Model of the Facial Emotions Expressions Based on Grouping Classes of Feature Vectors

The characteristic forms of facial expressions of the emotional states of a human are typical of a rather large degree of generalization on the basis of common physiological structures and the location of the muscles that form the human face. This circumstance is one of the main reasons for the commonality of human display of emotions that are reflected in the face. By the nature and form of facial expressions on the face it is possible, with high probability, to determine the emotional state of a human with some correction on the part of cultural characteristics and traditions of certain groups. In accordance with the existence of common mimic forms of emotional displays, an approach is proposed to create a model of recognition of emotional displays on a human’s face with relatively low requirements for the means of photo, video-fixation. The creation of the model is based on the implementation of the hyperplane classification of mimic displays of major emotional states. One of the main advantages of the proposed approach is the low computational complexity, which makes it possible to implement a system of recognition of changes in the emotional state of a human by mimic displays without the use of specialized equipment. In addition, the model formed on the basis of the proposed approach allows obtaining proper recognition accuracy with low requirements for quality image characteristics, which allows extending the scope of practical application to a great extent. One example of practical application is the control of the driver while driving a transport, complex production operator and other automatic visual surveillance systems. The set of identified emotional states is formed in accordance with the assigned tasks and gives the opportunity to focus on the recognition of facial expressions and to group characteristic structural displays based on the set of distinguished characteristic features.