Salient Object Detection Using Regional Contrast

Abstract

Image saliency is an important visual character which offers a priority to decide which part of the image is to be selected. Salient object is the most saliency part in visual attention and is widely used in computer vision application. We propose a novel salient object detection method based on regional contrast. We connect salient edges to a shape contour by contour detection algorithm. These salient object contours are very important cues for saliency map evaluation. Then we extract salient regions from contours. The saliency map evaluation of regions simultaneously considers the contour integrality and spatial coherence of regions. On the basis of edge manipulation, our method is compact and simple that performs efficiently. Experimental results demonstrate that our method show better accuracy of salient object boundary and performs favorably against several state-of-the-art methods.