多标签传播和边缘局部分析的显著性检测

Saliency detection based on multi-label propagation and local analysis of edge

针对显著性检测中显著目标周围模糊的现象,提出了一种基于多标签传播和边缘局部分析的图像显著性检测算法。首先提取最信赖的目标和背景标签,通过基于元胞更新策略的多标签传播算法,准确定位显著目标;然后提出了一种新颖的分析边缘局部线索并完成分类标记的策略,以消除目标边缘的模糊现象。公开数据集上的实验结果显示,检测结果有效地抑制了目标周围的背景,解决了目标周围的模糊问题,清晰均匀地突出了显著目标。

Aiming at fuzzy phenomena around edge of salient object in the saliency detection, a new image saliency detection algorithm based on multi-label propagation and local analysis of edge is proposed. Firstly, the most reliable object and background labels are extracted, then, the salient objects have been located accurately with the multi-label propagation algorithm based on cells updating strategies. A novel strategy is proposed for further optimization, which analyzes local clues of the extracted edge labels and accomplishes nodes classification to fall within the selected local range. Therefore, the fuzzy phenomena are eliminated around salient object edge. The experimental results on the public datasets show that the proposed algorithm can suppress the backgrounds around the salient object effectively, solves fuzzy phenomena problem around the salient object completely and highlights the salient objects clearly and uniformly.