Adam Schmidt

An Evaluation of Image Feature Detectors and Descriptors for Robot Navigation

Adam Schmidt, Marek Kraft, Andrzej KasiƄski

In: Computer Vision and Graphics

Series: Lecture Notes in Computer Science

Vol.: 6375

Year: 2010

Pages: 251--259

Publisher: Springer Berling/Heidelberg

Publisher's page: http://www.springerlink.com/content/up1177577j8k1665/

Abstract:

The detection and matching of feature points is crucial in many computer vision systems. Successful establishing of points correspondences between concurrent frames is important in such tasks as visual odometry, structure from motion or simultaneous localization and mapping. This paper compares of the performance of the well established, single scale detectors and descriptors and the increasingly popular, multi-scale approaches.