This paper examines the problem of tracking feature points in dynamic imagery. Two types of imaging geometries are examined: orthographic projection and perspective projection. In each case, the goal is to perform the feature point tracking with a minimal number of images. The feature point tracking problem is shown to be intractable and a grddey heuristic is presented.
All Science Journal Classification (ASJC) codes
- Signal Processing
- Computer Vision and Pattern Recognition
- Artificial Intelligence