Face expression recognition based on difference of improved AAM features and kernel-support vector machines
Wanjin Wang · Machinery Design and Manufacture · 2011
Active Appearance Model(AAM)is an efficient method,which extracts and combines the model features of the shape and appearance in order to extract facial expression features.Contrasting with other AAM fitting algorithms,inverse compositional algorithm has the fast and efficient characteristics.However,when the influence of different illumination and the nonlinearity are undetachable,the fitting effect of the algorithm could be worse.Thus,it presents a facial expression recognition method based on difference AAM feature and kernel support vector machine,and experiment is done in JAFFE facial expression database,which results show that the method can well remove the Illumination interference and resolve the problem of nonlinear identification,and obtain the robust recognition results as well.