MPEG-7区域综合特征匹配的图像检索算法

李悦; 覃团发

doi:10.11834/jig.20090927

图像理解和计算机视觉 | 浏览量 : 0 下载量: 251 CSCD: 0

PDF
导出
分享
收藏
专辑

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
2009年14卷第9期页码：1881
纸质出版：2009
DOI： 10.11834/jig.20090927
稿件说明：

移动端阅览

李悦, 覃团发. MPEG-7区域综合特征匹配的图像检索算法[J]. 中国图象图形学报, 2009,14(9):1881. DOI： 10.11834/jig.20090927.

Image Retrieval Based on Features Integrated Matching of Sub-regions for MPEG-7[J]. Journal of Image and Graphics, 2009, 14(9): 1881. DOI： 10.11834/jig.20090927.

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。

Abstract

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.