Video saliency detection by spatio-temporal sampling and sparse matrix decomposition

In this paper, we present a video saliency detection method by spatio-temporal sampling and sparse matrix decomposition. In the method, we sample the input video sequence into three planes: X-T slice plane, YT slice plane, and X-Y slice plane. Then, motion saliency map is extracted from the X-T and Y-T slices, and static saliency map is extracted from the X-Y slices by low-rank matrix decomposition. Finally, these maps are retransformed into the X-Y image domain and combined with central bias prior to obtain the video saliency maps. Extensive results on ASCMN dataset demonstrate that the proposed video saliency model can achieve higher subjective and objective performances. Key-Words: saliency detection, spatio-temporal sampling, sparse matrix decomposition, motion saliency, static saliency