A fuzzy descriptor for analyzing the differences of facial expression

摘要

This paper proposes an improved fuzzy semantic model (IFSM) to analyze the differences of facial expressions. Comparison with traditional methods. Our method has two advantages: (i) The proposed method breaks the image limit such as scale, illumination; (ii) According to fuzzy rules, the method can interpret expression features using semantic concepts. Then, the FEI and CK+ datasets are applied in analyzing the differences among various expressions. At last, in order to compare the performance, the state-of-the-art methods have been executed. The results illustrate our method has a better accuracy, but also can interpret expression features well.