Multi-focal tensors as invariant differential forms

Abstract

For each relative GL(V)-invariant tensor I∈ p1+1V .. pn+1V we construct a GL(V)-invariant weighted differential form η on (P V)n. Then η is expressed explicitly with respect to n-tuples of frames for tangent spaces at points of P V to obtain elements of a different tensor space. For certain invariants I, the resulting elements are shown to be the multi-focal tensors appearing in the machine vision literature (Demazure 1988, Longuet-Higgins 1981, Luong 1992, Faugeras 1993, Faugeras and Luong 2001, Hartley and Zisserman 2003). This generalizes the 3 multi-focal varieties known in dimension V = 4 to an infinite collection of special tensor subvarieties. We use this framework to exhibit a new system of degree 4 polynomial equations, reminiscent of the braid relation, satisfied by the Euclidean trifocal variety.

0

Turn this paper into a lesson

ArcXiv compiles a structured reading guide from this paper's metadata: plain-English importance, contributions, prerequisite concepts, which sections to read first, flashcards, and a quiz. Grounded in the abstract, never invented.

Discussion (0)

Sign in to join the discussion.

Loading comments…