基于非欧空间差异性度量的图像标注

社会媒体网站上图像数量的飞速发展,极大地激发了人们研究基于标注的图像语义检索的热情.然而,由于图像低层特征和高层语义间鸿沟的存在,导致许多基于标注的图像语义检索技术不能达到令人满意的效果.为此,文中提出一种新颖的图像标注方法——基于最大后验概率的高阶统计标注法.该方法首先使用高阶统计法计算图像的三阶差异性特征,以达到更好描述图像间相关性的目的;然后,使用基于高斯混合模型和差异性增量扩散的最大后验概率算法,来估计每个标签与图像内容的相关值;最后,通过Flick图像数据库上的实验,来验证文中所提算法的有效性.

Dissimilarity in Non-Euclidean Spaces Based Image Tagging

The overwhelming proliferation of digital images on media sharing webs has triggered the requirement of effective tools to retrieve images of interest using semantic concepts. Due to the semantic gap between low-level visual features and high-level semantic concepts of an image, however, the performances of many existing automatic image annotation algorithms are unsatisfactory. A novel image classification scheme, called the high order statistics based on maximum posterior is presented. The method firstly utilizes high order statistics to measure the triplet-dissimilarity to describe the relevancy among images, then utilizes a maximum posterior algorithm with the information of Gaussian mixture model and dissimilarity increment distribution to estimate the relevant scores of each tag. Experimental results on a Flickr image database demonstrate the effectiveness and efficiency of the proposed scheme.