基于广义贝叶斯网的图像模式识别框架研究  

Study on image pattern recognition framework based on generalized Bayesian network

作　　者：李甦[1] 谭永龙[1] 陈新亿[1] 

机构地区：[1]云南大学省电子计算中心,云南昆明650091

出　　处：《计算机工程与设计》2010年第19期4268-4271,共4页Computer Engineering and Design

基　　金：云南省自然科学基金项目(2008CD085)

摘　　要：为增强机器自适应筛选能力和减少人工干预并最终实现机器自动认知的目标,通过对贝叶斯网的改进,提出了广义贝叶斯网。在其基础上结合图像模式识别理论,定义了图像知识的集合运算、新知识加入等运算以及基于广义贝叶斯网的图像模式识别框架的知识映射变换、表达、存储的统一的抽象结构。在理论上找到了一种图像模式识别框架中知识表达的统一方法,而且对提高图像模式识别过程的机器参与度具有实际的指导意义。To enhance adaptive filtering capacity of the machine, reducing manual intervention and the eventual realization of the machine auto recognize goal, through the Bayesian network improvements, generalized Bayesian network is proposed. In combination ofimage pattern recognition, the set operation and new knowledge joining operation are defined, a unified abstract structure for image know- ledge mapping transformation, expression, storage is defined. This framework not only finds a unified image knowledge representation approach, but also has practical significance of improving the machine involvement in image pattern recognition.

关 键 词：数字图像处理 模式识别 模式识别框架 知识系统 对象表达 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]