基于Curvelet变换冗余字典的重力数据稀疏表示与重建  

Sparse representation and reconstruction of gravity data based on redundancy dictionary from Curvelet transform

作　　者：牛丽琨 吴美平[1] NIU Likun;WU Meiping(College of Mechatronic Engineering and Automation,National University of Defense Technology,Changsha 410073,China)

机构地区：[1]国防科学技术大学机电工程与自动化学院,长沙410073

出　　处：《物探化探计算技术》2018年第5期631-636,共6页Computing Techniques For Geophysical and Geochemical Exploration

摘　　要：压缩感知理论为低采样建立大面积、高精度的重力场模型提供了理论指导。构造合适的变换基,实现二维重力数据更为稀疏的表示,对于测量矩阵的设计和重构精度的提高具有重大意义。采用Curvelet变换作为核函数构造过完备冗余字典,选取不同尺度不同方向的曲波系数作为原子组成不同的字典,对EGM2008模型分辨率为1′×1′的某海域重力异常数据进行稀疏表示和重建,根据PSNR值,稀疏度和η值作为评估标准确定最优原子个数。最后采用OMP算法对重力数据稀疏表示并重建。The theory of compressed sensing provides theoretical guidance for the establishment of large area and high precision gravity model from low sampling.It's meaningful for the design of the measurement matrix and promotion of reconstruction effect to construct a suitable transform base and realize the more sparse representation of the two-dimensional gravity data.In this paper,the Curvelet transform is used as a kernel function to construct a redundant dictionary.Different types of curvature coefficients in different directions are selected as the dichotomies with different parameters.The gravity anomaly data of a certain gravity in the EGM2008 model are sparse and reconstructed.We use PSNR Value,sparse degree andηas the evaluation criteria to determine the optimal number of atoms.Finally,we use OMP algorithm to represent sparse gravity data and reconstruct it.

关 键 词：压缩感知 CURVELET变换 冗余字典 稀疏表示 

分 类 号：P631.4[天文地球—地质矿产勘探]