Abstract

Automatic segmentation of the primary object in a video clip is a challenging problem as there is no prior knowledge of the primary object. Most existing techniques thus adapt an iterative approach for foreground and background appearance modeling. However, these approaches may rely on good initialization and can be easily trapped in local optimal. In addition, they are usually time consuming and difficult for analyzing videos. To solve this problem, recent work introduces a new approach for automatic primary video object segmentation. The input is a plain video clip without any annotations and the output is a pixel-wise spatio-temporal foreground vs. background segmentation of the entire sequence. However how to handling essentially unconstrained settings, becomes very difficult task by using automatic primary video object segmentation based on Markov Random Field (MRF). So in this work proposed new primary video object segmentation by following Min-Cut Max Flow in MRF (MCMF-MRF) .This work present a MCMF-MRF technique for separating foreground objects from the background in a video. MCMF-MRF method is fast, fully automatic, and makes minimal assumptions about the video. This enables handling essentially unconstrained settings, including rapidly moving background, arbitrary object motion and appearance, and non-rigid deformations and articulations. Similar too many existing image and video object segmentation approaches, we cast the segmentation to a two-class node labeling problem in a MCMF-MRF. Within the MRF graph, each node is modeled as a super pixel, and will be labeled as either foreground or background in the segmentation process. It embeds the appearance constraint as auxiliary nodes and edges in the MCMF-MRF structure, and can optimize both the segmentation and appearance model parameters simultaneously in one MCMF. The extensive experimental evaluations validate the superiority of the proposed MCMF-MRF structure over the state-of-the-art methods, in both efficiency and effectiveness.

Paper Details