An Efficient Eigenstructure-Based Coherence Measure Via Dimensionality Reduction

Zhizhou Huo,Xiwu Liu,Xiaokai Wang,Wenchao Chen

DOI: https://doi.org/10.1109/lgrs.2019.2908434

IF: 5.343

2019-01-01

IEEE Geoscience and Remote Sensing Letters

Abstract:The eigenstructure-based coherence measure has been widely used in characterizing faults and other discontinuous structures in seismic quantitative interpretation. However, it still suffers from the high computational cost which is caused by the eigendecomposition of one large covariance matrix. In this work, we focus on the computational cost of the widely-used coherence measure and try to propose one efficient eigenstructure-based coherence measure but without lowering the visual quality. We replace calculating one large covariance matrix’s dominant eigenvalue with calculating two smaller covariance matrixes’ dominant eigenvalues and use these two dominant eigenvalues to construct one efficient eigenstructure-based coherence measure. The computational cost of the proposed efficient eigenstructure-based coherence measure is smaller than the computational cost of the common eigenstructure-based coherence measure but without decreasing visual quality. Finally, we use one 3-D field data example to illustrate the proposed efficient eigenstructure-based coherence measure’s efficiency.