Affine Structure and Motion from Points, Lines and Conics

  • Authors:

  • Fredrik Kahl;Anders Heyden

  • Affiliations:

  • Centre for Mathematical Sciences, Lund University, Box 118, S-221 00 Lund, Sweden. fredrik@maths.lth.se;Centre for Mathematical Sciences, Lund University, Box 118, S-221 00 Lund, Sweden. heyden@maths.lth.se

  • Venue:
  • International Journal of Computer Vision
  • Year:
  • 1999

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper several new methods for estimating scenestructure and camera motion from an image sequence taken by affinecameras are presented. All methods can incorporate both point, lineand conic features in a unified manner. The correspondence betweenfeatures in different images is assumed to be known.Three newtensor representations are introduced describing the viewing geometryfor two and three cameras. The centred affine epipoles can be used toconstrain the location of corresponding points and conics in twoimages. The third order, or alternatively, the reduced third ordercentred affine tensors can be used to constrain the locations ofcorresponding points, lines and conics in three images. The reducedthird order tensors contain only 12 components compared to the 16components obtained when reducing the trifocal tensor to affinecameras.A new factorization method is presented. The noveltylies in the ability to handle not only point features, but also lineand conic features concurrently. Another complementary method basedon the so-called closure constraints is also presented. The advantageof this method is the ability to handle missing data in a simple anduniform manner. Finally, experiments performed on both simulated andreal data are given, including a comparison with othermethods.