摘要
由于计算机自动提取的图像视觉特征与人所理解的图像内容存在巨大的差异,传统的低层的视觉特征(如颜色、纹理、形状等)CBIR(Content-Based Image Retrieval)系统的检索结果往往不尽如人意。近年来,根据概念级语义(如男孩、高兴、浪漫等)的CBIR引起了研究者的重视。本文对CBIR领域的大量文献进行了深入的分析,从工程角度综述了图像概念级语义的描述模型、概念级语义特征提取和概念级语义图像检索问题的研究进展,并阐述了作者的一些观点。
As the huge difference between the visual features of image extracted automatically by computer and the image contents appreciated by people,the search results of traditional CBIR （Content-Based Image Retrieval） system with lower-level visual features （such as color,texture,shape） are not satisfactory. In recent years,CBIR based on concept semantics （such as boys,happy, romantic, etc. ） attracts the attention of researchers. In this paper, based on profound research on large numbers of references in the domain of CBIR, a survey of the description model of concept semantics, semantic feature extraction,and the study progress of the CBIR based on concept semantic is made in details from the engineering perspective,and then some of the author＇s viewpoints are pointed out.
出处
《计算机科学》
CSCD
北大核心
2008年第7期206-212,共7页
Computer Science
基金
国家自然科学基金(编号:60473039)
江苏省重点科技攻关项目(BE2004093)
关键词
图像语义
图像描述与识别
语义鸿沟
相关反馈
Image semantic,Description and recognition of image,Semantic gap,Relevance feedback
