Saliency map fusion based on rank-one constraint
Saliency map fusion based on rank-one constraint
复制标题
DOI：
10.1109/icme.2013.6607523
复制
发表时间：
2013-07
期刊：
影响因子：
--
通讯作者：
Xiaochun Cao;Zhiqiang Tao;Bao Zhang;H. Fu;Xuewei Li
中科院分区：
文献类型：
--
作者：
Xiaochun Cao;Zhiqiang Tao;Bao Zhang;H. Fu;Xuewei Li
共显著是存在于多幅图像中的共同显著，在显著图中保持一致。一种显著检测方法为所有输入图像生成显著图，从而我们有一组图。每幅图像的显著区域由其在组中对应的显著图来提取。我们使用一个矩阵来组合所有显著区域。理想情况下，这些共同显著的区域是相似和一致的，因此矩阵排名看起来很低。在本文中，我们将这一一般的一致性准则形式化为秩酮约束，并提出了一个一致性能量来衡量矩阵的秩与一的逼近程度。我们将单个和多个图像显著图结合在一起，并在秩1约束下对这些图进行自适应加权，生成共显著图。我们的方法适用于两幅以上的输入图像，并且比现有的共显著方法具有更强的稳健性。在Benchmark数据库上的实验结果表明，该方法在共显著检测方面具有令人满意的性能。
Co-saliency is the common saliency existing in multiple images, which keeps consistent in saliency maps. One saliency detection method generates saliency maps for all the input images, so that we have a group of maps. Salient region of each image is extracted by its corresponding saliency map in the group. We use a matrix to combine all the salient regions. Ideally, these co-salient regions are similar and consistent, and therefore the matrix rank appears low. In this paper, we formalize this general consistency criterion as rankone constraint and propose a consistency energy to measure the approximation degree between matrix rank and one. We combine the single and multiple image saliency maps, and adaptively weight these maps under the rank-one constraint to generate the co-saliency map. Our method is valid for more than two input images and has more robustness than the existing co-saliency methods. Experimental results on benchmark database demonstrate that our method has the satisfactory performance on co-saliency detection.