.为解决此问题,提出一种基于双目视觉的运动检测方法,通过双摄像头的视差获取物体的深度信息,根据深度差检测运动物体.在得到运动区域后,利用运动物体在连续帧的深度信息,提出一种适应各种空间角度拍摄情形下的速度测定方法.实验表明,所提的方法能检测运动目标的准确轮廓,解决了单目视觉情形下始终存在的对光线敏感、阴影及黑洞的问题,实现了运动目标的实时检测和速度测定.