Motion and Structure from Optical Flow
Abstract
We show that a moving observer can (in principle) compute the relative depth map of the environment given only the instantaneous positional velocity field on his retina. The depth map (and surface normals) are computed by recovering the egomotion parameters which in turn determine the ratio of depth associated with any two neighbouring retinal points. The surface normal map may be computed to a degree of accuracy determined by the resolution of the retina, and by the accuracy of approximation obtainable in the solution of three non-linear equations of the third degree in three unknowns.
Cite
Text
Prazdny. "Motion and Structure from Optical Flow." International Joint Conference on Artificial Intelligence, 1979.Markdown
[Prazdny. "Motion and Structure from Optical Flow." International Joint Conference on Artificial Intelligence, 1979.](https://mlanthology.org/ijcai/1979/prazdny1979ijcai-motion/)BibTeX
@inproceedings{prazdny1979ijcai-motion,
title = {{Motion and Structure from Optical Flow}},
author = {Prazdny, K.},
booktitle = {International Joint Conference on Artificial Intelligence},
year = {1979},
pages = {702-704},
url = {https://mlanthology.org/ijcai/1979/prazdny1979ijcai-motion/}
}