基于局部异常行为检测的欺骗识别研究  

Methodologies for deception detection based on abnormal behavior

作　　者：夏凡[1] 王宏[1] 

机构地区：[1]清华大学智能技术与系统国家重点实验室,北京100084

出　　处：《智能系统学报》2007年第5期12-19,共8页CAAI Transactions on Intelligent Systems

基　　金：国家自然科学基金资助项目(60433030)

摘　　要：基于局部异常行为检测的欺骗识别研究是计算机视觉与情感计算领域中的一个新兴课题,其核心是利用计算机视觉技术从视频中检测、跟踪人的局部动作并进行分析,得出检测对象欺骗的可能性.它在安全监控、辅助面试、情感合成等领域均有着广阔的应用前景.从视觉技术的发展水平和常用处理方法入手,对欺骗识别研究的几个典型问题的研究现状进行了综述,并归纳了欺骗识别研究所涉及的相关理论与技术方案,最后结合课题研究给出了该研究的难点、急需解决的若干重要问题及未来的发展趋势.Research on deception detection based on visual cues for abnormal behavior is a new project in the area of computer vision and affective computing research. This research aims at detecting and tracking an individual＇s behavior via computer vision technology and analyzing the probability of it being deceptive. This has many promising applications in areas such as security surveillance, automated interview assistants, emotion synthesis, etc. This paper primarily focuses on overall methods and general characteristics of computer vision related to the detection process. A discussion on research challenges and future directions is also provided at the end of the paper.

关 键 词：情感计算 欺骗识别 计算机视觉 智能系统 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]