Abstract
This paper presents an efficient segmentation approach for non-rigid video object. Video object segmentation is a challenging problem in video processing and plays an important role in various applications such as video object coding and video retrieval. The change of video object may be very complex and is difficult to apply many assumptions such as motion smoothness, parametric shape and high gradient contour. We propose to formulate the video object segmentation problem as the maximum a posteriori probability (MAP) problem and define the probabilistic models in terms of distance between the object’s intensity distribution and that of its spatial- and temporal-neighborhood. Furthermore, in order to accurately estimate the likelihood and prior terms in the MAP problem, we employ a non-parametric method to estimate the probabilities. Our proposed non-parametric estimation mostly relies on the object’s intensity features and requires no time-consuming motion estimation. In addition, we further employ a contour evolution method in the MAP optimization step to iteratively refine the object’s contour. Our experiments demonstrate that the segmentation results are very promising even when the video objects are severely deformed or occluded.