Due to its effectiveness of identifying salient object while suppressing the background,boundary prior has been widely used in saliency detection recently.However,if the locations of salient regions are near the image border,the existing methods would not be suitable.In order to improve the robustness of saliency detection,we propose an improved saliency detection method using adaptive background template and spatial prior.Firstly,according to the rarity of salient object in the color space,a selection strategy is presented to establish the adaptive background template by removing the potential saliency superpixels from the image border regions,and a saliency map is obtained.A propagation mechanism based on K-means algorithm is designed for maintaining the neighborhood coherence of the above saliency map.Secondly,according to the aggregation of salient object,a new spatial prior is presented to integrate the saliency detection results by aggregating two complementary measures such as image center preference and the background template exclusion.Finally,the final salient map is obtained by fusing the above two salient maps.Quantitative experiments on four available datasets MSRA-1000,SOD,ECSSD and new constructed CBD demonstrate that our method outperforms other state-of-the-art saliency detection approaches.%目前,显著性检测已成为国内外计算机视觉领域研究的一个热点,但现有的显著性检测算法大多无法有效检测出位于图像边缘的显著性物体.针对这一问题,本文提出了基于自适应背景模板与空间先验的显著性物体检测方法,共包含三个步骤:第一,根据显著性物体在颜色空间上具有稀有性,获取基于自适应背景模板的显著图.将图像分割为超像素块,提取原图的四周边界作为原始背景区域.利用设计的自适应背景选择策略移除原始背景区域中显著的超像素块,获取自适应背景模板.通过计算每个超像素块与自适应背景模板的相异度获取基于自适应背景模板的显著图.并采用基于K-means的传播机制对获取的显著图进行一致性优化;第二,根据显著性物体在空间分布上具有聚集性,利用基于目标中心优先与背景模板抑制的空间先验方法获得空间先验显著图.第三,将获得的两种显著图进行融合得到最终的显著图.在公开数据集MSRA-1000、SOD、ECSSD和新建复杂数据集CBD上进行实验验证,结果证明本文方法能够准确有效地检测出图像中的显著性物体.
展开▼
