Optimal band selection for high dimensional remote sensing data using genetic algorithm

A 'fused' method may not be suitable for reducing the dimensionality of data and a band/feature selection method needs to be used for selecting an optimal subset of original data bands. This study examined the efficiency of GA in band selection for remote sensing classification. A GA-based algorithm for band selection was designed deliberately in which a Bhattacharyya distance index that indicates separability between classes of interest is used as fitness function. A binary string chromosome is designed in which each gene location has a value of 1 representing a feature being included or 0 representing a band being not included. The algorithm was implemented in MATLAB programming environment, and a band selection task for lithologic classification in the Chocolate Mountain area (California) was used to test the proposed algorithm. The proposed feature selection algorithm can be useful in multi-source remote sensing data preprocessing, especially in hyperspectral dimensionality reduction.

[1]  Jonathan Raper,et al.  Spatial evolutionary modeling , 2001, Spatial information systems.

[2]  John A. Richards,et al.  Remote Sensing Digital Image Analysis , 1986 .

[3]  Sankar K. Pal,et al.  Genetic Algorithms for Pattern Recognition , 2017 .

[4]  J.M. Daida,et al.  Evolving feature-extraction algorithms: adapting genetic programming for image analysis in geoscience and remote sensing , 1996, IGARSS '96. 1996 International Geoscience and Remote Sensing Symposium.

[5]  Fabio Maselli,et al.  Selection of optimum bands from TM scenes through mutual information analysis , 1993 .

[6]  Robert A. Neville,et al.  Spectral unmixing of hyperspectral imagery for mineral exploration: comparison of results from SFSI and AVIRIS , 2003 .

[7]  Xianfeng Zhang,et al.  Comparison of lithologic mapping with ASTER, hyperion, and ETM data in the southeastern Chocolate Mountains, USA , 2007 .

[8]  Sanghamitra Bandyopadhyay,et al.  Genetic algorithms for generation of class boundaries , 1998, IEEE Trans. Syst. Man Cybern. Part B.

[9]  David A. Landgrebe,et al.  Covariance Matrix Estimation and Classification With Limited Training Data , 1996, IEEE Trans. Pattern Anal. Mach. Intell..

[10]  John R. Jensen,et al.  Introductory Digital Image Processing: A Remote Sensing Perspective , 1986 .

[11]  H. Joel Trussell,et al.  Dimensionality reduction in hyperspectral image classification , 2004, 2004 International Conference on Image Processing, 2004. ICIP '04..

[12]  Mineichi Kudo,et al.  Comparison of algorithms that select features for pattern classifiers , 2000, Pattern Recognit..

[13]  Sebastiano B. Serpico,et al.  Introduction to the special issue on analysis of hyperspectral image data , 2001, IEEE Trans. Geosci. Remote. Sens..

[14]  P. Switzer,et al.  A transformation for ordering multispectral data in terms of image quality with implications for noise removal , 1988 .

[15]  Zbigniew Michalewicz,et al.  Parameter Control in Evolutionary Algorithms , 2007, Parameter Setting in Evolutionary Algorithms.

[16]  Christopher J. Brookes,et al.  A genetic algorithm for designing optimal patch configurations in GIS , 2001, Int. J. Geogr. Inf. Sci..

[17]  Cesare Alippi,et al.  Genetic-algorithm programming environments , 1994, Computer.

[18]  P. Groves,et al.  Methodology For Hyperspectral Band Selection , 2004 .

[19]  W. Farrand.  Mapping the distribution of mine tailings in the Coeur d'Alene River Valley, Idaho, through the use of a constrained energy minimization technique , 1997 .