Emotion Recognition based on Principal Component Analysis and Neural Network

• Main Authors: LIOU,SHIANG-WEI, 劉享緯
• Other Authors: LIEN,CHANG-HUA
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/98625436584738146426

碩士 === 國立高雄海洋科技大學 === 輪機工程研究所 === 104 === Human connect to each other based on standard emoticon. Emotions can affect mental stimulation, and mentally aware of their body language, such as anger and fear emotions are most commonly be discussed. Due to the autonomic nervous system situation is more obvious and easier to distinguish through the video image produced and can cause irritation of the autonomic nervous, reflex and physical changes. In general, autonomic nervous system is more commonly to be explored on a variety of emotional impact, autonomic nerves would be cognition, affected by the action, posture, the mental state of mind. Herein, we studied different facial expressions of recognition; vision for emotion affects the physiological changes related to the use of physiological signals can create a personal emotional interpretation. The study is focused on four kinds of facial emotions such as happy, surprise, sadness, and anger. These four kinds of emotions is using electromyography (EMG) to collect and detect the signals. In order to reduce the sample size, the principal components analysis physiological (PCA) method is used to reduce the signals dimension and feature extraction to obtain eigenvectors. Finally, the neural network classifies different facial expressions into various classifications, and then simulations demonstrate the effectiveness of the presented emotion recognition.