Motion Parallax from Catastrophies in Scale-space

Autor:	E. Argante, C. C. A. M. Gielen, Tjeerd M. H. Dijkstra
Rok vydání:	1993
Předmět:	Physics Scale (ratio) business.industry Gaussian Optical flow Scale space symbols.namesake Position (vector) Computer Science::Computer Vision and Pattern Recognition symbols Computer vision Artificial intelligence Affine transformation Parallax business Rotation (mathematics)
Zdroj:	ICANN ’93 ISBN: 9783540198390
DOI:	10.1007/978-1-4471-2063-6_54
Popis:	We present a way to extract affine motion parallax structure from time-varying images. In our method we construct scale-space by convolving the image with gaussians of all scales and then track the scale-space structure around annihilation-points, where spatial derivatives of the image disappear catastrophically because of the gaussian blurring. We show that the position change of these points corresponds to the image velocity and the scale change is related to the expansion of the image. The change in structure around these points is related to the rotation and deformation of the image. Simulations show our method to be robust against noise.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::4079f5a7ef8e6b236886217e5a7e615d https://doi.org/10.1007/978-1-4471-2063-6_54 Zobrazit plný text záznamu