A overview of 3D reconstruction based on factorization

The problem of recovering both structure and motion from 2D image sequences is an important and essential task in the field of computer vision. During the last two decades, many approaches have been proposed for different applications. Among which, factorization based methods are widely studied and attracted many attentions in the computer vision society, since this kind of approach deals with all the data in all images uniformly, thus may achieve good robustness and accuracy. The factorization methods of recovering the 3D structure and motion of an object from image sequence are introduced systematically and detailedly in this paper, from the point of view on rigid object and non-rigid object. Merits and drawbacks of the techniques mentioned are analyzed and compared. Finally, some problems of current theories and methods are proposed, the problems which need further research and the development trend of this domain are presented.