Sampling And Reconstruction Of Seawater Temperature Field Based On Compressed Sensing

As researchers have become more interested in oceanographic research,the seawater temperature as a key to understanding and studying and the physical and chemical processes of the Earth,is of crucial importance in its observation and reconstruction.In this paper,the ocean sampling method is studied,and a sampling points optimization scheme based on the seawater temperature variability and gradient is proposed,in order to optimize the allocation of sampling resources.In the field of reconstruction study,the K-SVD algorithm was used to train the sparse matrix.The sparse matrix which was adapted to the seawater temperature field data characteristics is applied to the reconstruction of seawater temperature field.Firstly,the basic contents of this paper is introduced.The first focuses on the analysis of the variability of seawater temperature field and the analysis of gradient.The mathematical expression is also given,and the application of the method is described.The second content introduces sparse representation,measurement matrix selection and signal reconstruction.Secondly,based on the analysis method of seawater temperature field,an optimal sampling points configuration method based on seawater temperature field variability and gradient is proposed.The method first analyzes the seawater temperature field variability among sub-sampling areas and designs a calculation method to calculate the sampling rate of each sub-sampling area in order to balance the sampling resources simultaneously.Sub-sampling areas with “richer information” are allocated more sampling resources.Then,cluster analysis is performed on the seawater temperature gradient field in each sub-sampling area,in order to distinguish the salient areas and the steady areas of temperature gradient field.The simulation results show that under the same conditions of sampling resources,the o optimal sampling points configuration method can improve the reconstruction accuracy of seawater temperature field.Furthermore,based on the K-SVD algorithm,a sparse matrix suitable for seawater temperature field is trained.Based on the sampling characteristics of the seawater temperature field,the sampling position coding matrix is selected as the measurement matrix.The ASMP algorithm is selected to be the reconstruction algorithm.Experimental results show that sparse matrices obtained by K-SVD algorithm improve compared with traditional sparse matrices in reconstruction accuracy of temperature field.Finally,the experiment of sampling method and reconstruction method is completed.The seawater temperature field data of the whole year are simulated through several comparative experiments to verify the effectiveness of the sampling method and reconstruction method proposed in this paper.The experimental results show that the proposed sampling configuration optimization method can well sample the regions with “richer information” of the seawater temperature field and ensure the balance of sampling resources.In the aspect of data reconstruction,the sparse matrix of seawater temperature field data obtained in this paper has greatly improved the reconstruction effect of seawater temperature field data.