一种新的基于分块颜色和形状的图像检索方法  

A New Approach for Image Retrieval Based on Color and Shape of Block

作　　者：郑霞[1] 周明全[1] 耿国华[1] 

机构地区：[1]西北大学计算机科学系,西安710069

出　　处：《计算机科学》2004年第5期184-186,共3页Computer Science

基　　金：国家自然科学基金项目(60271032);中科院自动化研究所模式识别国家重点实验室项目的资助

摘　　要：基于图像单一特征的检索方法只能反映图像的部分属性,应用范围受到限制。本文在分析图像颜色和形状特征的基础上,根据各自的特点,提出了一种新的基于分块颜色和形状的图像检索方法。首先,在图像分块的基础上提取颜色和特征点;其次,根据特征点提取图像的形状信息;最后,以颜色和形状作为综合特征,采用相应的匹配算法进行图像检索。实验证明该方法不仅准确度高,而且还可应用于基于对象或区域的图像查询中。The approach for image retrieval based on single feature only can represent partial attributes of images,so its application is limited. After analyzing color feature and shape feature in images and combining their characteristics, this paper proposes a new approach for image retrieval based on color and shape of blocks. Firstly, this approach abstracts color information and character-points in the blocks of image ; Secondly, according to character-points of each block.it abstracts shape information. Finally, integrating color and shape information, it employs proper matching algorithm to implement image retrieval. The results of the experiments show that this approach not only is more accurate but also can be applied to object-based or region-based image query.

关 键 词：分块颜色 形状 图像检索方法 块 特征点 综合特征 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]