一种基于区域的动态分块图象检索方法

摘要

本文提出一种新颖的基于内容的图象检索方法,基于运动子块分割并根据视觉特性对不同区域做不同的加权,比较各子块相似度,分析相似度矩阵,并检索查询物体.通过将图象分割细化,充分利用了原图的颜色位置信息,通过实验,实现了对特定物体进行检索.该物体检索方法可进一步发展,为特定的后续处理奠定基础,如在人脸识别等功能中发挥重要作用.%Content-based image retrieval (CBIR) has become an important research field in multimedia information processing. In this paper, we present a novel CBIR algorithm based on running sub-blocks with different similarity weights. By splitting the entire image into sub-blocks, we use color-layout information to retrieval images under the query of special object. Further more, after retrieval, some post-processing can be incorporated to enhance the CBIR systems for other applications.