Abstract | ||
---|---|---|
This paper presents a homography-based approach for tracking multi- ple planar templates with central catadioptric cameras (which include per- spective cameras). We extend the standard notion of homography to this wider range of devices through the unied projection model on the sphere. To enforce the same movement of the camera for the different planes, we parametrise the homography by the Lie algebra of the special euclidean group SE(3) and estimate the normal and depth for each plane. With this model, we use a minimisation technique to obtain a close to second-order convergence rate with a complexity similar to a rst-order approach. The proposed method takes into account the non-uniform resolution of the sensor and proved ro- bust to poor initial values for the plane normals. To assess the precision of the approach, the developed algorithm was tested on the estimation of the displacement of a mobile robot in a real application. We compare the results when the planes are tracked independently to the constrained case. We show that the proposed minimisation approach leads to better results in terms of speed and precision than current tracking algorithms. |
Year | Venue | Keywords |
---|---|---|
2006 | BMVC | convergence rate,second order,lie algebra,mobile robot |
Field | DocType | Citations |
Computer vision,Euclidean group,Computer science,Minimisation (psychology),Homography,Planar,Artificial intelligence,Rate of convergence,Mobile robot,Homography (computer vision),Catadioptric system | Conference | 10 |
PageRank | References | Authors |
0.72 | 17 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Christopher Mei | 1 | 534 | 25.90 |
Selim Benhimane | 2 | 475 | 32.79 |
Ezio Malis | 3 | 1322 | 80.65 |
Patrick Rives | 4 | 227 | 13.90 |