一种基于聚类的对等网络图像搜索机制  

Clustering-based image retrieval mechanism on P2P network

作　　者：张全信[1] 胡峰松[1] 欧阳竟成[1] 

机构地区：[1]湖南大学计算机与通信学院,湖南长沙410082

出　　处：《计算机应用》2006年第3期675-678,共4页journal of Computer Applications

基　　金：教育部骨干教师基金项目(教技司200065号)

摘　　要：将基于内容的图像检索应用到P2P网络中,提出了一种基于“类簇”的P2P网络信息搜索机制。结合聚类方法和Small-world原理,将相似的图像文件聚集到同一个“类簇”中,不同簇之间建立远距离连接;查询时先定位到所属的簇,在簇内广播查询,并根据查询记录更新节点路由表。仿真实验表明该搜索机制具有良好的搜索性能。The content-based image retrieval system was introduced into P2P network and a novel search model based on ＂duster clustering＂ was presentcd. Firstly, according to clustering algorithm and the small-world theory, image files with similar feature were collected together while dissimilar clusters were connected by shortcuts. Furthermore, a qucry reached its target cluster, and was broadcasted in the cluster. Founded on the previous query result, routing tables were updated. Simulation results show that the model is efficient.

关 键 词：对等网络 图像检索 聚类 路由表更新 

分 类 号：TP393.7[自动化与计算机技术—计算机应用技术]