Title
Barrel-Type Distortion Compensated Fourier Feature Extraction.
Abstract
Fourier based feature extraction is a common and powerful technique used in texture classification. In case of endoscopic imaging, often significant barrel-type distortions affect the feature extraction. Although images can be rectified using distortion correction techniques, in previous work feature extraction proved to suffer not just from geometric distortions, but also from effects within distortion correction. Distortion correction in combination with Fourier features has not been investigated so far. We introduce and evaluate three strategies to partially or completely compensate the geometric distortion in combination with Fourier features. With two methods, the interpolation within distortion correction can be omitted which should lead to a benefit in classification. Instead of making a general statement, we distinguish between certain frequencies and identify the positive and the negative aspects of the strategies.
Year
DOI
Venue
2013
10.1007/978-3-642-41914-0_6
ADVANCES IN VISUAL COMPUTING, ISVC 2013, PT I
Field
DocType
Volume
Computer vision,Distortion correction,Pattern recognition,Computer science,Interpolation,Fourier transform,Feature extraction,Artificial intelligence,Distortion
Conference
8033
ISSN
Citations 
PageRank 
0302-9743
5
0.45
References 
Authors
7
3
Name
Order
Citations
PageRank
Michael Gadermayr17415.65
Andreas Uhl21958223.07
Andreas Vécsei316718.36