3D seismic data reconstruction using an improved projection onto convex sets method based on prior support information

Reconstructing in different partitions independently is one of the effective strategies to implement 3D seismic data reconstruction. However, if this strategy is used, the underlying correlations between two adjacent partitions will be neglected. To solve this problem, this paper introduces curvelet transform and provides an improved projection onto convex sets (POCS) method with weighting operator, based on the continuity between two adjacent time slices, to reconstruct each time slice. For each time slice, we constructed this weighting operator based on the support estimation given by the adjacent reconstructed time slice in curvelet domain. Thereby, we provided a 3D seismic data reconstruction method using prior support information. The experiments for both synthetic data and field data demonstrated that our method is superior to the traditional POCS.