The structure tensor in projective spaces

作者: Klas Nordberg

DOI:

关键词:

摘要: The structure tensor has been used mainly for representation of local orientation in spaces arbitrary dimensions, where the eigenvectors represent and corresponding eigenvalues indicate type which is represented. Apart from being local, may be referred to as "object centered" since it describes relative a reference system. This paper proposes that basic properties can extended defined projective space rather than Euclidean space. result, "projective tensor", symmetric same way tensor, also uses eigensystem carry relevant information. However, instead orientation, represents geometrical primitives such points, lines, planes (depending on dimensionality underlying space). Furthermore, this useful property mapping operation forming affine hull points lines summation, e.g., sum two tensors amounts line passes through etc. "view each still scale, geometric primitive global image based implies combined, using meaningful over large regions.

参考文章(5)
Olivier D. Faugeras, What can be seen in three dimensions with an uncalibrated stereo rig european conference on computer vision. pp. 563- 578 ,(1992) , 10.1007/3-540-55426-2_61
Hans Knutsson, Carl-Fredrik Westin, Mats Andersson, Representing local structure using tensors II scandinavian conference on image analysis. pp. 545- 556 ,(2011) , 10.1007/978-3-642-21227-7_51
Josef Bigun, Gösta H. Granlund, Optimal Orientation Detection of Linear Symmetry international conference on computer vision. pp. 433- 438 ,(1987)
Hans Knutsson, GFosta H. Granlund, Signal processing for computer vision ,(1994)
Gunnar Farnebäck, Spatial Domain Methods for Orientation and Velocity Estimation Linköping University Electronic Press. ,(1999)