The Quantification of Human Facial Expression Using Trapezoidal Fuzzy Membership Function

Abstract

Fuzzy Inference System is an interesting approach. Major benefit of the FIS is, it permits the natural narration in linguistic terms of tribulations that can be resolved rather than in requisites of associations between accurate arithmetical points. This helps, handling with the complicated systems in easy way, is the major motive why fuzzy system is broadly incorporated in practice. In the present research paper, an effective approach is proposed that quantifies the human facial expression using Mamdani implication based fuzzy logic system. The recent principle engages in retrieving arithmetical values from person’s face and feed them to a fuzzy classifier. Fuzzification and Defuzzification process issues trapezoidal fuzzy membership function for input as well as output. The diverse characteristic of this method is its effortlessness and maximum correctness. Experimental outcome on Image dataset depicts excellent accomplishment of the proposed methodology. In this paper, a legitimate procedure proposed for quantification of human facial expression from the features of the face by means of Mamdani type fuzzy inference system, which is proficient to set up a convenient membership association involving the various dimensions of the happy expression. Values representing features of the face are fed to a Mamdani-type fuzzy classifier. This system recognizes three levels of same happy expression namely Normal, Bit Smiley and Loud Laugh. The total output expressions for this proposed scheme is three. Another discrete element of the proposed methodology is the membership method model of expression outcome which stands on various surveys and readings of psychology.