Issue No. 07 - July (1997 vol. 19)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/34.598230
<p><b>Abstract</b>—The need to generate new views of a 3D object from a single real image arises in several fields, including graphics and object recognition. While the traditional approach relies on the use of 3D models, we have recently introduced [<ref rid="bibi07331" type="bib">1</ref>], [<ref rid="bibi07332" type="bib">2</ref>], [<ref rid="bibi07333" type="bib">3</ref>] simpler techniques that are applicable under restricted conditions. The approach exploits image transformations that are specific to the relevant object class, and learnable from example views of other "prototypical" objects of the same class.</p><p>In this paper, we introduce such a technique by extending the notion of linear class proposed by Poggio and Vetter. For <it>linear object classes</it>, it is shown that linear transformations can be learned exactly from a basis set of 2D prototypical views. We demonstrate the approach on artificial objects and then show preliminary evidence that the technique can effectively "rotate" high-resolution face images from a single 2D view.</p>
3D object recognition, rotation invariance, deformable templates, image synthesis.
T. Vetter and T. Poggio, "Linear Object Classes and Image Synthesis From a Single Example Image," in IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 19, no. , pp. 733-742, 1997.