Spatial and spectral decorrelation of lossless data compression for LANDSAT-TM imagery

Presents a spatial and spectral decorrelation scheme for lossless data compression of remotely sensed imagery. The statistical properties of Landsat-TM data are investigated. Measurements of the statistical information indicate that some bands have strong spectral correlation which can be used for interband prediction or spectral decorrelation; whereas other bands may be most suitable for spatial decorrelation within single band. Experiments show that data compression has been improved using both spatial and spectral nature of the remotely sensed data.<>