Abstract | ||
---|---|---|
A modular approach to finding fast SIFT correspondences in single-image matching applications is proposed. Our algorithm exploits properties of the SIFT descriptor vector to find shortcuts to the most likely matches in a feature set. We are able to converge approximately 15 times faster than a linear search, and, respectively, four and five times faster than both PCA-SIFT and SURF (both of which use descriptor vectors that contain far fewer dimensions than SIFT), at near-equivalent recall and precision performance. |
Year | DOI | Venue |
---|---|---|
2009 | 10.1109/WACV.2009.5403099 | Applications of Computer Vision |
Keywords | Field | DocType |
feature extraction,image matching,SIFT descriptor,keypoint descriptor properties,scale invariant feature transform,single-image matching | Scale-invariant feature transform,Computer vision,GLOH,Pattern recognition,Computer science,Image matching,Precision and recall,Feature extraction,Feature set,Artificial intelligence,Modular design,Linear search | Conference |
ISSN | ISBN | Citations |
1550-5790 | 978-1-4244-5497-6 | 5 |
PageRank | References | Authors |
0.46 | 6 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Geoffrey Treen | 1 | 5 | 0.46 |
Anthony Whitehead | 2 | 143 | 20.84 |