Compression of 3D facial data

The paper describes a facial mesh compression method based on an optimization algorithm to improve the quality of rendered facial animation. The proposed facial mesh compression algorithm removes the center vertex of a hexagon which consists of six triangles and represents that area with four triangles. To find the optimal vertex for removal, an area based distortion measure is employed. Different reshape models are also considered to minimize the distortion between the original and reconstructed mesh. From different reshape models, one optimum shape is selected to represent the original facial mesh. The optimization algorithm chooses one vertex which yields minimum distortion. Experimental results show that the proposed algorithm reduces data by 20% while maintaining the quality of the rendered facial animation.