Study on a method of restoring the 3D virtual nail model from the 2D image data based on the GFFD deformation method

This paper proposes a novel method of generating a 3D virtual nail model from multiple camera image data based on the GFFD deformation method. 3D virtual nail models for individuals can be generated easily and rapidly with considerably low cost. The outline of the whole procedures is as follows. First, we take three pictures of an individual´s nail. Second, we extract operation points of the nail in each image, which is arranged on the outline and surface of the nail. Third, we obtain the corresponding 3D positions of the operation points using a stereo computing method. Finally, we deform the 3D generic virtual nail model into an individual´s one with GFFD based on the obtained 3D positions of the operation points. We made some experiments to evaluate whether the errors of the restored nail shape model is within the acceptable error range. By comparing the restored nail model with the model obtained by scanning the nail with a 3D digitizer, we evaluated the accuracy of the restored model. The result of the experiment shows that we can get individual nail models with sufficient accuracy.