Abstract | ||
---|---|---|
We present a practical, stratified autocalibration al- gorithm with theoretical guarantees of global optimality. Given a projective reconstruction, the first stage of the algo- rithm upgrades it to affine by estimating the position of the plane at infinity. The plane at infinity is computed by glob- ally minimizing a least squares formulation of the modulus constraints. In the second stage, the algorithm upgrades this affine reconstruction to a metric one by globally mini- mizing the infinite homography relation to compute the dual image of the absolute conic (DIAC). The positive semidefi- niteness of the DIAC is explicitly enforced as part of the optimization process, rather than as a post-processing step. For each stage, we construct and minimize tight convex relaxations of the highly non-convex objective functions in a branch and bound optimization framework. We exploit the problem structure to restrict the search space for the DIAC and the plane at infinity to a small, fixed number of branching dimensions, independent of the number of views. The convex relaxation techniques presented here are gen- eral enough that we expect them to be of use to computer vision researchers solving optimization problems in multi- view geometry and elsewhere. Experimental evidence of the accuracy, speed and scal- ability of our algorithm is presented on synthetic and real data. MATLAB code for the implementation is made avail- able to the community for facilitating further research. |
Year | DOI | Venue |
---|---|---|
2007 | 10.1109/ICCV.2007.4409114 | Rio de Janeiro |
Keywords | Field | DocType |
image processing,least squares approximations,optimisation,MATLAB code,dual image of the absolute conic,globally optimal affine-metric upgrades,infinite homography relation,least squares formulation,modulus constraints,nonconvex objective functions,optimization process,post-processing process,problem structure,stratified autocalibration algorithm | Affine transformation,Least squares,Plane at infinity,Mathematical optimization,Branch and bound,DIAC,Computer science,Regular polygon,Homography,Scalability | Conference |
Volume | Issue | ISSN |
2007 | 1 | 1550-5499 E-ISBN : 978-1-4244-1631-8 |
ISBN | Citations | PageRank |
978-1-4244-1631-8 | 6 | 0.44 |
References | Authors | |
26 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Manmohan Chandraker | 1 | 451 | 25.58 |
Sameer Agarwal | 2 | 10328 | 478.10 |
David Kriegman | 3 | 7693 | 451.96 |
Serge J. Belongie | 4 | 12512 | 1010.13 |