Using the Linear Combinations between the samples for improving the Nonparametric Weighted...

Using the Linear Combinations between the samples for improving the Nonparametric Weighted Feature Extraction method in the Hyperspectral Images

SHARE

Abstract

Using the Linear Combinations between the samples for improving the Nonparametric Weighted Feature Extraction method in the Hyperspectral Images

Mohse Ghamary Asl
Student
K.N Toosi University,
Email: [email protected]

M. Reza Mobasheri
Dr.
K.N Toosi University
Email: [email protected]

M. Javad Valadan Zouj
Dr.
K.N Toosi University
Email: [email protected]

Barat Mojarradi
Dr.
K.N Toosi University
Email: [email protected]

In this paper, an improvement method is proposed for improving the Nonparametric Weighted Feature Extraction method, that is used for high dimensional pattern recognition problems. NWFE method is based on a nonparametric extension of scatter matrices, that the Mean parameters of them are computed separately for each sample, using by weighted summation of the other samples. The weights of each of these samples are computed based on their Euclidian distance from the under consideration sample (the sample that we want to calculate its weighted mean). However, using the distance parameter only, can not express the scatterings of samples, completely; and their Linear Combinations are effective for this purpose. In this paper, the Results of the Nonparametric Weighted Feature Extraction method has improved by using the Linear Combination between the samples.

Index Terms:

NWFE (Nonparametric Weighted Feature Extraction), Linear Combination, Discriminant Analysis, Dimensionality Reduction.