摘要
本文给出一种基于颜色的图像检索算法,将DSQ量化与GLQ聚类算法相结合建立图像颜色直方图,避免了动态聚类算法对初始聚类类别数及相应聚类中心敏感,影响聚类结果。该算法首先采用DSQ对图像颜色进行初始量化,确定初始聚类中心;然后在此初始聚类中心基础上进行GLQ聚类;最后,采用对称相似度算法利用颜色直方图进行图像相似性匹配。试验证明,算法对图像有较好的检索性能。
In this paper,An algorithm for Color_based Image Retrieval is proposed,and requires that color histogram is made by com-bination of Dependent Scalar Quantization with Generalized Lloyd Algorithm and avoids the affection on the amount and center of initialization lustering classes by dynamic clustering algorithms. Color of the image is quantized with Dependent Scalar Quantization algorithm and clustered with Generalized Lloyd Algorithm,then,symmetric resembility algorithm is used in image matching. Experi-me...
出处
《微计算机信息》
北大核心
2008年第12期286-287,272,共3页
Control & Automation
基金
国家自然科学基金项目(90607008)
关键词
图像检索
颜色量化
聚类
颜色直方图
Image Retrieval
Color Quantization
Clustering
Color Histogram
