Multi-Layers Saliency Detection Based on Spectral Density Peaks Clustering

Abstract

Saliency detection obtains the whole salient object by simulating human visual system, but the detection accuracy is seriously influenced by the small-scale and highcontrast regions in foreground or background, especially when dealing with objects with complex construe. This problem is common in natural images and forms a fundamental challenge for prior methods. To solve the above problem, we propose the multi-layers approach based on spectral density peaks clustering, in which the information from different saliency layers are fused to extract saliency object in complex scene, and we propose a criterion of salient region merging in order to light whole salient object. The experimental results show that the proposed method improves saliency detection on many images that cannot be handled well traditionally.