Research On Remote Sensing Image Reconstruction Method Based On Compressive Sensing

The development of remote sensing image exhibits high-resolution, hyper-spectral and multi-temporal trend, but the traditional Shannon-Nyquist sampling theorem is used in the existing remote sensing imaging system, which is difficult to deal with the hardware acquisition, transmission and storage process of massive data. Compressed sensing reconstruction of remote sensing image is the use of the reconstruction algorithm, which allows the computer to deal with sparse representation of the image and the reconstruction process has double value of theory and practice. In this paper, the construction process of sparse representation model of remote sensing image is introduced, and the two kinds of optimization algorithm, generalized iterative shrinkage algorithm (GISA) and the augmented Lagrange method (ALM) optimization reconfiguration process, provides a new idea for remote sensing image reconstruction method.The research contents of this paper include the following aspects:(1)An adaptive improved joint sparse representation model is constructed. In order to make the remote sensing image sufficiently sparse, the two kinds of basic prior knowledge based on remote sensing image, namely, the local smoothness of 2D spatial domain and the non local self similarity in 3D transform domain, the regularization method is used to control the proportion of the information content of 2D and 3D separately by using 2 parameters, which can weight the overall sparsity of remote sensing images, and preserve the integrity information and structure of remote sensing image. In order to improve the quality of reconstruction, the block compression method is introduced before pretreatment. The remote sensing image is divided into a number of non overlapping image blocks, and then each image block is compressed sensing sampling and measured, which reduces the amount of data stored in the sensor, and also reduces the amount of computation, this will be able to deal with commonly used image resolution, and solves the previous deal directly with the whole image compressed sensing sampling and cannot handle resolution slightly larger images.(2)The improved joint sparse representation model is based on the remote sensing image, uses two different optimization methods to optimize the solution, which achieves a remote sensing image compression and reconstruction algorithm. First of all, a simple and efficient generalized iterative shrinkage algorithm (GISA) is used to optimize the joint sparse representation model, which realizes remote sensing image reconstruction algorithm. By adjusting the optimal p value to a certain extent, which reduces the redundancy between the data and greatly reduces the number of observations required to reconstruct the signal. In the calculation. GISA algorithm is more simple and efficient, it reduces the computational complexity, can better suppress the noise and ringing effect, protect edge details of the image. Secondly, to solve the improved joint sparse representation model and realize remote sensing image reconstruction algorithm, which combined total variation (TV) method with augmented Lagrange method. TV method greatly reduces the number of iterations and can achieve a good reconstruction effect after initialization, so as to shorten the reconstruction time. Using the augmented Lagrange method to optimize algorithms can effectively reduce the computational complexity, speed up the algorithm and can better preserve the image edge and details.(3)Simulation results verify the effectiveness of the two algorithms. At present, the most representative compressive sensing reconstruction algorithms are compared in the experiment, such as total variation (TV) method, multi hypothesis (MH) method, and cooperative sparse (CoS) method. The results of the experiment are compared at follow aspects:subjective image quality, objective PSNR value, the number of iterations. In order to solve the problem of remote sensing image data acquisition and transmission, a new method is provided.