Advanced Techniques for Scene Analysis
Abstract
The overall objective of this project is to develop advanced techniques for scene analysis. Specifically, the problems of image registration, tracking, and change detection are considered. In this project, the following new techniques have been developed: 1) robust feature-based algorithm for object tracking, 2) motion-segmentation-based technique for change detection, 3) a target detection algorithm that consists of image differencing, maximum-margin classifier, and diversity combining, 4) a rotation-invariant transform for change detection, 5) a depth-based image registration algorithm, 6) an image registration algorithm that leverages wavelet, 7) a machine learning algorithm to automatically recover 3D surface from sparse 3D points, 8) an automatic surface fitting method for 3D reconstruction from 2D video sequence, and 9) a depth-based image registration method via geometric segmentation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 2010
- Accession Number
- ADA530357
Entities
People
- Dapeng Wu
Organizations
- University of Florida