基于邻域相关性和帧间连续性的前景目标分割.docVIP

基于邻域相关性和帧间连续性的前景目标分割 王传旭1,2，张祥光1,2，原春锋2，刘云2 (1. 中国海洋大学信息学院，青岛 266003 2.青岛科技大学信息学院，青岛 266061) 摘 要： 像素级前景分割方法大都基于各像素独立的假设(例如：MOGs和HMMs), 然而这种假设往往导致前景物体的不完整。本文克服了这一缺点，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。 关键词：前景目标二值图；局部相关系数；帧间相关系数；前景分割 中图法分类号：TP301.6 文献标识码：A 文章编号：26120 Integrating Local Correlation and Interframe Continuity for Robust Foreground Object Segmentation WANG Chuan-xu1,2, ZHANG Xiang-guang1,2, YUAN Chun-feng2, LIU Yun2 (1.Institute of Information, China Ocean University, Qingdao, 266003, China 2. Institute of Information, Qingdao University of Science and Technology, Qingdao, 266061, China) Abstract: Per-pixel foreground segmentation methods are mostly based on the assumption that each pixel is independent (e.g. MOGs[2][3][7][9] and HMMs[4]), which makes the segmented foreground objects not integral. In this paper, a novel technique without this deficiency is proposed. Firstly, a binary map of initial foreground segmentation is achieved by performing Bayesian strategy according to spectral, spatial, and temporal features, where the foreground map is fragmented due to independence hypothesis among pixels. Secondly, pixels’ cross correlation in neighborhood of each foreground object patch is calculated considering spatial homogeneity. The pixels’ cross correlation 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 in order to compensate small holes within foreground object. Experiments show that this method is robust in complicated background traffic scene video and can obtain more integral foreground objects. Key words: foreground binary map, local correlation, inte