中图法分类号tp301文献标识码a文章编号10089612014.docVIP

中图法分类号：TP301 文献标识码：A 文章编号：1006-8961(2014) - - 论文引用格式： 融合双层信息的显著性检测 姜霞霞1 李宗民1,2 匡振中2 刘玉杰1 1.中国石油大学(华东)计算机与通信工程学院, 青岛，266580； 2.中国石油大学(华东)地球科学与技术学院, 青岛，266580 摘 要 目的：根据已有工作在颜色及结构显著性描述方面的缺陷，本文提出了一种新的图像显著性检测方法。方法：本文方法在不同的图像区域表达上从颜色与空间结构角度计算图像的显著性，充分考虑图像的特征与像素聚类方式之间的适应性。首先，根据颜色复杂度、边缘与连通性等信息，将图像从像素空间映射到双层区域表示空间。然后，根据两个层次空间的特性，与每个图像区域的边界特性，计算图像的结构和颜色显著度。最后，由于不同图像表示中的显著性信息存在互补性，将所有这些信息进行融合得到最终的显著性图。结果：本文在公认的MSRA-1000数据集上验证所提出的方法并与目前国际上流行的方法进行对比。实验结果表明，本文方法在精确率、召回率以及绝对误差（分别为75.03%、89.39%、85.61%） 关键词 显著性检测；双层表示；颜色；空间结构 Image saliency detection based on two-layer information fusion Jiang Xia-Xia1 Li Zong-Min1,2 Kuang Zhen-Zhong2 Liu Yu-Jie1 1. College of Computer and Communication Engineering, China University of Petroleum (Huadong), Qingdao, 266580; 2. School of Geosciences, China University of Petroleum (Huadong), Qingdao, 266580 Abstract Objective: Image saliency detection is a method to eliminate the redundant information of the image and has been used in many computer vision applications, such as adaptive compression of images, content-aware image editing, and image retrieval. In this paper, a new image saliency detection method is proposed to compute the image saliency from different perspectives. In fact, there exist many methods computing the saliency and most of them use different types of feature to detect saliency in single regional representation, but few of them take the adaptability between the feature and the image representation into consideration. Method: According to the different characters of different types of regional representations, we compute the image saliency from different angles by using a wide variety of information, such as color, so on. The method consists of three basic steps. First, based on the information of the edges and the connectivity, image is mapped from the pixel space to a two-layer regional representation space. The first layer is more related to the spatial structure of the image and the second one is more superior