On-board DPCM compression of hyperspectral data

Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) images are commonly adopted by the research community to evaluate compression algorithms suitable for hyperspectral data. The calibrated images collected in 1997 show sample value distributions which contain artificial regularities introduced by the conversion of raw data values to radiance units. Being optimal on images having flat histograms, classical DPCM methods do not obtain their best performances. Conversely, the lowest lossless bit rates are achieved by algorithms based on lookup-table (LUT) that significantly exploit these artifacts. The main consequence is that these performances can be misleading if they are extrapolated to images produced my more recent advanced sensors and especially to uncalibrated data. In fact, LUT-based algorithms do not achieve the best compression performances on the set of the 2006 AVIRIS images both raw and calibrated. The consequence is that LUT algorithms are not suitable for on-board lossless and lossy compression. Conversely, DPCM based algorithms can be easily adapted for on-board requirements. Goal of this paper is to provide a thorough comparison of compression methods especially on the 2006 AVIRIS data set. Both calibrated data (radiances) and raw data (digital counts) have been compressed. Results confirm that advanced DPCM-based methods, whose idea was originally developed by the authors in 2001, provide the best compression results. Relying on this idea, a new scheme is devised capable of working on the acquired data in BIL/BIP formats by properly choosing the causal prediction support and adopting an MMSE adaptive DPCM (MA-DPCM). Such a scheme is suitable for on-board implementations since data manipulations are reduced, thus limiting memory requirements.