Correlation-based Band Ordering Method for Lossless Compression of Hyperspectral Images

O. Kubasova, P. Toivanen, and J. Mielikäinen (Finland)


Compression, Hyperspectral Images, Correlation, Minimum Spanning Tree, Prim's Algorithm.


An efficient band ordering algorithm for lossless compression of hyperspectral images is presented in this paper. Main purpose of the proposed algorithm is to increase compression ratio dealing with hyperspectral images. At the first stage the algorithm considers interband correlation dependencies in a spectral image and uses them for optimal band ordering. Then linear prediction and entropy coding is applied for each image pixel. Band ordering based on correlation is shown to be an effective method for ordering of bands for lossless compression of hyperspectral images. The obtained results are very good, the achieved compression ratio is 1 percent better compared to the use of natural ordering for the image bands. Also the computation time is very short compared to the other band ordering methods.

Important Links:

Go Back