The Research And Implementation Of Coding Method In Remote Sensing Image Compression

Remote sensing image compression is an important research in information andsignal processing. It is widely used in geological mapping, atmospheric radiation,water inversion, object recognition and so on.On the whole, the compressionmethod can be summarized into three methods, the first compression method isbased on Transform; the second compression method is based on prediction and thethird compression method is based on vector quantization. In remote sensing imagecompression method based on transform, using tensor to decompositionhyperspectral image is a novel technology, which not only removes the correlationof space, but also removes the correlation of spectrum. Then it combines3D-SPECKencoding method, the image is compressed without the loss of information.In addition, the paper also study the entropy coding part of the JPEG XRcompression standard, its compression performance close to JPEG2000and thecomputational complexity is much lower. Because of its low memory consumption,low computational complexity, low memory consumption and high efficiencyencoding, it not only supports both lossy compression and lossless compression, butalso supports regional decoding, HDR and so on. The time of compression can beimproved, So JPEG XR entropy coding complemented in the FPGA has importantapplications.Firstly, this paper introduces the tensor decomposition and JPEG XR standards,analyses the advantages and disadvantages of these two algorithms. Secondly, thearticle proposes two methods of remote sensing image compression which arehyperspectral images compression method based on tensor decomposition andoptical images compression based on JPEG XR. Remote sensing image compressionperformed a comprehensive analysis from software to hardware. Finally, the paperconducted simulation experiments for the two compression methods. Inhyperspectral images compression method based on tensor decomposition, thispaper compares the3D-DWT which improve SNR about5dB with other methods.During optical images compression based on JPEG XR, this paper compares withthe traditional JPEG2000algorithm. The encoding and decoding time and signal tonoise ratio of the reconstructed image are compared under the same compressionrate, the compression time of JPEG XR is1/4of JPEG2000.