Improved SIFT performance evaluation against various image deformations

Author

Liu Li ; Yu Hongyang

fYear

2014

Firstpage

172

Lastpage

175

Abstract

Scale Invariant Feature Transform (SIFT) is an effective algorithm in feature detection and scene matching. In this paper, we proposed an improved Scale Invariant Feature Transform (SIFT) based on ^D2OG keypoints detector for better real time performance and explored the performance of 64D, 96D and 128D SITF feature descriptors on standard test datasets. Results shows that the improved Scale Invariant Feature Transform has a big progress in the real time performance and the 64D, 96D SIFT feature descriptors performs as well as the traditional 128D SIFT feature descriptor for image matching at a significantly reduced computational cost.

Keywords

feature extraction; image matching; natural scenes; transforms; 128D SITF feature descriptor; 64D SITF feature descriptor; 96D SITF feature descriptor; D²OG keypoint detector; computational cost reduction; feature detection; image deformations; image matching; improved SIFT performance evaluation; real time performance; scale invariant feature transform; scene matching; standard test datasets; Algorithm design and analysis; Computer vision; Detectors; Feature extraction; Noise; Standards; Transforms; D²OG; SIFT; image matching; scale; viewpoint;

fLanguage

English

Publisher

ieee

Conference_Titel

Information Technology and Artificial Intelligence Conference (ITAIC), 2014 IEEE 7th Joint International

Print_ISBN

978-1-4799-4420-0

Type

conf

DOI

10.1109/ITAIC.2014.7065029

Filename

7065029