Image Saliency Detection Based on Spatial Distribution Statistics of Image Patch
Chao Jia, Shuqing Lin, Fanshu Kong · 2020
According to biological visual attention mechanism, an image saliency detection algorithm based on spatial distribution statistics of image patch is proposed in this paper. Firstly, the structure of image patch is constructed by image segmentation; Then, color clustering is carried out on image patch, and the Color eigenvector and spatial distribution dispersion of clustering colors are calculated; After that, the statistical characteristics model of clustering color spatial distribution of image patch is used to detect the salient region of image; Finally, the image salient region is enhanced by calculating the distance between each pixel and the center of the image salient region. Experiments show that this detection algorithm has a higher precision and recall rate, can significantly suppress complex texture and noise, and remove the influence of complex background.