Separation of simultaneous sources via coherence pass robust Radon operators

We have developed a new iterative method for simultaneous source separation (deblending). The proposed technique adopts the robust sparse Radon transform to define a coherence pass operator, which, used in conjunction with the steepest descent method, guarantees solutions that honor simultaneous source records. We show that a substantial improvement in convergence is attainable when we adopt a robust coherence pass projection, such as the robust sparse Radon transform by ADMM method. The improvement is a consequence of having an iterative deblending algorithm that applies intense denoising to erratic blending noise in its initial iterations. The coherence pass robust Radon operator acts as a data projection operator that conserves coherent signals and annihilates incoherent blending noise right from the start of the iterative process. We compare the algorithm with its non-robust version and show that a coherence pass robust Radon operator can attain high-quality results for both synthetic and real data examples.