Band selection for hyperspectral remote sensing data through correlation matrix to improve image clustering

Hamed Gholizadeh

doi:10.1117/12.2027032

23 September 2013 Band selection for hyperspectral remote sensing data through correlation matrix to improve image clustering

Hamed Gholizadeh

Author Affiliations +

Proceedings Volume 8870, Imaging Spectrometry XVIII; 88700D (2013) https://doi.org/10.1117/12.2027032
Event: SPIE Optical Engineering + Applications, 2013, San Diego, California, United States

Abstract

Hyperspectral remote sensing is capable of providing large numbers of spectral bands. The vast amount of data volume presents challenging problems for information processing, such as heavy computational burden. In this paper, the impact of dimension reduction on hyperspectral data clustering is investigated from two viewpoints: 1) computational complexity; and 2) clustering performance. Clustering is one of the most useful tasks in data mining process. So, investigating the impact of dimension reduction on hyperspectral data clustering is justifiable. The proposed approach is based on thresholding the band correlation matrix and selecting the least correlated bands. Selected bands are then used to cluster the hyperspectral image. Experimental results on a real-world hyperspectral remote sensing data proved that the proposed approach will decrease computational complexity and lead to better clustering results. For evaluating the clustering performance, the Calinski-Harabasz, Davies-Bouldin and Krzanowski-Lai indices are used. These indices evaluate the clustering results using quantities and features inherent in the dataset. In other words, they do not need any external information.

Citation Download Citation

Hamed Gholizadeh "Band selection for hyperspectral remote sensing data through correlation matrix to improve image clustering", Proc. SPIE 8870, Imaging Spectrometry XVIII, 88700D (23 September 2013); https://doi.org/10.1117/12.2027032

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available