The paper presents a method for groundplane estimation from image-pairs even if unstructured environment and motion. In a typical outdoor multi-camera system the observed objects might be very different due to the noise coming from lighting conditions, camera positions. Static features such as color, shape, and contours cannot be used for image matching in these cases. In the paper a method is proposed for matching partially overlapping images captured by video cameras. Using co-motion statistics, which is followed by outlier detection and a nonlinear optimization, does the matching. The described robust algorithm finds point correspondences in two images without searching for any structures and without tracking any continuous motion. Real-life outdoor experiments demonstrate the feasibility of this approach.
|Number of pages||8|
|Journal||Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|
|Publication status||Published - Dec 1 2004|
ASJC Scopus subject areas
- Theoretical Computer Science
- Computer Science(all)