The Application of OPTSPACE Algorithm and Comparison with LMAFIT Algorithm in Threedimensional Seismic Data Reconstruction via Lowrank Matrix Completion


Chen Xue







File Size:

5.7 MB

View File:

Access Count:



This report is focused on three-dimensional seismic data reconstruction with randomly missing data on a regular grid. Ma (2013) developed a rank-reduction method that transforms three-dimensional seismic data reconstruction problem into low-rank matrix completion (MC) problem with the “texture-patch transformation”, and resolved the MC problem from the perspective of the nuclear-norm minimization. Aiming at achieving a higher-quality reconstruction with small computational complexity in time, this report followed the general framework and the low-rank matrix completion idea in Yang et al. (2011) and Ma (2013), generalized the three-dimensional texture-patch transform, and settled the low-rank matrix completion problem from two other perspectives, the rank-r matrix approximation problem and low-rank factorization problem. Furthermore, this report applied two corresponding matrix completion algorithms, a gradient descent algorithm on the Grassman manifold (OptSpace) and the low-rank matrix fitting algorithm (LMaFit), to resolving the MC problem, and finally compared the performance of the proposed methods by conducting numerical experiments on simulated seismic data and the field data set.

Press the Back button in your browser.

Copyright 2016, Chen Xue: Please note that the reports and theses are copyright to their original authors. Authors have given written permission for their work to be made available here. Readers who download reports from this site should honor the copyright of the original authors and may not copy or distribute the work further without the permission of the author, Chen Xue.

Accessed by: ec2-44-192-79-149.compute-1.amazonaws.com (
Accessed: Friday 09th of June 2023 11:16:16 AM