In computer vision, to speed up pixel-level algorithm such as object recognition,lo-calization, matching and image saliency detection, this is can be achieved by convert pixel-level algorithm to superpixel-level algorithm.In this paper, we proposed a method to com-pute superpixel grid for image, which can guarantee superpixel shape regular and hold a reasonable adjacent relation between superpixels.In this way, the pixel-level image algo-rithms can be applied to the super-pixel level.So that image algorithm and achieve better and faster results.Image co-saliency detection is a valuable technique to high-light perceptually salient regions in image pairs. In this paper, we propose a self-contained co-saliency detection algorithm based on superpixel affinity matrix. We first compute both intra and inter similar-ities of superpixels of image pairs. Bipartite graph matching is applied to determine most reliable inter similarities. To update the similarity score between every two superpixels, we next employ a GPU-based all-pair SimRank algorithm to do propagation on the affinity ma-trix. Based on the inter superpixel affinities we derive a co-saliency measure that evaluates the foreground cohesiveness and locality compactness of superpixels within one image. |