Orientation Calculation of human face using Symbolic Computing techniques and ANFIS

Abstract

Face recognition techniques have come a long way since last six decade. There are several techniques which mainly focus on improving the performance of face recognition. In the present work symbolic computing and ANFIS is use. ANFIS classifier is used as neuro-fuzzy approach and symbolic computing is used for calculating the degree of orientation of human face. By using ANFIS mapping can be done between the input and output of the developed system. The whole process is divided into two part: modeling and understanding. The modeling phase is carried out using template of face image with different pose. Each template is representing a dummy face and an angle of orientation as a symbol. Different geometrical feature are extracted from dummy face which is store in data base as template and angle as symbol. This process has been carried out using switching pattern of the face. The understanding phasetesting process is carried out using unknown-view of human face image and sufficient amount of feature has been extracted and it is match with the templates which are already store in database. The feature extracted from unknown view of face image and matched with templates starting from first template. With the help of ANFIS training is done and during testing if features are matched with template feature then the switching process is stop and symbol is used as angle of orientation. Is features not matches then the next template is used to match the features in testing phase. If the classification treshhld value is 60% then employ fuzzy c means–clustering technique and check for the biometric code using Bidirectional associative memory technique (BAM) of artificial neural network. The algorithms have been tested and the result obtain so far yield to 90% accuracy for true and positive calculation of angle.