Share Email Print

Proceedings Paper

Spatiotemporal filtering for visual motion estimation from real images
Author(s): Patrizia Baraldi; Massimo Tistarelli; Giulio Sandini; Francesca Gandolfo
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

The effects of regularization techniques applied to the estimation of visual motion are investigated. A new method to compute the optical flow from a sequence of time-varying images is used. It is based on second-order derivatives of the image brightness and it is applied to evaluate 3-D motion parameters like the time-to-impact from translational motion. In particular, quantitative results are presented on the influence that spatio-temporal filtering of real image sequences with 2-D and 3-D (where the third dimension is time) smoothing operators has on the estimation of such parameters. The experiments show that, performing a three-dimensional filtering of the image sequence, remarkable accuracy can be reached, even using a small spatial kernel. Moreover, computing the optical flow only on intensity edges it gives the same 3-D motion parameter estimate as considering the whole field, provided that a small spatial kernel is used. The experiments presented in this paper demonstrate the applicability of the methods to a large number of computer vision applications.

Paper Details

Date Published: 14 February 1992
PDF: 11 pages
Proc. SPIE 1613, Mobile Robots VI, (14 February 1992); doi: 10.1117/12.135182
Show Author Affiliations
Patrizia Baraldi, Univ. of Genoa (Italy)
Massimo Tistarelli, Univ. of Genoa (Italy)
Giulio Sandini, Univ. of Genoa (Italy)
Francesca Gandolfo, Univ. of Genoa (Italy)

Published in SPIE Proceedings Vol. 1613:
Mobile Robots VI
William J. Wolfe; Wendell H. Chun, Editor(s)

© SPIE. Terms of Use
Back to Top
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?