A Novel Feature Combination Methods for Saliency-Based Visual Attention

Abstract

In the field of visual attention, bottom-up or saliency-based visual attention allows primates to detect non-specific conspicuous objects or targets in cluttered scenes. Simple multi-scale “feature maps” detect local spatial discontinuities in intensity, color, orientation, and are combined into a “saliency” map. In this paper, we propose a Saliency Map based on feature weighted, in which the Rough Sets is used to assign the weighting for every feature. This method measures the contribution of each conspicuity map obtained from the feature maps to saliency map. And it also carries out a dynamic weighting of individual conspicuity maps. We obtain results, which enrich the theory of Saliency detection. We use the real data of natural scenes to demonstrate the effectiveness of the algorithm.