摘要
研究了基于图像理解和意图识别的多Agent自主协同方法,研究对象是空中目标,估计跟踪关系,帮助观察者有效地做出决策.列出了两种跟踪模型:一是通过估计敌方飞机的飞行路径来调整自己的前进方向;二是一直向敌方方向跟踪.基于两种方向,提出了一种复合函数来估计跟踪趋势的概率,同时假设Agent之间正态分布的关系来估计概率以便决策合作策略.将贝叶斯网络有效地应用在这两种跟踪模型中.所提出的算法均被应用于仿真序列图实验,证实了所提方法的有效性.
A new approach of multi-agent self-determining cooperation based on image understanding and intention recognition is presented.The algorithm is to estimate the relationship among agents in image which helps the observer make decision efficiently with agents refer to flying objects.We list two kinds of tracking model to analyse the problem.One of the models is to estimate the enemy′s trace to adjust its flight path,and the other is just to move towards the direction of the enemy.We set a complex function to ...
出处
《华中科技大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2008年第S1期42-45,共4页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金重点资助项目(60736010)
国防预研基金资助项目
关键词
图像理解
图像识别
多智能体协作
意图估计
贝叶斯网络
image understanding
image reognition
multi-agent cooperation
intention estimation
Bayesian network