A New Approach For Facial Expression Recognition Based On Parallel Adaptive Nerue-Fuzzy System
as in interaction between human and machine we need interaction and machine should be able to recognize the humans emotion when it is in relation with machine, it seems necessary to have a system that can recognize some emotions of humans. In this essay we give the image to the system and after detection of face in image, we detect the area of brows, eyes, nose and mouth. And in this areas we detect of facial points and the distance between these points are use in system. In our proposed method, in order to recognize the facial expression we use ANFIS in two different methods, in the first method: we used one ANFIS for recognition 3 facial expressions (happiness, sadness and normal) and in the second method we use for each facial expression an ANFIS. System (in the first method) after training has the 92.74 percent recognition on FG-Net database. Also with using from an ANFIS for each facial expression (Total 3 ANFIS for 3 facial expressions and 1 ANFIS on output of 3 ANFIS in the second method) system has the 96.28 percent recognition on FG-Net database that has relative to the first method 3.54 percent improvement.
Nerue-Fuzzy System ANFIS facial expression recognition
Seyed AH Mohammadi Mahdi Yaghoobi
Artificial Intelligence department Islamic Azad University, Mashhad Branch Mashhad, Iran
国际会议
重庆
英文
129-132
2011-01-21（万方平台首次上网日期，不代表论文的发表时间）