基于流行排序的前景背景显著性检测算法

为准确提取图像显著区域,提出基于流行排序的前景背景显著性检测算法。首先,采用SLIC(simple linear iterative clustering)方法对经平滑处理的图像进行超像素分割。然后以超像素作为图中节点,采用自适应参数计算节点之间的权重以解决因采用固定值导致的图像效果不理想的问题。其次,在计算背景查询节点时,通过阈值剔除边界超像素中不属于背景的像素,以保留合适的查询节点,避免因显著目标位于图像边界而错把非背景像素标记为背景查询节点的问题。最后,因前景优先方法可以有效抑制背景噪声,而背景优先方法对背景噪声抑制不足,但可均匀突出前景目标。因此,采用相乘或者取平均的方式融合前景背景显著图以得到最终的显著图。在公开数据集MSRA、SED2及ECSSD上与其他算法进行实验对比,实验结果证明了算法的有效性。

saliency detection combined foreground with background based on manifold ranking

In order to detect salient object accurately in the image, a method based on manifold ranking which combining foreground and background is proposed. Firstly, the SLIC (Simple Linear Iterative Clustering) algorithm was adopted to divide the input images which have been smoothed into super-pixels. And instead of using fixed values, the weight of nodes was calculated using adaptive parameter to solve the problem that the image effect is not ideal because of the fixed value. Secondly, to avoid the problem that we often mistake the salient object as background query node when it is located at the image boundary. In this paper, we used threshold to eliminate boundary super-pixels which are not belonging to background. Finally, since the foreground method can suppress background noise well, while background method can highlight the foreground object evenly although it is not enough to suppress noise. Thus, we get the final map after integrate foreground and background saliency map which obtained by manifold ranking. Experimental results on benchmark databases MSRA, SED2 and ECSSD demonstrate the proposed method performs well.