Hashem Tamimi and Andreas Zell

Global Robot Localization using Iterative Scale Invariant Feature Transform

In Proc. of the 36th International Symposium on Robotics (ISR), Tokyo, Japan, November 2005.


Abstract

We propose an approach that can reduce the feature extraction time of the Scale Invariant Feature Transform (SIFT). The main idea is to search for the keypoints around a set of randomly generated particles rather than to perform exhaustive search in the whole difference of Gaussian pyramid. The proposed approach makes it possible to de- fine the required number of keypoints in advance. A relation between the number of required features and the feature extraction time can be drawn. The matching time of two sets of keypoints is consequently minimized. To demonstrate its performance, the approach is applied to the robot global localization problem. The results are successful, with much less keypoints and less computation time than the original SIFT.

Download

[pdf]

Bibtex 

@INPROCEEDINGS{Tamimi05-isr1, 
 AUTHOR = "Hashem {Tamimi} and Andreas {Zell}", 
 TITLE = "Global Robot Localization using Iterative Scale Invariant Feature Transform", 
 BOOKTITLE =    "In Proc. of the 36th International Symposium on Robotics (ISR)", 
 ADDRESS ="Tokyo, Japan", 
 YEAR = "2005" 
}