An efficient virtual aesthetic surgery model based on 2D color photograph

Abstract

In this paper, we propose a virtual aesthetic surgery (VAS) system using a deformation technique based on a radial basis function (RBF) and blending technique that combines the deformed facial component with the original face. The proposed VAS system is composed of three main steps. First, various deformation templates are matched to facial components by a multiresolution active appearance model (MAAM), which is trained by 2D color face images. Next, the VAS system computes the degree of deformation for lattice cells on the free-form deformation (FFD) using the proposed RBF. The deformation error is compensated for by the coefficients of the mapping function, which is recursively solved by the singular value decomposition (SVD) technique using the sum of squared error (SSE) between the deformed control points and target control points on the base curves. Finally, the deformed facial component is blended with the original face using a blending ratio that is computed by the modified Euclidean distance transform. Experimental results show that the proposed deformation and blending techniques are very efficient in terms of smoothness, accuracy, and distortion.