An information theory perspective on computational vision
An information theory perspective on computational vision作者机构:Department of StatisticsUniversity of California at Los AngelesLos AngelesCA 90095USA
出 版 物:《Frontiers of Electrical and Electronic Engineering in China》 (中国电气与电子工程前沿(英文版))
年 卷 期:2010年第5卷第3期
页 面:329-346页
学科分类:0808[工学-电气工程] 0809[工学-电子科学与技术(可授工学、理学学位)] 08[工学] 0805[工学-材料科学与工程(可授工学、理学学位)] 0702[理学-物理学] 0812[工学-计算机科学与技术(可授工学、理学学位)]
主 题:computer vision pattern recognition information theory minimum description length Markov random field(MRF)model stochastic grammars
摘 要:This paper introduces computer vision from an information theory *** discuss how vision can be thought of as a decoding problem where the goal is to find the most efficient encoding of the visual *** requires probabilistic models which are capable of capturing the complexity and ambiguities of natural *** start by describing classic Markov Random Field(MRF)models of *** stress the importance of having efficient inference and learning algorithms for these models and emphasize those approaches which use concepts from information *** we introduce more powerful image models that have recently been developed and which are better able to deal with the complexities of natural *** models use stochastic grammars and hierarchical *** are trained using images from increasingly large ***,we described how techniques from information theory can be used to analyze vision models and measure the effectiveness of different visual cues.