Research On Irregular Graph Based Visual Salient Object Detection

Humans have the capability to rapidly localize the most interesting region of an image scene by using an inherent selective visual attention mechanism.It is also important that the computational visual system can simulate the human visual selective-attention mechanism,so as to handle the information overload problem.In recent years,saliency detection represents a vital pre-processing stage of computer vision.Because the human visual selective-attention mechanism is inherently complex,visual attention modeling remains challenging,despite its achievements in recent years.In recent years,many graph-based saliency propagation methods have been proposed and achieved outstanding results in the field of salient object detection.Most existing propagation-based salient object detection methods construct a 6)-regular graph for saliency propagation.Applying a regular graph to a vast smooth region is potentially prone to unnecessary or prolonged propagation errors,leading to the excessive highlighting of the background regions.To mitigate such problems,this paper substitutes the conventional 6)-regular graph with an adaptive irregular graph for saliency value propagation,thereby avoiding unnecessary iterations over a vast smooth region.The proposed method first performs a clustering analysis based on the smoothness,color,and other features of regions.The new graph boosts an adaptive link density by considering the clustering result.In addition,this paper proposes a seeding strategy for the propagation.Based on the experimental studies of six major benchmark datasets,the proposed method performed favorably against the other state-of-the-art methods,both quantitatively and qualitatively.The proposed method furthermore efficiently suppresses the background regions and uniformly highlights the salient regions.