Linear Object Classes and Image Synthesis from a Single Example Image
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 techniques that are applicable under restricted conditions but...
| Main Authors: | , |
|---|---|
| Language: | en_US |
| Published: |
2004
|
| Online Access: | http://hdl.handle.net/1721.1/6635 |
| _version_ | 1826212782297579520 |
|---|---|
| author | Vetter, Thomas Poggio, Tomaso |
| author_facet | Vetter, Thomas Poggio, Tomaso |
| author_sort | Vetter, Thomas |
| collection | MIT |
| description | 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 techniques that are applicable under restricted conditions but simpler. 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. In this paper, we introduce such a new technique by extending the notion of linear class first proposed by Poggio and Vetter. For linear object classes 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. |
| first_indexed | 2024-09-23T15:37:50Z |
| id | mit-1721.1/6635 |
| institution | Massachusetts Institute of Technology |
| language | en_US |
| last_indexed | 2024-09-23T15:37:50Z |
| publishDate | 2004 |
| record_format | dspace |
| spelling | mit-1721.1/66352019-04-11T02:52:37Z Linear Object Classes and Image Synthesis from a Single Example Image Vetter, Thomas Poggio, Tomaso 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 techniques that are applicable under restricted conditions but simpler. 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. In this paper, we introduce such a new technique by extending the notion of linear class first proposed by Poggio and Vetter. For linear object classes 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. 2004-10-08T20:35:58Z 2004-10-08T20:35:58Z 1995-03-01 AIM-1531 http://hdl.handle.net/1721.1/6635 en_US AIM-1531 13231252 bytes 887715 bytes application/postscript application/pdf application/postscript application/pdf |
| spellingShingle | Vetter, Thomas Poggio, Tomaso Linear Object Classes and Image Synthesis from a Single Example Image |
| title | Linear Object Classes and Image Synthesis from a Single Example Image |
| title_full | Linear Object Classes and Image Synthesis from a Single Example Image |
| title_fullStr | Linear Object Classes and Image Synthesis from a Single Example Image |
| title_full_unstemmed | Linear Object Classes and Image Synthesis from a Single Example Image |
| title_short | Linear Object Classes and Image Synthesis from a Single Example Image |
| title_sort | linear object classes and image synthesis from a single example image |
| url | http://hdl.handle.net/1721.1/6635 |
| work_keys_str_mv | AT vetterthomas linearobjectclassesandimagesynthesisfromasingleexampleimage AT poggiotomaso linearobjectclassesandimagesynthesisfromasingleexampleimage |