| Yin Lou | Noah Snavely | Johannes Gehrke |
Abstract
Many new computer vision applications are utilizing large-scale datasets of places derived from the many billions of photos on the Web. Such applications often require knowledge of the visual connectivity structure of these image collections---describing which images overlap or are otherwise related---and an important step in understanding this structure is to identify {\em connected components} of this underlying image graph. As the structure of this graph is often initially unknown, this problem can be posed as one of exploring the connectivity between images as quickly as possible, by intelligently selecting a subset of image pairs for feature matching and geometric verification, without having to test all $O(n^2)$ possible pairs. We propose a novel, scalable algorithm called MatchMiner that efficiently explores visual relations between images, incorporating ideas from relevance feedback to improve decision making over time, as well as a simple yet effective {\em rank distance} measure for detecting outlier images. Using these ideas, our algorithm automatically prioritizes image pairs that can potentially connect or contribute to large connected components, using an information-theoretic algorithm to decide which image pairs to test next. Our experimental results show that MatchMiner can efficiently find connected components in large image collections, significantly outperforming state-of-the-art image matching methods.
Paper
MatchMiner: Efficient Spanning Structure Mining in Large Image Collections.
Yin Lou, Noah Snavely, and Johannes Gehrke.
ECCV'12, Florence, Italy. [Poster][BibTex]
@inproceedings{lou2012matchminer,
author = {Lou, Yin and Snavely, Noah and Gehrke, Johannes},
title = {MatchMiner: Efficient Spanning Structure Mining in Large Image Collections},
booktitle = {Proceedings of the 12th European Conference on Computer Vision},
year = {2012},
location = {Florence, Italy}
}
Datasets
Acknowledgements
We thank anonymous reviewers for their valuable comments. This research was supported by the NSF under grants IIS-0964027 and a grant from Intel.