MPEG-7区域综合特征匹配的图像检索算法  

Image Retrieval Based on Features Integrated Matching of Sub-regions for MPEG-7

作　　者：李悦[1] 覃团发[1] 

机构地区：[1]广西大学计算机与电子信息学院,南宁530004

出　　处：《中国图象图形学报》2009年第9期1881-1885,共5页Journal of Image and Graphics

基　　金：广西高校人才小高地建设创新团队计划项目(桂教人[2007]71号);科技部科技型中小型企业技术创新基金项目(06C26224500454);广西科学基金项目(桂科自0991058)

摘　　要：为改进现有基于颜色-纹理特征的图像检索算法在反映空间分布信息方面的不足,提出了一种基于MPEG-7区域综合特征匹配的图像检索算法(FIM-RBIR):首先按类粗分辨率对图像进行区域划分,再综合图像区域的颜色、纹理特征实现有效检索。实验结果表明,FIM-RBIR算法比SIMPLIcity系统的检索效果有了比较显著的改进,平均检索性能提高了4.33%,具有较好的平均查准率和鲁棒性。In order to improve the current image retrieval methods based on color and texture which are deficient in describing the spatial relations, in this paper, a new method called features integrated matching for region based image retrieval（FIM-RBIR） is proposed. At first, the image is segmented using a coarser resolution scheme, then the color and texture features of sub-regions are combined to achieve the image retrieval. The experimental results show that the FIM- RBIR algorithm is more accurate than the SIMPLicity method, the average retrieval performance has been raised by 4.33% , better average precision ratio and robustness has been obtained.

关 键 词：基于内容的图像检索 MPEG-7 图像区域 颜色 纹理 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]