3D head texture using multiple kinect

Texture for 3D models is one important component of computer graphics. In this paper, we propose a method for generating a coherent, integrated and seamless texture on 3D head models. In the process, the color images and depth data captured by Kinect are established a correspondence to build initial texture on the head surface. Then the visible mismatching seams in the adjacent texture regions are effectively removed by an iterative coarse-to-fine strategy, resulting in an accurate aligned texture. In the following, the illumination variations among different color images are eliminated by color fusion in gradient domains. Finally, one texture is generated for the whole 3D head model by non-texture repair. The experiments demonstrate that the coherent and seamless texture method for 3D head is effective and robust.