Abstract | ||
---|---|---|
This paper addresses the problem of recovering 3D structure of small archeological objects under affine projection. Affine projection is an approximation to the perspective projection which works quite well when object size and depth are small compared with the distance between the camera and the object. A new technique for affine reconstruction from two affine images is developed, which consists in decomposing the problem into two subproblems: contour matching based on invariant descriptors and 3D recovering based on matching results and calibration parameters. First we compute affine invariants descriptors on extracted contours and then we estimate affine transformation that best align the matched contours: a scale factor a, a shift value l(0) and an affinity matrix A are computed. This step establishes point-to-point matching that preserves the circular order of contours points. Second we use matching results in addition with camera parameters, previously determined, to retrieve the whole 3D model. In order to validate the proposed method, some experiences on small archeological 3D objects in stereovision context are exposed. |
Year | Venue | Field |
---|---|---|
2012 | COMPUTATIONAL MODELLING OF OBJECTS REPRESENTED IN IMAGES: FUNDAMENTALS, METHODS AND APPLICATIONS III | Computer vision,Harris affine region detector,Computer science,Affine invariant,Artificial intelligence,3D reconstruction |
DocType | Citations | PageRank |
Conference | 0 | 0.34 |
References | Authors | |
0 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Maweheb Saidani | 1 | 0 | 0.34 |
Faouzi Ghorbel | 2 | 361 | 46.48 |