Segmentation in Dynamic Image Sequences by Isolation of Coherent Wave Profiles
Abstract
A segmentation and velocity estimation technique is presented which treats each object (either moving or stationary) as a distinct intensity wave profile. The Fourier components of wave profiles — and equally of objects — which move with constant velocity exhibit a regular frequency-dependent phase change. Using a Hough transform which embodies the relationship between velocity and phase change, moving objects are isolated by identifying the subset of the Fourier components of the total image intensity wave profile which exhibit this phase relationship. Velocity is measured by locating local maxima in the Hough space and segmentation is effected by re-constituting the moving wave profile — the object — from the Fourier components which satisfy the velocity/phase-change relationship for the detected velocity.
Cite
Text
Vernon. "Segmentation in Dynamic Image Sequences by Isolation of Coherent Wave Profiles." European Conference on Computer Vision, 1996. doi:10.1007/BFB0015545Markdown
[Vernon. "Segmentation in Dynamic Image Sequences by Isolation of Coherent Wave Profiles." European Conference on Computer Vision, 1996.](https://mlanthology.org/eccv/1996/vernon1996eccv-segmentation/) doi:10.1007/BFB0015545BibTeX
@inproceedings{vernon1996eccv-segmentation,
title = {{Segmentation in Dynamic Image Sequences by Isolation of Coherent Wave Profiles}},
author = {Vernon, David},
booktitle = {European Conference on Computer Vision},
year = {1996},
pages = {293-303},
doi = {10.1007/BFB0015545},
url = {https://mlanthology.org/eccv/1996/vernon1996eccv-segmentation/}
}