Sparse Coding based Compression of Spectrally Uncorrelated Hyperspectral Data Using Haar Wavelet Transform


Alaydin J. G. , Töreyin B. U.

24th Signal Processing and Communication Application Conference (SIU), Zonguldak, Turkey, 16 - 19 May 2016, pp.1945-1948 identifier

  • Publication Type: Conference Paper / Full Text
  • City: Zonguldak
  • Country: Turkey
  • Page Numbers: pp.1945-1948

Abstract

Sparse coding based compression of hyperspectral imagery yields better rate-distortion performance especially for low bit-rates when compared with other state-of-the-art methods in the literature. In this paper, an on-line dictionary learning based lossy compression method is proposed yielding even a better rate-distortion performance, thanks to the spectral decorrelation achieved by the Haar wavelet transform. The hyperspectral data is decorrelated in the spectral dimension using a single-level Haar transform which is followed by a dictionary learning step over the low-subband data. The higher subband is further compressed in a lossless manner using JPEG2000. Rate-distortion results are obtaind for AVIRIS hyperspectral data. Results indicate that the spectral decorrelation coupled with sparse dictionary learning of low-subband images yield superior performance over existing hyperspectral data compression schemes.