Recognition of expression which uses face feature

摘要

To recognize face expression is very important issue in natural language understanding. In this paper, a new method for recognizing face expression, which consists of a preprocessing face expression image process, a detecting face area process, a extracting the characteristic process, and a recognizing the expression process, is presented. The recognized expressions in our research are seven kinds of expressionless, happy, surprise, disgust, fear, sadness and anger. In the method, for improving the system robustness, the characteristic for each face expression is extracted by using FACS, and then they are compared with a face expression dictionary using a minimum distance method.