Minimum Surface Bhattacharyya feature selection

Author

Gonzalez, Jose Andres ; Mendenhall, Michael J. ; Merenyi, Erzsebet

Author_Institution

Air Force Inst. of Technol., Electr. & Comput. Eng., Wright-Patterson AFB, OH, USA

fYear

2009

fDate

26-28 Aug. 2009

Firstpage

1

Lastpage

4

Abstract

This paper introduces a novel feature selection method called minimum surface Bhattacharyya (MSB). The method is applicable for multiple class problems utilizing supervised training. The minimum surface method selects features by means of inter-class separability. For the purposes of this paper, the method is applied to a hyperspectral data set with high correlations among the features. The method shows promise for hyperspectral analysis due to its speed and demonstrated capacity to improve classification performance.

Keywords

data handling; image processing; learning (artificial intelligence); hyperspectral analysis; hyperspectral data set; hyperspectral images; machine learning; minimum surface Bhattacharyya feature selection method; pattern classification performance; supervised training; Decision trees; Filters; Histograms; Hyperspectral imaging; Image storage; Military computing; Performance analysis; Prototypes; Runtime; Sorting; Bhattacharyya coefficient; dimensionality reduction; feature selection; machine learning;

fLanguage

English

Publisher

ieee

Conference_Titel

Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, 2009. WHISPERS '09. First Workshop on

Conference_Location

Grenoble

Print_ISBN

978-1-4244-4686-5

Electronic_ISBN

978-1-4244-4687-2

Type

conf

DOI

10.1109/WHISPERS.2009.5289044

Filename

5289044