基于内容检索中的模糊处理

Fuzzy Processing in Content-Based Retrieval

基于内容的检索是多媒体信息处理中的一个重要问题，图像、视频作为多媒体中最直观、最形象的内容，对它们的检索和查询是多媒体内容处理的一个重要方面。以往对图像检索往往借助于图像理解等领域发展起来的模式识别技术，但检索在很多方面又不同于模式识别，它通常并不需要精确匹配，而且由于检索中人一机交互性起着很重要的作用，使得在检索中有必要利用模糊的方法来提出查询要求，计算机也可以通过模糊相似性匹配来给出结果。本文提出了以信任度、可能性测度、权重有隶属函数概念作为模糊相似性匹配的基础来检索图像的方法。

Content-Based Retrieval is an important issue in Multimedia in formation processing.As the most direct and vivid content,image and video imformation's retrieval is an important aspect in multimedia content processing.It has been always using pattern recognition developed in image understanding or other fields.But retrieval is different from pattern recognition in many aspect.As human-computer interactions play an important role in image retrieval,it is possible and necessary to use fyzzy method ot propose query request,and computer can also give fuzzy result.In this paper,we put forward the concept of belief measure,possible measure, membership function and weight as the base of fuzzy similarity match,with these functions we can retrieve images.