Computer Science – Computer Vision and Pattern Recognition
Scientific paper
2011-11-21
Computer Science
Computer Vision and Pattern Recognition
This paper has been withdrawn
Scientific paper
We address the problem of distributed matching of features in networks with vision systems. Every camera in the network has limited communication capabilities and can only exchange local matches with its neighbors. We propose a distributed algorithm that takes these local matches and computes global correspondences by a proper propagation in the network. When the algorithm finishes, each camera knows the global correspondences between its features and the features of all the cameras in the network. The presence of spurious introduced by the local matcher may produce inconsistent global correspondences, which are association paths between features from the same camera. The contributions of this work are the propagation of the local matches and the detection and resolution of these inconsistencies by deleting local matches. Our resolution algorithm considers the quality of each local match, when this information is provided by the local matcher. We formally prove that after executing the algorithm, the network finishes with a global data association free of inconsistencies. We provide a fully decentralized solution to the problem which does not rely on any particular communication topology. Simulations and experimental results with real images show the performance of the method considering different features, matching functions and scenarios.
Aragues Rosario
Montijano Eduardo
Sagues Carlos
No associations
LandOfFree
Distributed Multi-view Matching in Networks with Limited Communications does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with Distributed Multi-view Matching in Networks with Limited Communications, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Distributed Multi-view Matching in Networks with Limited Communications will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-376747