Constrained Texture Mapping Algorithms For 3D Face Model

Computer-Aided 3D craniofacial reconstruction is a technology to recover the 3D face model from skull data using computer graphics,which can be applied to criminal investigation, archeology,medical visualization,face animation and so on.In this paper,we propose an approach of constrained texture mapping for 3D face model to strengthen the realism of the craniofacial reconstruction results.The constrained texture mapping algorithm is an approach of texture mapping,and it can maintain the alignment of the corresponding features in the texture mapping process by the predefined constrained points.Firstly,we mark constrained points on the model and texture,the constrained points are used to match the texture coordinate for feature points,and we interpolate these points onto the 3D mesh.In this step,we optimize the new mesh with local optimization procedure(LOP) to make sure that the mesh is Delaunay triangle mesh.Secondly, we divide the texture by Delaunay triangulation.According to the topology of the constrained points in the texture,we divide the mesh into triangles by approximated shortest path.Thirdly, we use the Radial Basis Function(RBF) interpolating to get the un-constrained points texture coordinate from the association of the constrained points.Finally,we optimize the result to reduce the distortion of texture.Compared with other methods,our method doesn't need to parameterize 3D mesh to 2D space.In our algorithm,we divide the facial model to small triangles and then we use the RBF to implement the texture mapping.In this way,we ensure that the result of the texture mapping is valid and make the result smoother.Furthermore,we developed a constrained texture mapping system with the aforementioned method.In our system,a good texture-mapping result can immediately be generated by semi-automatically marking some constrained points on the corresponding 3D face model and texture image,and the experimental results validate our method.