Lip Contour Modeling Based on Active Shape Model

We demonstrate a robust method of modeling lip contour using Active Shape Model(ASM), which is only able to deform in ways characteristic of the class of objects it represents. The method first labels the training set, then uses Procrustes Analysis to align the coordinates got by labeling, and then does Principle Component Analysis(PCA) on the aligned data, with which we can get the modes of lip contour variation, and the data got by PCA can be used to build the lip contour model. Experimental results indicate that the first four modes of the model account for different variations of the lip and subsequent modes describe more finer contour details, and also the average difference between the contour reconstructed by the model and the correspondingly original contour of the training set is all less than 0.6 pixels wide, which shows the model has strong generalization capabilities.