Texturing and deforming models with casual images

Using images to texture 3D models to create realistic 3D models is an important task in computer graphics. However, if the object shape of the casual image is not similar to the target model, the textured model would be strange since the image is severely distorted. In this extended abstract, we present a novel texturing and deforming approach for mapping the texture and object shape of a casual image to a 3D model simultaneously based on an alternating least-square approach. Through a photogrammetric method, we project the target model onto the source image according to the estimated camera parameters. Then, the target model is deformed according to the object shape of the source image while minimizing the image distortion. The processes are performed iteratively. Our method can achieve texture mapping, shape deformation, and detail-preserving at once, and can obtain more reasonable texture mapped results than traditional methods.