Research On Hyperspectral Image Classification Algorithm Based On Albedo And Sparse Representation

Hyperspectral image classification is always an important content in remote sensing field,which can be applied in forestry,fishery,agriculture and so on.Sparse representation and cooperative representation are widely applied in hyperspectral image classification.However,information redundancy,hughes phenomenon,the same thing with different spectrums and different things with the same spectrum bring great difficulties to hyperspectral image classification.In order to improve classification accuracy,based on cooperative representation and sparse representation,and combined with the albedo feature in the image,the following two improved hyperspectral image classification algorithms are proposed.In view of not making full use of spectral information,spatial information and albedo features of hyperspectral images,hyperspectral image classification algorithm based on spatial spectrum fusion and albedo restoration is proposed.Firstly,the algorithm restores the albedo features of images.Secondly,the spectral and spatial information of the processed image is extracted by using Spatial spectrum feature learning network.Finally,the Laplacian feature mapping matrix of manifold learning algorithm is introduced into the kernel cooperative representation as a regular term,which increases the classification decision ability of the algorithm.Finally,the Laplacian feature mapping matrix of the manifold learning algorithm is introduced into the kernel cooperative representation as a regular term,which increases the classification decision ability of the algorithm.Compared with other algorithms in different data sets,the experimental results show that the algorithm is effective.Aiming at the problem of low accuracy of hyperspectral image classification caused by information redundancy and high similarity between samples,joint sparse representation hyperspectral image classification algorithm based on fused albedo restoration is proposed.On the basis of restoring albedo,the algorithm carries out joint sparse representation for the image with.The correlation coefficient is added to calculate the correlation between pixels,in order to determine whether pixels can be used for joint sparse representation.The results of experiments on different data sets show that the effectiveness and excellent classification performance of the algorithm.