Facial Expression Features Extraction Based on Local Parameterized Models
Shuai Shi · Ceshi jishu xuebao · 2007
An approach to extracting motion parameters of facial expression based on local parameterized models is proposed in this paper.First,a set of image sequence is preprocessed,and the facial features which can reflect expression are extracted.Secondly,the method of block-match is used to track the critical feature points.Then,the affine motion equation based on the result of block-match and the local parameterized model is created.Finally,a nonlinear constraint optimization way is adopted to estimate the motion parameters.The experiment result demonstrates that the method is feasible.