PSYCH OpenIR  > 认知与发展心理学研究室
自动微表情识别研究
其他题名Investigations on the automatic micro-expression recognition system
吴奇
学位类型博士
导师傅小兰
2012-05
学位授予单位中国科学院研究生院
学位授予地点北京
学位专业心理学
关键词微表情 自动微表情识别 静态特征 动态信息 动态特征
摘要在现实生活中,有时人们会通过实施欺骗的方式来达到自己的目的,这往往需要人们对自身真实的情感进行压抑和隐藏。然而,这些被压抑和隐藏的真实情感有时会以一种非常快速的面部表情的形式被表达出来,这种表情被称为微表情。研究者认为,微表情是一种可用于谎言和危险意图检测的有力行为线索。然而,由于微表情的持续时间很短,普通人很难及时捕获或准确识别微表情。为此,在微表情的研究与应用过程中,人们往往需要使用编码工具对包含微表情的视频进行人工逐帧编码。这样的方法非常费时费力,使得目前的微表情研究和应用工作均进展缓慢。构建自动微表情识别系统,不仅能够为微表情研究者提供微表情分析的必需工具,而且在临床、司法及反恐等领域也具有广泛且重要的应用价值。 本研究基于计算机视觉和心理学已有研究,研发了基于静态特征的自动微表情识别系统。该系统综合利用了静态图像中的纹理与形状信息,并通过逐帧编码的方式,实现了对视频中的微表情的自动捕获与识别。 本研究还考察了微表情动态信息在微表情识别中的作用。实验结果显示,在接受过METT训练且METT训练有效的情况下,人依然不能利用微表情的动态信息对高强度微表情进行识别;但当微表情强度较低时,动态地呈现微表情将有助于提高微表情的识别准确率。 最后,本研究在上述研究结果的基础上,对基于动态特征的自动微表情系统的系统框架进行了初步探讨。本研究为进一步研发适用于任何情境的鲁棒的基于动态特征的自动微表情识别系统提供了重要的研究基础
其他摘要In situations in which individuals are motivated to conceal or repress their true emotions, their facial expressions may leak despite their efforts to conceal them. These leakages can be very useful for deception detection and many of these leakages are manifested in the form of micro-expressions. However, it is difficult for human to detect micro-expressions. Due to this incompetence, researchers have to manually inspect large amount of videos in frame by frame manner. It has become the greatest impediment for micro-expression studies. The combination of computer science and psychology research fields can provide the technical adjuncts to assist researchers and practitioners in micro-expression analysis. In this study, a novel approach for automatic micro-expression recognition is presented. By extracting the features of textures and shapes from static images, the final system are able to automatically spot and recognize the micro-expressions. This system will prove to be a very useful tool for the researchers who are interested in investigating the generation of micro-expressions. To build a system that is robust and applicable in clinical practice, research activity, national security, and criminal investigations, the system mentioned above must to be capable of extracting the facial dynamics from facial expressions. However, at present stage, it is difficult for the researchers to utilize the dynamic features of micro-expressions because of the appropriate psychological base for this method is still missing. To solve this problem, the effects of dynamic information on micro-expression recognition were investigated in this study. Results showed that, subjects were unable to utilize the dynamic information of micro-expressions to recognize intense micro-expressions even after receiving the METT training program. However, when the intensity of micro-expressions was low, the recognition accuracy of the subjects was promoted by dynamically presenting the micro-expressions. Based on the investigations mentioned above, this study proposed a preliminary system framework for the dynamic automatic micro-expression system. This study provides the algorithmic base and psychological base for building an automatic micro-expression recognition system which is robust across all situations.
学科领域基础心理学
语种中文
文献类型学位论文
条目标识符http://ir.psych.ac.cn/handle/311026/20417
专题认知与发展心理学研究室
作者单位中国科学院心理研究所
推荐引用方式
GB/T 7714
吴奇. 自动微表情识别研究[D]. 北京. 中国科学院研究生院,2012.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
自动微表情识别研究.pdf(2736KB)学位论文 限制开放CC BY-NC-SA浏览 请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[吴奇]的文章
百度学术
百度学术中相似的文章
[吴奇]的文章
必应学术
必应学术中相似的文章
[吴奇]的文章
相关权益政策
暂无数据
收藏/分享
文件名: 自动微表情识别研究.pdf
格式: Adobe PDF
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。