3D bilaterally symmetric object shape recovery by shading

Abstract

Nowadays, the 3D models are widely adapted in many applications such as video games, movies and TV commercials. The traditional way to construct a 3D model is using 3D software by professional artist. It cannot be easily done by untrained people. Therefore, my project applies another method to recover the 3D shape from images, which can simplify the modeling process. In this project, we aim to present an approach for shape recovery based on integrating geometric and photometric information. We try to recover 3D bilaterally symmetric objects from a single image. The symmetric property provides us geometric information and the intensity image provides us photometric information. The basic idea is that an image can be regarded as a pair of images. Each image of the pair is half of the object with different viewing and lighting directions since it is a bilaterally symmetric object. By integrating the photometric and geometric information can yield a dense correspondence between pairs of symmetric points, and a result, a dense shape recovery of the object.