Interesting moving object segmentation based on selective visual attention and Markov random field

This paper proposes an approach that combines selective visual attention with the Markov random field (MRF) framework for segmenting moving objects of interest in video sequences. The gray level and motion features are first extracted from the input image. Then moving regions are obtained by combining motion features with connected component labeling. The shape features of the moving regions are extracted and compared with the predefined shape features of interesting objects, and initial object mask is generated by a set of selected moving regions. Finally, MRF classification is exploited to further classify every pixel in the initial object mask into either an object or background. Experimental results demonstrate the good segmentation performance of the proposed approach.