Research On Multi-source Remote Sensing Image Fusion Model And Algorithm

Multi-source remote sensing image fusion technology refers to the combination of different types of data from multiple sensors to provide more comprehensive information,which helps to improve accuracy in application scenarios.This paper has conducted in-depth research on multi-spectral image fusion and hyperspectral image fusion.Multispectral image fusion technology combines low spatial resolution multispectral images（MSI）and high spatial resolution panchromatic images（PAN）to obtain MSI with high spatial and spectral resolution.Hyperspectral image fusion technology usually fuses a low spatial resolution hyperspectral image（HSI）with a high spatial resolution multispectral image to obtain HSI with high spatial and spectral resolution.This paper proposes a multispectral image fusion model using cartoon-texture similarity,and a HSI-MSI-PAN fusion model based on spatial-spectral total variation regularization.The solving algorithms of the model are respectively proposed,and the main work contents are as follows:This paper proposes a multispectral image fusion model using cartoon-texture similarity.After decomposing of a PAN image,it can be found that its cartoon component contains global structure information,and its texture component contains local detail information.This enables that the fused high-spatial resolution MS image can preserve the global and local spatial details（e.g.,high-order information）well after leveraging the similarities of cartoon and texture components from PAN and MS images.To explore such cartoon-texture similarities,we describe cartoon similarity as gradient sparsity,formulated as a reweighted total variation term.Meanwhile,we use group low rank constraint for texture similarity that is presented as repetitive texture patterns.By incorporating a data fidelity term for preserving the spectral information on the basis that the down-sampled fused MS image is consistent with the MS image,we further formulate the multi-spectral image fusion as an optimization problem,and use the alternating direction multiplier method to divide the objective function into multiple sub-problems to solve the optimization problem.Extensive experiments are conducted on a series of satellite datasets.The qualitative and quantitative results demonstrate that our method outperforms the state-of-the-art pansharpening methods in terms of both visual effect and objective metrics.The source images of the hyperspectral image fusion technology are usually HSI with low spatial resolution and MSI with high spatial resolution.HSI with high spatial and spectral resolution can be obtained through fusion.Considering that the information contained in these three types of data is complementary,this paper proposes a fusion model of HSI,MSI and PAN based on spatial-spectral total variation regularization.The hyperspectral fusion method proposed in this paper is a model-based method.We first decompose the fusion target into two parts:a subspace matrix and a coefficient matrix based on matrix decomposition.The subspace matrix is obtained from the low-resolution HSI using vertex component analysis,so that the solution of the high-dimensional data is transformed into the solution of the low-dimensional coefficient matrix,and the7)₂ norm is used as the constraint of the residual error between the down-sampled version of HR-HSI and HR-PAN,LR-MSI and LR-HSI,correspondingly establish three data fitting items to retain spectral information and spatial information.In addition,based on the smoothness of the three-dimensional coefficient matrix,this paper uses spatial-spectral total variation regularization to establish the regularization term of the coefficient matrix,thereby establishing the target energy function of the coefficient matrix.Finally,this paper proposes a solution based on the alternate direction multiplier method algorithm to solve the target energy function,and conduct a comprehensive experiment on a series of simulated satellite data and real satellite data sets to verify the effectiveness of proposed method.Since the results of remote sensing image fusion will eventually be applied to various fields,this paper designs experiments to simulate application scenarios to further verify the effectiveness of our fusion method.Specifically,it includes an experiment of vegetation coverage change analysis of multispectral fusion data and an experiment of vegetation index analysis of hyperspectral fusion data.The experimental results verify the feasibility of the two fusion methods proposed in this paper in practical applications.