Gabor-max-DCT feature extraction techniques for facial gesture recognition

Abstract

Facial gesture recognition is an application based on pattern recognition which has applications in expression identification in criminal cases, investigation, and security purpose, human–machine interaction but still accuracy, illumination and occlusion are the research issues which have to improve. Key research issue of facial gesture identification is improving the accuracy of system which is measured in term of recognition rate and accuracy of system mostly depends on optimization of feature extraction. In the facial gesture, edge pattern (shape) and texture are unique pattern which have to extract from facial image. In this paper, Gabor filter is used to extract edge pattern from face but Gabor produces high-dimensional matrix with redundant edge information which is reduced optimally by proposed maximum discrete cosine transformation in order to improve accuracy of system.