Fazel, Zahra, Famouri, Mahmoud, Nazemi, Azadeh ORCID: 0000-0002-1138-309X and Azimifar, Zohreh (2018) Quick sift(QSIFT), an approach to reduce SIFT computational cost. In: 2017 Artificial Intelligence and Signal Processing Conference (AISP), 25-27 Oct. 2017, Shiraz, Iran.
Abstract
SIFT has been proven to be the most robust local rotation and illumination invariant feature descriptor. Being fully scale invariant is the most important advantage of this descriptor. The major drawback of SIFT is time complexity which prevents utilizing SIFT in real-time applications. This paper describes a method to increase the speed of SIFT feature extraction using keypoint estimation and approximation instead of keypoint calculation in various scales. This research attempts to decrease SIFT computational cost without sacrificing performance and propose quick SIFT method (QSIFT). The recent researches in this area have approved that direct feature value computation is more expensive than the value extrapolation. Consequently, the contribution of this research is to reduces the time execution without losing accuracy.
Metadata
Item Type: | Conference or Workshop Item (Paper) |
---|---|
Event Type: | Conference |
Refereed: | Yes |
Uncontrolled Keywords: | Interest points; Feature detector; Feature descriptor; Feature extraction; Feature matching; natural image statistics; real-time; Lighting; Cameras; Real-time systems; Signal processing; Computational efficiency; Time complexity |
Subjects: | UNSPECIFIED |
DCU Faculties and Centres: | DCU Faculties and Schools > Faculty of Engineering and Computing > School of Computing |
Published in: | 2017 Artificial Intelligence and Signal Processing Conference (AISP), Proceedings. . IEEE. |
Publisher: | IEEE |
Official URL: | http://dx.doi.org/10.1109/aisp.2017.8515128 |
Copyright Information: | ©2017 The Authors |
ID Code: | 23502 |
Deposited On: | 10 Jul 2019 14:13 by Azadeh Nazemi . Last Modified 03 Sep 2020 15:58 |
Documents
Full text available as:
Preview |
PDF
- Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
935kB |
Metrics
Altmetric Badge
Dimensions Badge
Downloads
Downloads
Downloads per month over past year
Archive Staff Only: edit this record