'Nulling' Filters and the Separation of Transparent Motions

Abstract

Using low-order global motion hypotheses and the assumption that there are no more than two motions at a single point, it is possible to successfully decompose motion stimuli that contain additively combined transparent layers. It is assumed that the space of flow fields is sufficiently smooth that a relatively coarse sampling of the flow parameter(s) will produce a set of vector fields that can be combined to reasonably approximate the actual motions in the scene. The definition of support from an exclusively spatial notion is extended to include the spatio-temporal energy domain. The key insight is that when processing transparent motion displays, the support of a motion hypothesis should exist over both a region of space and velocity, so that it can be isolated both spatially and in terms of local velocity.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Cite

Text

Darrell and Simonecelli. "'Nulling' Filters and the Separation of Transparent Motions." IEEE/CVF Conference on Computer Vision and Pattern Recognition, 1993. doi:10.1109/CVPR.1993.341150

Markdown

[Darrell and Simonecelli. "'Nulling' Filters and the Separation of Transparent Motions." IEEE/CVF Conference on Computer Vision and Pattern Recognition, 1993.](https://mlanthology.org/cvpr/1993/darrell1993cvpr-nulling/) doi:10.1109/CVPR.1993.341150

BibTeX

@inproceedings{darrell1993cvpr-nulling,
  title     = {{'Nulling' Filters and the Separation of Transparent Motions}},
  author    = {Darrell, Trevor and Simonecelli, Eero Peter},
  booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year      = {1993},
  pages     = {738-739},
  doi       = {10.1109/CVPR.1993.341150},
  url       = {https://mlanthology.org/cvpr/1993/darrell1993cvpr-nulling/}
}