A Fusion Method of Smile and Laugh Expression Classification

This paper proposed to build a smile expression classification system on data sets of GENKI that can represent real-world environments, and tested its implementation, in which we got the optimal recognition rate up to 86.197%. To deal with the features extraction problems, hybrid features (i.e., Gabor, PHOG, PLBP) are used, using hybrid recognition algorithms (i.e., GentleBoost, SVM) to classify, in this paper. Experiments showed the effectiveness of our methods.