PROJECT TITLE :

Structure-Sensitive Saliency Detection via Multilevel Rank Analysis in Intrinsic Feature Space - 2015

ABSTRACT:

This paper advocates a completely unique multiscale, structure-sensitive saliency detection methodology, which can distinguish multilevel, reliable saliency from various natural pictures in an exceedingly robust and versatile means. One key challenge for saliency detection is to guarantee the whole salient object being characterized differently from nonsalient background. To tackle this, our strategy is to design a structure-aware descriptor based on the intrinsic biharmonic distance metric. One profit of introducing this descriptor is its ability to simultaneously integrate local and global structure data, which is very valuable for separating the salient object from nonsalient background in a very multiscale sense. Upon devising such powerful shape descriptor, the remaining challenge is to capture the saliency to create sure that salient subparts really stand out among all doable candidates. Toward this goal, we conduct multilevel low-rank and sparse analysis within the intrinsic feature area spanned by the form descriptors outlined on over-segmented super-pixels. Since the low-rank property emphasizes much more on stronger similarities among super-pixels, we have a tendency to naturally get a scale space along the rank dimension during this means. Multiscale saliency can be obtained by merely computing variations among the low-rank elements across the rank scale. We tend to conduct in depth experiments on some public benchmarks, and build comprehensive, quantitative analysis between our technique and existing state-of-the-art techniques. All the results demonstrate the superiority of our technique in accuracy, reliability, robustness, and flexibility.