Compressed Sensing of 3D Marine Environment Monitoring Data Based on Spatiotemporal Correlation

Abstract

In the compressed sensing process of three-dimensional marine environmental monitoring data(3D-MMD), the traditional one-dimensional preprocessing methods and direct three-dimensional methods exhibit insufficient performance. In this study, we propose a novel compressed sensing observation and reconstruction method for 3D-MMD based on spatiotemporal correlation. Firstly, by analyzing the characteristics of two-dimensional expansion matrix group of 3D-MMD, we found that the expanded data in longitude dimension is more relevant. Then, according to the circular distribution of global longitude, the two-dimensional matrix sets are overlapped and grouped. The first and last matrices in the group are designated as the key matrix, and the difference matrix of adjacent matrices is calculated. Finally, the sampling rate of the key matrix and the difference matrix is allocated reasonably to achieve efficient observation and reconstruction. Theoretical analysis and simulation results showed that the proposed algorithm can greatly reduce the number of observations and relieve the pressure of system acquisition and storage while ensuring the reconstruction performance.

Keywords

• Compressed sensing
• three-dimensional data
• measurement in groups
• multiple measurement vector
• spatio-temporal correlation
• marine monitoring data