Extracting Visual Saliency Based on Multi-scale Receptive Field Template

A novel approach to extracting visual saliency based on receptive field is described in this paper. Visual saliency may be applied potentially in object recognition and image retrieval etc. In this paper, its two key problems are solved: saliency measurement and scale detection. Based on the receptive field hypothesis in psychology, the approach measures visual saliency in terms of the statistical features difference between target region and background region, and detects the scale of target region adaptively with a set of multi-scale measurements in the multi-resolution framework. Its definition of visual saLiency is more close to feeLing of human vision than current ones by introduction of the theories in psychology and physiology. It is capabiLity of detecting the target varying in a wide scale range with low computation burden by combing multi-scale template with multi-resolution framework. The satisfying experimental results on real images of remote sense and natural scene show the approach is effective.