图像检索中基于长期学习的动态用户模型  

A Long-Term Learning-Based Dynamic User Model in Image Retrieval

作　　者：张亮[1] 周向东[1] 张琪 施伯乐[1] 

机构地区：[1]复旦大学计算机与信息技术系,上海200433 [2]Department of Computer Science,University of North Carolinaat Chapel Hill USA

出　　处：《软件学报》2005年第2期233-238,共6页Journal of Software

基　　金：国家自然科学基金;国家高技术研究发展计划(863);上海市自然科学基金~~

摘　　要：提出一个结合长期学习的概率检索模型,即通过估算一个动态的用户模型来预测目标图像的概率分布.系统以反馈序列为用户模式,通过与反馈日志中用户模式的匹配,动态地调整对当前用户的预测.与 PicHunter 传统用户模型相比,此系统基于长期学习的方法,通过对用户反馈的观察,动态地调整当前的用户模型,从而更精确地推断当前用户的反馈行为.对 1.1 万幅图像数据库的实验表明,与已知方法相比,此方法对检索精度有明显的改善.This paper proposes a probabilistic model incorporating long-term learning to estimate a dynamic user model. By using RF sequence as the user pattern, the approach can gradually update the prediction of current user based on matching the current user pattern with the user patterns in log. Compared with the invariant user model in PicHunter, the model is capable of dynamically adjusting when more user actions are observed, thus provides more accurate prediction for probability distribution. Experimental results on 11 000 images show that this approach can improve the retrieval accuracy apparently.

关 键 词：基于内容的图像检索 用户模型 长期学习 用户相关反馈 Bayesian公式 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]