Emotion Recognition System featuring a fusion of Electrocardiogram and Photoplethysmogram Features

The research aims to evaluate the latency for emotion identification technology to upgrade the quality of human-computer interaction (HCI). The purpose of the research is to Design automatic systems that can recognize emotions automatically. This study is considered to be fruitful for many applications including in healthcare settings and HCI. The system is developed to evaluate different human emotions through Electrocardiogram (ECG) signals and photoplethysmograph (PPG) signals. The paper presents the recognition of human emotions based on some noticeable features of ECG and PPG signals. For this purpose, four emotional states such as disgust, fear, happy and sad are specified. Frequency transforms and statistical measures have been used to analyze several signals for these states. Multiple pre-processing techniques are used to filter out noise in the bioelectrical signals for true detection of the subject's emotional state. A total of 35 features were extracted and various classifiers to train the system. The best accuracy of this system is 85.7
^%
, yielded at Ensemble Bagged Trees.