Research On Compression Method Of Hyper-spectral Image

Hyperspectral remote sensing images obtain image and spectral information of the earth's surface at the same time, through analyzing and processing the spectral data reasonably and effectively, the images can be widely used in various fields. However, the large amount of data in the hyperspectral image has caused great difficulties in its transmission and storage, which severely limits its application and development, therefore, it is necessary to carry out effective compression, hyperspectral image compression and processing technology has become an important part of the modern science and technology. Based on the existing research results, this thesis studies the method of hyperspectral image compression deeply. The main contents of the article are:First, in view of the existing algorithms, the spectrum characteristics of hyperspectral images are not fully utilized, a hyperspectral image compression scheme based on adaptive band clustering PCA combined with JPEG2000 is proposed. Based on the AP algorithm, an adaptive band clustering algorithm is designed, PCA operations is performed on each band group after clustering, finally, the static image compression standard JPEG2000 is used to encode all the principal components. The experimental results show that the proposed scheme can significantly improve the compression performance compared with the contrast algorithm.Second, aiming at the traditional data compression framework, the high speed sampling process will inevitably lead to the serious waste of resources, the coding end is too complex, and so son, the compressed sensing theory is introduced into the hyperspectral image compression, and an in-depth study of the existing hyperspectral image compression algorithm based on compressed sensing mode is studied in the present study.Third, a new compression scheme for hyperspectral images is proposed which based on variable projection rate sub block compressive sensing and reconstruction of optimized inter spectral prediction. At the encoder, all bands of the hyperspectral image are divided into several groups, the class center of each group is the reference band, and the rest is the common band. Then, different bands are used to separate the compressed sensing with different precision in order to obtain the measured value. At the decoder, reconstruction of the reference band using sparse adaptive matching pursuit algorithm, for reconstruction of the common band, a new model of optimized inter spectral prediction combined with SAMP algorithm is designed: firstly, the common band is predicted by means of the reconstructed reference band, and compressive projecting it, then calculating the residual error of the projection value of prediction before and after for the common band, finally, the SAMP algorithm is used to reconstruct the residual error, which is used to correct the prediction value. The experimental results show that the proposed scheme can significantly improve the overall compression performance compared with other similar algorithms.