The Sparse Representation And Compression Algorithm Research Of Hyperspectral Image

Hyperspectral remote sensing technology developed in the late20th century, is emergingin the field of multidisciplinary cross disciplines such as electromagnetic, optics, signalprocessing. Compared with the traditional remote sensing technology, hyperspectral remotesensing technology can obtaining the ground information, at the same time obtainingabundant spectral information, making it widely used in agriculture, forestry, geological,environment, military and other fields. The continuously improving of hyperspectral remotesensing image spatial resolution and spectral resolution makes the amount of dataexponentially increasing, which brought great pressure to the transmission and storage. So thehyperspectral image compression algorithm is of vital significance to the development ofhyperspectral remote sensing technology.In actual application the amount of data hyperspectral remote sensing image is too large,and the growing contradictions have happened between information accessed and datatransmission. In this paper, based on the redundant dictionary of the sparse representation ofhyperspectral remote sensing image and the compression algorithm were studied. The mainresearch content is as follows:Firstly, this paper implements the sparse representation of hyperspectral images based onthe redundant dictionary. The method of sparse representation method based on redundantdictionary uses less the amount of data to better describe the characteristic information ofhyperspectral image, which is a kind of more effective way to represent hyperspectral image.Secondly, this paper studies a compression method of hyperspectral remote sensingimage based on the sparse representation. This method based on the sparse representation ofhyperspectral remote sensing images uses the bit plane coding method to the coefficient planeand we can obtain a higher compression ratio.Thirdly, the paper completes the reconstruction of hyperspectral remote sensing image,and obtains good effect.In this paper, we carry on a large number of simulation experiments, which show that thealgorithm can achieve good compression effect and good reconstruction results, and it has thegood generality in different hyperspectral image library.