基于邻域相关性和帧间连续性的前景目标分割

Integrating Local Correlation and Interframe Continuity for Robust Foreground Object Segmentation

提出了一种新的运动目标分割算法。首先利用像素的颜色、空间的和帧间的特性信息结合贝叶斯判别定理对视频图像进行粗分割,得到一个前景目标的二值图,由于该类方法基于像素间彼此独立的假设,导致分割出的前景目标不完整存在很多空洞。其次,基于前景目标局部邻域空间的一致性假设,计算该邻域内像素间的互相关系数;同时,基于背景的帧间连续性和前景的不连续性,计算像素帧间的互相关系数。最后,依据像素的互相关系数在该邻域内进行二次判决,以填补粗分割中前景目标内部的空洞。实验表明,在复杂背景交通视频中该分割算法具有较强的鲁棒性,并能获得更完整准确的前景目标。

A novel technique without foreground segmentation is achieved this deficiency is proposed. Firstly, a binary map of initial by performing Bayesian strategy according to spectral, spatial, and temporal features, where the foreground map is fragmented due to independence hypothesis among pixels. Secondly, the cross correlation of pixels in neighborhood of each foreground object patch is calculated by spatial homogeneity. The cross correlation of pixels between two frames regarding background interframe continuity and foreground discontinuity is also computed. Finally, pixels in each neighborhood are reclassified according to the above cross correlations to compensate small holes within foreground object. Experiments show that the method is robust in complicated background traffic scene video and obtains more integral foreground objects.