交互式图像检索中的相关反馈技术研究进展  

Relevance Feedback in Content-based Image Retrieval:the State of the Art

作　　者：谭晓阳[1] 孙正兴[1] 张福炎[1] 

机构地区：[1]南京大学多媒体计算机研究所

出　　处：《南京大学学报（自然科学版）》2004年第5期639-648,共10页Journal of Nanjing University（Natural Science）

基　　金：国家自然科学基金(69903006);教育部高等学校骨干教师资助项目[教技司(2000)65号];中国博士后科学基金(中博基金[1997]11号)

摘　　要：相关反馈是近年来交互式图像检索领域研究的重要方向。首先提出了基于相关反馈的图像检索系统框架,并在此基础上从机器学习的角度分析了相关反馈学习的算法模型、样本获取、分布密度估计,及其在其特定应用背景下的困难和挑战。进而对图像相关反馈技术的研究现状进行调查总结,从聚类和分类两个方面对各种相关反馈算法在基于内容的图像检索中的应用进行了较为深入地研究和比较。最后对相关反馈技术发展趋势进行了展望,指出了该技术与图像语义抽取、用户模型建立以及软计算技术之间存在的密切关系。Motivated by the fast growth of image databases, content-based image retrieval (CBIR) has received widespread research interest recently, where a typical user query is represented by a dynamic combination of visual and semantic descriptions of the desired image or class of images. Unfortunately, users often have difficulty specifying such descriptions. To alleviate those problems, users' queries are often refined interactively through mining the relevance of the feedback images, with the parameters combining the features automatically adjusted to adapt to the users' original needs. The aim of this paper is to clarify some of the issues raised by this new technology by reviewing the current capabilities and limitations of the relevance feedback (RF) techniques from the viewpoint of machine learning. A concept CBIR framework with RF techniques embed in is proposed first to facilitate the description, and the basic components of a RF-based CBIR system are illustrated as well. By using the proposed model, the paper proceeds with reviewing various RE techniques existing in the literatures from three aspects. The data which can be used as the source of RF-mining is investigated firstly. To analysis the obtained feedbacks, some assumptions about the distribution of the data must be made. We describe two kinds of probable distribution assumptions commonly seen in literatures, i. e., Gaussian distribution and mixture models, and the advantages and shortcomings of both assumptions are compared. Thirdly, we identify six challenges one may encounter in the practice of applying RE technique in CBIR context and point out that there is not such perfect algorithm that can deal with all the mentioned difficulties well at the same time. Next, we focus on the classical RE methods, which mainly originate from the pattern recognition or machine learning area. Roughly say

关 键 词：相关反馈 基于内容的图像检索 信息检索 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]