监控场景中的运动物体提取技术研究

针对监控场景中因存在遮挡而无法有效地提取出完整的运动序列这一问题，提出了一种将ViBe前景检测算法和改进后的粒子滤波跟踪算法相结合的跟踪提取方法。首先用ViBe来提取出场景中所有运动物体的前景轮廓；其次用粒子滤波来检测和跟踪目标物体；最后通过与目标物体的关联轮廓求交运算以及跟踪区域的反馈调节完成对目标物体运动帧序列的提取。当运动物体发生遮挡时，采用将跟踪区域内所检测到的前景轮廓重新加入到目标物体的关联轮廓中以保证后续可以继续用关联轮廓交集来提取。实验结果表明，该方法能够很好地保证提取的质量，并有效地解决了局部遮挡与全局遮挡情况下运动物体完整运动序列的提取。

Research on moving objects extraction in surveillance scene

Complete motion sequences sometimes can not be extracted accurately and effectively due to the presence of ambient occlusion in the surveillance scene. In view of that, a novel strategy combining Vibe algorithm with improved particle filter tracking is proposed. Firstly, extract all of the foreground contours of moving objects with ViBe, and then use the particle filter to detect and track the target object. Finally, the target object’s sequences can be extracted by intersection between contours and the target object, and by regulation of the tracking area. When the target object is occluded, the associated foreground contour detected in the tracking area can rejoin related contours to ensure the subsequent extraction. Experiments show that the algorithm can completely extract moving object’s motion sequences under both partial occlusion and global shelter conditions with good performances.