The Research And Design About Marine Route In Gravity Measurement Based On Compressed Sensing

The Earth's gravity model is the basic information for common needs in various fields and disciplines.Compressive sensing theory can avoid inefficient intermediate links,obtain a compressed representation of gravity signal by solving an optimization problem through a little measurement,break through the bottleneck of Shannon sampling theorem,and realize the gravity data reconstruction which is actually sub-nyquist sampling.In 2015,Yang Yapeng's doctoral dissertation applied compression sensing theory to gravity field modeling for the first time.This paper solves the problem of high sampling rate and unscientific measurement mode.Based on EGM2008 model,the sparse representation and reconstruction of gravity data are further studied.By designing the route to achieve the same accuracy at the same time significantly reduce the sampling rate.The main research results are as follows:Firstly,the sparse representation of gravity data based on the Curvelet transform redundant dictionary is carried out to overcome the selective defects that the wavelet bases only have horizontal,vertical and diagonal directions.By choosing different curved-wave coefficients in different scales as different atoms and different dictionaries,the gravity data are sparsely expressed and reconstructed.The PSNR values,sparsity and values are used as the evaluation criteria to determine the optimal number of atoms.At last,the gravity data is sparsely expressed by orthogonal matching algorithm,and the gravity data is reconstructed by using the sparse coefficient obtained.Secondly,the gravity data are reconstructed by using spectral projected gradient(SPG)algorithm.Two typical routes were selected,sampled and measured for one sea area,real gravitational field was simulated based on EGM2008 model,and SPG algorithm was used to reconstruct gravity data.The results of two kinds of route reconstruction were analyzed by sampling rate and peak signal-to-noise ratio.Finally,using genetic algorithm,the sea area is divided into small blocks,and by initializing the population,calculating the fitness,crossing and mutation,and constantly cycling,eventually generating the best route.From the final result,we can see that the reconstruction accuracy(about 40dB)similar to that of Yang Yapeng's paper can be achieved by the method proposed in this paper,but the sampling rate is greatly reduced from about 50% to about 5%.