基于内容的图像检索技术研究

在对海量的图像数据进行检索时,传统的基于数值/字符的信息检索技术并不能满足要求.因此,基于内容的图像检索技术(CBIR:Content-Based Image Retrieval)的研究应运而生,并引起了广泛关注.本文主要讨论CBIR研究中的一些关键问题:图像的内容特征及其提取、特征之间的相似度计算、查询条件的表达、检索性能的评价、压缩域的图像检索技术等等,并指出了一些可值得深入研究的方向.

Research on Content-Based Image Retrieval Techniques

With the rapid increasing of image data,an effective and efficient image retrieval system is needed for finding useful images in large databases or on Internet.Because of the complexity of images,the traditional information retrieval techniques,which are based on value and/or text comparison,are not sufficient.Then the research is shifted to the techniques of content-based image retrieval (CBIR).In this paper,some key issues on CBIR are discussed:image features extraction,similarity measures,the expression of query conditions,retrieval techniques in compressed-domain,the performance evaluation,and so on.Some future research directions are also given in the conclusion.