Share Email Print
cover

Proceedings Paper

Band selection method based on spectrum difference in targets of interest in hyperspectral imagery
Author(s): Xiaohan Zhang; Guang Yang; Yongbo Yang; Junhua Huang
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

While hyperspectral data shares rich spectrum information, it has numbers of bands with high correlation coefficients, causing great data redundancy. A reasonable band selection is important for subsequent processing. Bands with large amount of information and low correlation should be selected. On this basis, according to the needs of target detection applications, the spectral characteristics of the objects of interest are taken into consideration in this paper, and a new method based on spectrum difference is proposed. Firstly, according to the spectrum differences of targets of interest, a difference matrix which represents the different spectral reflectance of different targets in different bands is structured. By setting a threshold, the bands satisfying the conditions would be left, constituting a subset of bands. Then, the correlation coefficients between bands are calculated and correlation matrix is given. According to the size of the correlation coefficient, the bands can be set into several groups. At last, the conception of normalized variance is used on behalf of the information content of each band. The bands are sorted by the value of its normalized variance. Set needing number of bands, and the optimum band combination solution can be get by these three steps. This method retains the greatest degree of difference between the target of interest and is easy to achieve by computer automatically. Besides, false color image synthesis experiment is carried out using the bands selected by this method as well as other 3 methods to show the performance of method in this paper.

Paper Details

Date Published: 25 October 2016
PDF: 8 pages
Proc. SPIE 10156, Hyperspectral Remote Sensing Applications and Environmental Monitoring and Safety Testing Technology, 101560J (25 October 2016); doi: 10.1117/12.2244818
Show Author Affiliations
Xiaohan Zhang, Air Force Aviation Univ. (China)
Guang Yang, Air Force Aviation Univ. (China)
Yongbo Yang, Air Force Aviation Univ. (China)
Junhua Huang, Air Force Aviation Univ. (China)


Published in SPIE Proceedings Vol. 10156:
Hyperspectral Remote Sensing Applications and Environmental Monitoring and Safety Testing Technology

© SPIE. Terms of Use
Back to Top