A study of facial emotion recognition techniques to examine micro-expressions

Abstract

Humans communicate with one another by speaking, gesticulating with their bodies, and expressing facial emotions. Among these methods, expressing emotions play an important role. Since human beings naturally use facial expressions to convey their emotions. Micro-expressions are perceptive facial expressions that last only a few seconds. Micro-expressions, as compared to regular facial expressions, will expose the majority of the latent, unconcealed emotional states. However, because of their shorter length, micro-expressions are more difficult to find. As a result, interest in micro-expression has grown in many fields, including defence, psychology, and computer vision, in recent years. This paper provides a brief overview of current methodologies for detecting human micro-emotions, with a focus on the LBP, LBPTOP, DCNN, 3DHOG, MMPTR, and DTCM feature extraction filter methods, which have been found to be more accurate. The theoretical accuracy of the LBP-TOP Feature Extraction method with SVM and KNN classifier combination was discovered to be better than the theoretical accuracy of all approaches. As a result, this paper also discusses those two classifiers.