We investigate the use of the L/sub /spl infin// cost function in geometric vision problems. This cost function measures the maximum of a set of model-fitting errors, rather than the sum-of-squares, or L/sub 2/ cost function that is commonly used (in least-squares fitting). We investigate its use in two problems; multiview triangulation and motion recovery from omnidirectional cameras, though the results may also apply to other related problems. It is shown that for these problems the L/sub /spl infin// cost function is significantly simpler than the L/sub 2/ cost. In particular L/sub /spl infin// minimization involves finding the minimum of a cost function with a single local (and hence global) minimum on a convex parameter domain. The problem may be recast as a constrained minimization problem and solved using commonly available software. The optimal solution was reliably achieved on problems of small dimension.
[1]
R. Hartley.
Triangulation, Computer Vision and Image Understanding
,
1997
.
[2]
Richard Szeliski,et al.
High-quality Image-based Interactive Exploration of Real-World Environments 1
,
2003
.
[3]
Richard I. Hartley,et al.
Chirality
,
2004,
International Journal of Computer Vision.
[4]
Alexandru Tupan,et al.
Triangulation
,
1997,
Comput. Vis. Image Underst..
[5]
Bernhard P. Wrobel,et al.
Multiple View Geometry in Computer Vision
,
2001
.
[6]
Richard Szeliski,et al.
Image-based interactive exploration of real-world environments
,
2004,
IEEE Computer Graphics and Applications.
[7]
William H. Press,et al.
Numerical recipes in C
,
2002
.
[8]
R. Vanderbei.
LOQO user's manual — version 3.10
,
1999
.