Use of independent component analysis for lossless compression of ultraspectral sounder data

Abstract

Independent component analysis has been known for its success in blind source separation. It features a decorrelation capability beyond second-order moments. Recently ICA has been used in lossy compression for target detection in hyperspectral imager data where loss of unimportant features does not affect the detection result. For the ultraspectral sounder data, it is better to be lossless compressed for the ill-posed retrieval of geophysical parameters. In this paper we will try to use ICA in the lossless compression of ultraspectral sounder data. The compression result shows that ICA compares favorably with JPEG2000, JPEG-LS, and CCSDS IDC 5/3 on the ten standard AIRS granules dataset.

Lossless Compression of Ultraspectral Images Using Recursive Least Square with Constant Gain Vectors

“…Vector quantization (VQ) is applied to each partition separately. In [2], independent component analysis (ICA) was applied for lossless compression of ultraspectral data. Wei and Huang [3] proposed a modified Tunstall code which grows an optimal non-exhaustive parse tree by assigning the complete codewords only to top probability nodes in the infinite tree, based on an infinitely extended parse tree.…”