Fisher Ratio Optimization under Volterra Filtering Model for Identification of Polarimetric Air/Air Range Profiles of Aircrafts
- 1 April 2007
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 10975659,p. 817-821
Abstract
This paper introduces an original supervised classification method based on the optimization of the Fisher ratio under the Volterra filtering model. Optimum Volterra filters are derived for the problem of discriminating 2 classes. They have properties yielding some decision strategy that aims to reject (i.e. take no decision) data corresponding to unlearned classes. An original extension of this solution is then proposed for the problem of discriminating more than 2 classes. Performances of our method are evaluated by classifying real data of jet fighters range profiles. Influence of the filtering model order is investigated. Interest of the waveform polarization states is also analyzed. Results obtained show that our method outperforms the nearest neighbor classification method in term of error probability, while yielding equivalent performances in term of correct identification probability. Moreover, it is shown that good rejection probability of unlearned classes can be obtained, particularly with fully-polarized data.Keywords
This publication has 14 references indexed in Scilit:
- KPCA plus LDA: a complete kernel Fisher discriminant framework for feature extraction and recognitionIEEE Transactions on Pattern Analysis and Machine Intelligence, 2005
- Trained detection of buried mines in SAR images via the deflection-optimal criterionIEEE Transactions on Geoscience and Remote Sensing, 2004
- Why can LDA be performed in PCA transformed space?Pattern Recognition, 2003
- Identification of Volterra kernels using interpolationIEEE Transactions on Instrumentation and Measurement, 2002
- Volterra Filtering of Noisy Images of Curves2002
- Multiclass linear dimension reduction by weighted pairwise Fisher criteriaIEEE Transactions on Pattern Analysis and Machine Intelligence, 2001
- Multichannel linear and quadratic adaptive filtering based on the Chandrasekhar fast algorithmIEEE Transactions on Signal Processing, 1999
- Nonlinear Component Analysis as a Kernel Eigenvalue ProblemNeural Computation, 1998
- A Unifying View of Second-Order Measures of Quality for Signal ClassificationIEEE Transactions on Communications, 1980
- The optimum nonlinear features for a scatter criterion in discriminant analysisIEEE Transactions on Information Theory, 1977