Image Salient Object Detection Based on Perceptually Homogeneous Patch
Chao Jia, Shaoqiang Li, Weili Chen, Fanshu Kong · 2019
In this paper, a salient object detection algorithm is proposed, which use the local and global information of image, and combine with the spatial relationship between the perceptually homogeneous patch. First, color clustering is carried out on the image, and the image is divided into a set of perceptually homogeneous patches which is non-overlapping. Then, the saliency of each patch is calculated by combining the local and global information of the patch. Finally, the saliency of each patch is enhanced by combining spatial relationship. The proposed algorithm was tested using the largest publicly available data sets, Experimental results show that this algorithm can achieve higher precision and better recall rate, and significantly reduce the impact of complex texture on the calculation of salient patch.