Efficient facial expression detection by using the Adaptive-Neuro-Fuzzy-Inference-System and the Bezier curve

Abstract

Human beings have long been interested in detecting gestures of other human beings, especially facial gestures. The emotional states of humans are difficult to computational glean from their facial gestures; none the less, this recognition task is widely applied in daily life. In human communication the facial expressions recognition plays very important role and it is applied in many real worlds’ applications. Present methods for facial expression recognition in static images are not completely utilized and considered features for muscle movements and facial features, which show dynamic, statics, and geometric as well as the emergence characteristics of the facials expressions. In our paper, we develop the system that can recognize the basic emotional states efficiently and in real time based on human facial expressions. This system uses the adaptive neuro-fuzzy-inference system (ANFIS) and the Bezier curves. The system was evaluated based on the human facial images obtained from the databases of Japanese female facial expression and Cohn Kanade. The proposed approach achieves high correct recognitions rate inthe comparisons of the state of the art techniques.