基于感兴趣区域的彩色图像检索  

Colorized Image Retrieval Based on ROI

作　　者：王冰[1] 

机构地区：[1]中国科学技术大学电子工程与信息科学系,安徽合肥230027

出　　处：《计算机仿真》2008年第7期216-219,共4页Computer Simulation

摘　　要：随着图像信息的大规模应用,图像检索具有越来越重要的应用价值。基于内容的图像检索成为今年来的研究重点。提出一种基于感兴趣区域(RO I)和颜色—纹理特征的联合彩色图像检索新方法。采用此方法,将图像经由对图像进行小波变换,得到小波变换域的颜色局部能量,进而得到图像分割后各区域颜色和纹理综合特征,并采用用户定义感兴趣区域即RO I(Region-of-Interest)对相似度进行修正。实验结果证明,该方法能够快速有效地得到图像特征和图像中的感兴趣物体,取得很好的检索效果。As the image data has been used more and more widly,image retrieval becomes much more important.Image Retrieval which based on aspect of content has been a significant researching point in the researching field of colorized image rerieval nowadays.This paper designs a novel colorized image retrieval method which combine the region-of-interest（ROI） and color-texture feature.As to this novel method,users are able to get the the value of local energy from the colorized images after the process of wavelet transform,and then the color-texture feature of the regions after segmentation are obtained from the local energy in discrete wavelet transform domain,meanwhile the user defined ROI is adopted to revise the image similarity.Experiment results prove that our methods can obtain the image feature and the object of interest efficiently,and can achieve excellent retrieval performance.

关 键 词：图像检索 感兴趣区域 小波变换 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]