Research On Hyperspectral Image Compression For Remote Sensing Satellite

With the improvement and perfection of the remote sensing technology, the rise and the development toward ground resolution and spectrum resolution of remote sensing devices for plenty of remote sensing data have been seen, from which derive an obvious problem that how so big a mass of data can be transmitted to the surface from an outer space platform. The experts at system design have to keep it in mind that how to transmit so big amount of data with the restricted downward link of satellites. Consequently, the future remote sensing satellite systems are most likely to be limited by the band width of the link. For this purpose, the compression of the remote sensing data is necessitous.Hyperspectral images consist of one-dimension spectrum and two-dimension space information. They are wildly used in resource exploration, environmental inspection, military detection, geographic prospecting, and so forth. Besides the properties of common remote sensing images, the much greater number of spectrum bands and the higher spectrum resolution are the unique, hence the larger data quantity.On the basis of the research in wavelet theory and wavelet-based image coding techniques, the author presents the technique of hyperspectral remote sensing image compression for satellites. In the first part the wavelet analysis theory is systematically summarized in the perspective of signal analysis and digital signal processing. In succession, the excellent of the wavelet image coding techniques are introduced with the focus on the EBCOT coding algorithm. At the last part, aiming at the properties of hyperspectral images, three compression schemes are presented: DCT + EBCOT , differential method + EBCOT and predictive method + EBCOT . The three schemes are simulated and compared with others. With the 8:1 compression rate, experimental results show that these two compression schemes can both meet the basic requirements of compressing Hyperspectral images.At the end of this paper, all the work having been done is briefly reviewed and some ideas of the following work are presented.