Salient object detection of fusing foreground seeds and center priori
Lingkang Gu · Journal of Physics Conference Series · 2019
In this paper, a salient object detection algorithm based on foreground seeds and center priori fusion is proposed. Firstly, using corner detection and edge linking alogrithms obtains two convex hulls, and the approximate position of the target region is preliminarily determined by their intersection points. Then using the convex hull edge as the standard, the similarity of the hyperpixel in the convex hull is detected, and removing superpixels similar to most external edges. Finally, a center priori model fused with the foreground seeds, and the final significant image is obtained by using Bayesian optimization framework. By comparing and merging different visual features, it is shown that this proposed algorithm is very effective in the saliency object detection.