TU Darmstadt / ULB / TUbiblio

Improving SURF Image Matching Using Supervised Learning

Mousselly-Sergieh, Hatem and Egyed-Zsigmond, Elöd and Döller, Mario and Coquil, David and Pinon, Jean-Marie and Kosch, Harald (2012):
Improving SURF Image Matching Using Supervised Learning.
In: Proceedings of the 2012 Eighth International Conference on Signal Image Technology and Internet Based Systems, IEEE Computer Society, In: SITIS '12, ISBN 978-0-7695-4911-8,
DOI: 10.1109/SITIS.2012.42, [Online-Edition: https://ieeexplore.ieee.org/document/6395100/],
[Conference or Workshop Item]

Abstract

Key points-based image matching algorithms have proven very successful in recent years. However, their execution time makes them unsuitable for online applications. Indeed, identifying similar key points requires comparing a large number of high dimensional descriptor vectors. Previous work has shown that matching could be still accurately performed when only considering a few highly significant key points. In this paper, we investigate reducing the number of generated SURF features to speed up image matching while maintaining the matching recall at a high level. We propose a machine learning approach that uses a binary classifier to identify key points that are useful for the matching process. Furthermore, we compare the proposed approach to another method for key point pruning based on saliency maps. The two approaches are evaluated using ground truth datasets. The evaluation shows that the proposed classification-based approach outperforms the adversary in terms of the trade-off between the matching recall and the percentage of reduced key points. Additionally, the evaluation demonstrates the ability of the proposed approach of effectively reducing the matching runtime.

Item Type: Conference or Workshop Item
Erschienen: 2012
Creators: Mousselly-Sergieh, Hatem and Egyed-Zsigmond, Elöd and Döller, Mario and Coquil, David and Pinon, Jean-Marie and Kosch, Harald
Title: Improving SURF Image Matching Using Supervised Learning
Language: English
Abstract:

Key points-based image matching algorithms have proven very successful in recent years. However, their execution time makes them unsuitable for online applications. Indeed, identifying similar key points requires comparing a large number of high dimensional descriptor vectors. Previous work has shown that matching could be still accurately performed when only considering a few highly significant key points. In this paper, we investigate reducing the number of generated SURF features to speed up image matching while maintaining the matching recall at a high level. We propose a machine learning approach that uses a binary classifier to identify key points that are useful for the matching process. Furthermore, we compare the proposed approach to another method for key point pruning based on saliency maps. The two approaches are evaluated using ground truth datasets. The evaluation shows that the proposed classification-based approach outperforms the adversary in terms of the trade-off between the matching recall and the percentage of reduced key points. Additionally, the evaluation demonstrates the ability of the proposed approach of effectively reducing the matching runtime.

Title of Book: Proceedings of the 2012 Eighth International Conference on Signal Image Technology and Internet Based Systems
Series Name: SITIS '12
Publisher: IEEE Computer Society
ISBN: 978-0-7695-4911-8
Uncontrolled Keywords: SURF, Image Matching, Classification
Divisions: 20 Department of Computer Science
20 Department of Computer Science > Ubiquitous Knowledge Processing
Date Deposited: 31 Dec 2016 14:29
DOI: 10.1109/SITIS.2012.42
Official URL: https://ieeexplore.ieee.org/document/6395100/
Identification Number: TUD-CS-2012-0373
Related URLs:
Export:

Optionen (nur für Redakteure)

View Item View Item