利用相关反馈技术的图像检索研究  

Research of Image Retrieval System by Relevance Feedback

作　　者：刘恒[1] 刘琳[2] 马涛[1] LIU Heng,LIU m-,MA Tao （1.Physics ＆ Information Engineering College,Henan Normal University,Xinxiang 453007,China; 2.Deparment of College Educational Administration,Henan Normal University,Xinxiang 453007,China）

机构地区：[1]河南师范大学物理与信息工程学院,河南新乡453007 [2]河南师范大学教务处,河南新乡453017

出　　处：《电脑知识与技术》2007年第11期842-844,852,共4页Computer Knowledge and Technology

摘　　要：介绍了一种支持语义的图像检索系统—PIcsearch（PICTURE Search），该系统获取图像低层特征（颜色）时采用基于区域的主颜色提取算法．综合考虑了图像的像素统计特征和空间位置信息同时节省存储空间和计算时间。提出了高级视觉特征的语义查询。在图像库上构建一个可扩展的语义网络，利用一种基于用户相关反馈的机器学习策略来改进这种语义网络，以解决低层特征向高层语义特征的过渡问题，使检索能够体现高层次语义属性。实验证明，PICsearch能有效通过人机协同工作，弥补了计算机理解能力的不足，提高了检索效率。To bridge the semantic gap,a semantic-based image retrieval system--PICsearch（PICTURE Search） is proposed in which,prior to retfieval,region-based representative color feature are extracted from the image. The project synthetically take the pixelsstatistic feature and the spatial information into account,and at the same time,save the storage space and the calculate time. The semantic query of high-grad vision feature is proposed in this paper.Based on the image database,we construct a Word net which can be extended, and we improve the Word net by machine study strategy using the method of user-based relevance feedback,in this way we can ＇resolve the transition problem from the low-level feature to the high-level semantics. Experimental resuits show that by cooperation between human and computer,we can make up the computer＇s limited ability of understanding and enhancing the effect of retrieval.

关 键 词：基于内容的图像检索 主颜色 相关反馈 语义网络 语义图像检索 

分 类 号：TP391[自动化与计算机技术—计算机应用技术;自动化与计算机技术—计算机科学与技术]