Determining temporal persistence and consistent edge motion from natural images
Abstract
Temporal persistence and consistent edge motion are introduced as measures of temporal continuity in images; the former is the period of time over which a spatial feature has stably persisted in the image plane, while the latter occurs when an edge moves in a manner consistent with its previous trajectory. These features are presently computed from natural imagery on the basis of an event-driven method for the determination of edges. This computation places constraints on the variation of the intensity surface. Previous image motion is used as a low-level context for the determination of whether current motion measurements are the result of modeling intensity variations. The approach is demonstrated with natural imagery.
- Publication:
-
IN: AIAA Computers in Aerospace Conference
- Pub Date:
- 1987
- Bibcode:
- 1987coae.conf..287K
- Keywords:
-
- Computer Vision;
- Image Processing;
- Motion Perception;
- Edges;
- Low Pass Filters;
- Spatial Filtering;
- Instrumentation and Photography