Research On HSI Image Compression Technology Based On Compressed Sensing

The hyperspectral image can achieve the spatial information and spectral information simultaneously,and due to this unique characteristic,it is widely used in various fields.However,the increase of spectral information makes the hyperspectral image have the large amount of data,which is bad for the storage and transmission of data.The hyperspectral data,which is different from ordinary 2D image and video images,is a kind of 3D images.The compression of hyperspectral images is a new problem in the field of image compression.So analyzing the characteristics of hyperspectral image itself and studying the new suitable method for hyperspectral image compression is a worthy of study field.Compressive sensing(CS)is a novel sampling theory.CS theory is proved that a small number of incoherent measurements of a compressible signal contain enough information for exact reconstruction.Since the CS rate is far lower than the Nyquist rate,the requirements for sensors are greatly relaxed so that the problem of high cost caused by the blind pursuit of an excessively high-resolution sensor can be avoided.As one of the crucial issues,the CS reconstruction algorithm plays a key role in the application of CS theory and affects its practical implementation;hence it has been a hot study since it was presented.Under this background,this dissertation deeply studied the CS reconstruction algorithms for hyperspectral images in order to find effective and robust reconstruction algorithms,and also discussed the implementation framework of compressive imaging.The main contributions and innovation points of the dissertation are taken as follows:Firstly,this thesis introduces the compressive sensing theory,analyze the characteristics of hyperspectral image and achieve the relevant conclusions.The hyperspectral images have strong spectral correlation compared to the ordinary 2D digital images.The analysis of the hyperspectral image data is the preparation for the following compression method.Secondly,hyperspectral image adaptive grouping compression algorithm based on JSM-1 model is proposed aiming at the shortcomings of the existing research algorithm.The algorithm,taking advantage of the spectral similarity of the hyperspectral image,adaptively divides the hyperspectral image into several groups.Then this thesis applies the distributed compressive sensing JSM-1 model in hyperspectral image compression.The spectral similarity of the hyperspectral image is used,and the compression efficiency is improved.Experiments show the feasibility of the algorithm.Finally,hyperspectral image adaptive grouping compression algorithm based on JSM-1 model considers the spectral correlation but not enough.Adaptive grouping compression algorithm based on KLT transform is presented.The algorithm make KLT transform applying the strong correlation of the hyperspectral residual image CS data,and is further compressed on the basis of the algorithm in the chapter four,which can improve the compression effect.Experiments show the effectiveness of the proposed algorithm.