Improving SURF Image Matching Using Supervised Learning

Abstract : Keypoints-based image matching algorithms have proven very successful in recent years. However, their execution time makes them unsuitable for online applications. Indeed, identifying similar keypoints 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 keypoints. 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 keypoints that are useful for the matching process. Furthermore, we compare the proposed approach to another method for keypoint 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 keypoints. Additionally, the evaluation demonstrates the ability of the proposed approach of effectively reducing the matching runtime.
Document type :
Conference papers
Complete list of metadatas

https://hal.archives-ouvertes.fr/hal-01353098
Contributor : Équipe Gestionnaire Des Publications Si Liris <>
Submitted on : Wednesday, August 10, 2016 - 4:22:26 PM
Last modification on : Friday, January 11, 2019 - 5:09:21 PM

Identifiers

  • HAL Id : hal-01353098, version 1

Citation

Hatem Mousselly-Sergieh, Elod Egyed-Zsigmond, Mario Döller, David Coquil, Jean-Marie Pinon, et al.. Improving SURF Image Matching Using Supervised Learning. The 8th International Conference on Signal Image and Internet Systems (SITIS 2012), Nov 2012, Naples, Italy, Italy. pp.230-237. ⟨hal-01353098⟩

Share

Metrics

Record views

256