基于反馈背景模型的城市道路交叉口前景目标检测

发布时间：2018-11-04 14:50

【摘要】：为准确检测城市道路交叉口监控视频中缓慢行驶或短时停留的前景目标,提出了一种基于前景目标反馈的背景模型检测方法.首先基于观测样本的像素值构建背景模型,利用计数器观测像素点检测为前景或背景的次数并描述当前场景的交通状态和稳定性,其次根据场景自适应阈值判断当前像素点为前景点或背景点,最终通过交通状态和场景的稳定性更新背景模型.采用基于真实的交叉口视频场景对算法的有效性进行了定性与定量分析.实验结果表明,该算法在复杂的城市道路交叉口场景中检测出缓慢行驶或短时停留车辆的性能优于其他方法,同时能够满足城市道路交叉口智能视频监控实时性和准确性的要求,为交叉口前景目标的行为分析奠定了基础.
[Abstract]:A background model detection method based on foreground target feedback is proposed in order to accurately detect the foreground target in the surveillance video of urban road intersection which is slow moving or staying for a short time. Firstly, the background model is constructed based on the pixel value of the observation sample, and the number of times the pixels are detected as foreground or background by the counter is used to describe the traffic state and stability of the current scene. Secondly, according to the adaptive threshold of scene, the current pixel is determined as foreground point or background point. Finally, the background model is updated by traffic state and scene stability. The validity of the algorithm is analyzed qualitatively and quantitatively based on real intersection video scene. Experimental results show that the proposed algorithm is superior to other methods in detecting slow-moving or short-stay vehicles in complex urban road intersections. At the same time, it can meet the requirements of real-time and accuracy of intelligent video surveillance of urban road intersection, which lays a foundation for the behavior analysis of intersection foreground target.
【作者单位】：西安文理学院西安市物联网应用工程实验室;东南大学交通学院;湖北经济学院物流工程学院;
【基金】：国家自然科学基金中国西安科技创新基金~~
【分类号】：TP391.41;U495

【相似文献】