基于颜色、边缘综合特征的图像检索技术研究  

Research of image retrieval based on color-edge features

作　　者：胡浩[1] 张曦煌[1] 

机构地区：[1]江南大学信息工程学院,江苏无锡214122

出　　处：《计算机工程与设计》2009年第1期148-151,193,共5页Computer Engineering and Design

摘　　要：在基于内容的图像检索技术(CBIR)中,颜色作为图像的一种重要视觉信息,已得到广泛应用。相对于图像的几何特征而言,颜色具有一定的稳定性,对缩放、平移、旋转具有很强的鲁棒性。目前基于颜色特征的检索技术很多,但也都有各自的局限性。提出一种新的图像检索算法,基于颜色、边缘二种特征的综合来完成图像的检索。在颜色特征中采取了聚类方法优化像素点颜色量化过程。通过加权处理,将基于颜色直方图的检索技术与基于边缘直方图的检索技术相结合,实现图像检索性能的显著提高。Within the image retrieval based on content, being an important vision information of image, are used broadly. The color is as to geometric characteristic of the picture, have certain tability, to scale, rotate, have a strong one stupid and getting wonderful quite. By now, there are lots of image retrieval technology based on color features, but also have their limitation. A new image retrieval method proposed, based on color-edge multi-features. In the field ofretrievaling based on color, we use clustering to quantify the color. By assigning a weighting value, the technology based on color features and the technology based on edge features combine to a multimethod. It has better retrieval performance.

关 键 词：图像检索 聚类 颜色直方图 边缘直方图 相似性度量 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]