Dense Match Summarization for Faster Two-view Estimation
This work addresses a computational efficiency problem for computer vision researchers and practitioners using dense matchers in two-view estimation, representing an incremental improvement.
The paper tackles the runtime bottleneck of robust two-view relative pose estimation from dense correspondences by proposing an efficient match summarization scheme, achieving comparable accuracy to using full dense matches with 10-100x faster runtime.
In this paper, we speed up robust two-view relative pose from dense correspondences. Previous work has shown that dense matchers can significantly improve both accuracy and robustness in the resulting pose. However, the large number of matches comes with a significantly increased runtime during robust estimation in RANSAC. To avoid this, we propose an efficient match summarization scheme which provides comparable accuracy to using the full set of dense matches, while having 10-100x faster runtime. We validate our approach on standard benchmark datasets together with multiple state-of-the-art dense matchers.