Development of an Automatic ECG-based Emotion Classification Algorithm

• Main Authors: Chien-HanHung, 洪千涵
• Other Authors: Jeen-Shing Wang
• Format: Others
• Language: en_US
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/hr92t2

碩士 === 國立成功大學 === 電機工程學系碩博士班 === 101 === This thesis presents a real-time ECG morphology feature based R-wave detection algorithm and an automatic ECG-based emotion classification algorithm for R-wave detection and human emotion classification, respectively. At first, we adopt a musical induction method to collect participants’ ECG signals without any deliberate laboratory setting, which can induce participants’ real emotional states. Next, the proposed real-time R-wave detection algorithm is presented to detect R-waves in ECG signal based on the ECG morphological features. Afterward, we develop an automatic ECG-based emotion classification algorithm to classify human emotions elicited by listening to music. Physiological ECG features generated from time-, frequency-domain, and nonlinear analyses are utilized to find the emotion-relevant features and to correlate them with emotional states. Subsequently, we develop a sequential forward floating selection-kernel-based class separability based (SFFS-KBCS-based) feature selection algorithm and utilize the generalized discriminant analysis (GDA) to effectively select significant ECG features associated with emotions and to reduce the dimensions of the selected features, respectively. Classifications of positive/negative valence, high/low arousal, and four types of emotion (Joy, Tension, Sadness, and Peacefulness) are performed by least squares support vector machine (LS-SVM) classifiers. The results show that the sensitivity, positive predictive value, and detection error rate of the real-time R-wave detection algorithm can achieve 99.97%, 99.89%, and 0.14%, respectively, and the average delay time of the proposed algorithm is only 15.1ms. Furthermore, the correct classification rates of the positive/negative valence, high/low arousal, and four types of emotion classification tasks are 82.78%, 72.91%, and 61.52%, respectively. Keywords: real-time R-wave detection, ECG, automatic emotion classification algorithm, musical induction.