Nonparametric bottom-up saliency detection by self-resemblance

Abstract

We present a novel bottom-up

On spatio-temporal saliency detection in videos using multilinear PCA

“…Finally, we compare our MPCA based method with four state-of-the art methods: a method that fuses optical flow and color features (OF) [17], the self-resemblance method (SF) [21], the phase discrepancy based saliency detection method (PD) [12], and a method that uses LBP features extended to temporal domain (LBP) [16].…”

Rare: A new bottom-up saliency model

“…We compared our model against Bruce's AIM [1] [12], SUN [5,15] (2008), Seo [13,14] (2009) and finally, Jian Li [6] in 2011.The proposed RARE method correctly detects the main blobs of interest in the images. A difference with other methods is also the nature of the final saliency maps which have a high resolution (even low-pass filtered in Figure 3) compared to other methods.…”

“…In this paper, we model the bottom-up approach on still images. The literature [1,3,4,5,6,7,10,12,13,14,15] is very active in the field. Those models have various technical approaches even if they all derive from the same idea: they look for contrasted, rare, surprising, novel, worthy to learn, less compressible, maximizing the information areas.…”

Inverse Bilateral Filter for Saliency

“…The map infers the most probable locations of the subjects of the photograph according to highly distinct salient cues. Saliency by Self-Resemblance method [43] computes socalled local regression kernels, which measure the likeness of a pixel to its surroundings. Visual saliency is then computed using self-resemblance measure resulting saliency map where each pixel indicates the statistical likelihood of saliency of a feature matrix.…”