A Radon-domain sparsity enhancement algorithm based on 3D deconvolution

The Radon transform serves as a critical tool for multiple suppression as well as high-precision imaging for seismic primary reflections, while its focusing capability directly affects the outcomes of seismic data processing. Conventional least-squares Radon transforms based on frequency-curvature domain L2-norm constraints tend to exhibit scissor-like trailing artifacts due to the limitation of finite aperture. Sparse Radon transforms have been developed to enhance Radon-domain focusing performance by incorporating sparsity constraints, which nevertheless face the limitation in achieving adequate convergence of energy clusters. To overcome this limitation, we propose a Radon-domain sparsity enhancement algorithm based on 3D deconvolution theory. The proposed algorithm integrates a seismic wavelet and a curvature-direction smoothing function to construct a deconvolution operator, which is then applied to 3D deconvolution of Radon-domain data within an iterative inversion framework using the alternating direction method of multipliers (ADMM). This approach compresses energy in the curvature dimension via the deblurring mechanism grounded in sparse inversion theory, thereby notably improving the concentration and resolution of energy clusters. Synthetic and field data tests demonstrate that, compared with conventional least-squares and time-domain sparse Radon transforms, the proposed deconvolution-based sparsity-enhanced method substantially improves the discriminability of Radon-domain energy clusters, leading to more accurate identification and suppression of multiple reflections.