Based On Spectral Correlation Of Hyperspectral Image Compression Method

Hyperspectral imaging technology is the foreland of the remote sensing development in the 21st century and is one of the most important focuses of the remote sensing domain. Hyperspectral images can provide much more information than multispectral images do and can solve many problems which can not be solved by multispectral imaging technology. However this advantage is at the cost of massy quantity of data that brings difficulties of images' storage and transmission. So it is very important to search a high effective hyperspectral images' data compression algorithm .Compared with other traditional remote sensors' data, hyperspectral images' data has both spatial correlation and spectral correlation. In order to get better compression result , we must consider remove the spatial correlation as well as the spectral correlation.This paper, based on particularly analyzing information characters in hyperspectral image data and nowadays compression algorithms .will investigate compression algorithm applying for hyperspectral images by analyzing spectral correlation .The main research results is followed:1) advancing the theory that the eigenvectors of hyperspectral images that are composed of the resemble objects have almost the same distribution;2) according to the above theory, this paper advance the near-KL transformation theory;3) implementing the algorithm as KL transformation + single band wavelet transformation + two dimentionai SPIHT coding;4) advancing and implementing the algorithm as KL transformation + multi-bands wavelet transformation + two dimensional SPIHT coding;5) implementing the algorithm as three dimensional wavelet transformation + three dimensional SPIHT coding;...