FACIAL EXPRESSION RECOGNITION TECHNOLOGY BASED ON FUZZY FUSION

摘要

This paper proposes a new method to recognize facial expression using fuzzy theory fusing local facial information after reference a great deal of expression recognition method. Segment the image into three parts, eyebrow, eyes and mouth, then recognize these by trained SVMs and fuse the result with fuzzy relation matrix. In the training, this paper chooses the Japanese women facial expression database (JAFFE) and radial basis function (RBF) as kernel function, while gauss function is adopted as membership function in the fusion. However this paper designs and implement the FFFER system of expression recognition based on fuzzy set theory. The experiment result show it is has a good effect on recognition rate.