An image saliency detection algorithm based on color and space information
Liyuan Feng, Peizhi Wen, Yuanyuan Miao, Ying Zhou · 2017
This study aims to correct inaccurate results of salient region detection. This paper presents an image saliency detection algorithm based on color and space information. First, image based SLIC (simple linear iterative clustering) method was used, then the image color and space information was fused to calculate the salient features of the region, so that the foreground and background were separated to an extent. Specifically, an optimized saliency map computing algorithm, which uses the operator of similarity measures between superpixels, was employed. This prevented background areas from being detected by mistake and provided a detection result that satisfies the human visual attention mechanism. Experiments show that the proposed algorithm produces satisfying saliency maps.