Publica
Hier finden Sie wissenschaftliche Publikationen aus den FraunhoferInstituten. Using the KarhunenLoeve expansion for feature extraction on small sample sets
 European Center for Mechatronics; Society of Instrument and Control Engineers of Japan; IEEE Industrial Electronics Society: 24th Annual Conference of the IEEE Industrial Electronics Society, IECON 1998. Proceedings. Vol.3 Piscataway, NJ, USA: IEEE, 1998 ISBN: 0780345037 ISBN: 0780345045 ISBN: 0780345053 S.15821586 
 IEEE Industrial Electronics Society (IECON Annual Conference) <24, 1998, Aachen> 

 Englisch 
 Konferenzbeitrag 
 Fraunhofer IPA () 
 KarhunenLoeve; Merkmalsextraktion; PCA  Principle Component Analysis; Signalverarbeitung 
Abstract
A general task in the field of signal processing and control consists in representing signals by a small number of features, i.e. the mapping of a ndimensional signal onto a mdimensional feature vector. Examples for widely used features are position of maxima, minima, gradients or certain integral values of the signal curve, although it is obvious that the original curves can hardly be reconstructed from these features. The KarhunenLoeve (KL) expansion on the other hand provides the means to approximate the ndimensional sample vectors of a random distribution by m features such that the meansquare error is minimized. However, these m basis vectors of the KLexpansion may not be appropriate for representing signals which are not distributed according to the covariance matrix of the given sample set. In this paper a method is presented which improves the representation of signals with a distribution deviating from the given distribution by reducing the dependence of the feature ve ctors on the sample set. Feature vectors spanning a subspace of slowly varying functions are combined with basis vectors of the KLexpansion in order to increase the robustness of the features.