An efficient implementation of eigenstructure-based coherence algorithm via dimensionality reduction

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