Nonrigid Registration Algorithm in the Face Recognition Problem

Algorithms for aligning point clouds in the 3D space under the assumption of the nonrigidity of these clouds play an essential role in solving the problems of searching for particular facial parts and key points that could be used to recognize the facial expressions with various face deformations. The common nonrigid iterative closest point registration techniques are conventionally based on the affine point-to-point approach. In this study, to apply a nonrigid iterative closest point algorithm to facial surfaces, a model of the facial surface is built as a 3D point cloud and then used with a new nonrigid iterative closest point algorithm. The efficiency of the proposed algorithm is illustrated by the results of computational modeling.