Recognition of emotions using non-contact breadth analyzer

Abstract

Automation of emotion detection has been done in many methods. The method followed here is detection of emotions through the pattern of respiration. A non-contact infrared temperature sensor is used to detect the pattern of respiration. The signals acquired from the sensor in analyzed. 10 statistical features are extracted. Feature reduction is done by applying Principle component analysis and 4 PCAs are obtained. These features are applied to supervised and unsupervised classification algorithm. Trilayered neural network has an accuracy of 95.5%