%0 Journal Article
%T 基于多特征融合及背景建模的微表情Apex帧检测<br>Micro-Expression Apex Frame Detection Based on Multi Feature Integration and Background Modeling
%A 马韵洁
%A 王佐成
%A 吴艳平
%A 王飞
%J Computer Science and Application
%P 147-157
%@ 2161-881X
%D 2024
%I Hans Publishing
%R 10.12677/CSA.2024.141016
%X <span style=\"font-size:14px;\">微表情是人们在情感在波动时体现在面部的细微变化。根据心理学的研究，微表情在心理治疗等领域有着广泛应用，而Apex帧能够表达微表情的最丰富信息，为了更准确地提取Apex帧，本文研究了图像序列中面部微表情中的Apex帧检测问题。本文提出了一个新的检测Apex帧方法，在频率域中对面部区域采用分块方法进行背景建模，检测出脸部运动区域，随后通过统计运动区域的面积达到检测Apex帧的目的。将提出的方法应用于CASME、CASME II等数据集中，实验结果表明本文提出的方法能够有效地探测定位到Apex帧。</span><br />
<span style=\"font-size:14px;\">Micro-expressions are subtle facial changes that reflect fluctuations in emotions. According to psy-chological research, micro-expressions find widespread application in areas such as psychotherapy. Among these expressions, the Apex frame encapsulates the richest information. In order to accu-rately extract Apex frames, this study investigates the detection problem within facial mi-cro-expressions in image sequences. This paper proposes a novel method for detecting Apex frames. In the frequency domain, a block-based approach is employed for background modeling in facial regions, identifying regions of facial movement. Subsequently, the detection of Apex frames is achieved by statistically analyzing the area covered by the moving regions. The proposed method is applied to datasets such as CASME and CASME II, with experimental results demonstrating its effec-tiveness in detecting and locating Apex frames.</span>
%K 微表情，特征融合，Apex帧，分块方法，计算机视觉<br>Micro-Expression
%K Feature Integration
%K Apex Frame
%K Chunking Method
%K Computer Vision
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=80316