Authors:
Marc Steven Krämer
;
Simon Hardt
and
Klaus-Dieter Kuhnert
Affiliation:
Univesity of Siegen, Germany
Keyword(s):
Visual Odometry, Motion Estimation, Space Robotics, SIFT, SURF, BRIEF, ORB, KAZE, AKAZE, BRISK.
Related
Ontology
Subjects/Areas/Topics:
Applications
;
Computer Vision, Visualization and Computer Graphics
;
Feature Selection and Extraction
;
Human-Computer Interaction
;
Methodologies and Methods
;
Motion and Tracking
;
Motion, Tracking and Stereo Vision
;
Pattern Recognition
;
Physiological Computing Systems
;
Robotics
;
Sensors and Early Vision
;
Software Engineering
;
Theory and Methods
;
Virtual Environments
Abstract:
Image features are used in many computer vision applications. One important field of use is the visual navigation.
The localization of robots can be done with the help of visual odometry. To detect its surrounding, a robot
is typically equipped with different environment sensors like cameras or lidar. For such a multi sensor system
the exact pose of each sensor is very important. To test, monitor and correct these calibration parameters, the
ego-motion can be calculated separately by each sensor and compared. In this study we evaluate SIFT, SURF,
ORB, AKAZE, BRISK, BRIEF and KAZE operator for visual odometry in a space scenario. Since there was
no suitable space test data available, we have generated our own.