Deblending using a space-varying median filter

Conclusions

I have proposed a novel non-iterative approach for deblending simultaneous-source data. In order to utilize the strong ability of median filter (MF) to remove spike-like blending noise and to solve the problem of MF in harming useful signal, I propose a space-varying median filter (SVMF). The SVMF can adapt to different regions with varying window length. The proposed SVMF can detect the useful signal and blending noise by signal reliability (SR), defined as the local similarity between the data initially filtered using MF and the original noisy blended data, and then squeeze and stretch the filtering window according to SR. The SVMF can be used in common-receiver gather, common-midpoint gather, and common-offset gather, where the blending noise appears as spike-like noise along the spatial dimension. Synthetic and field data examples show that the proposed deblending approach can efficiently and effectively separate the blending data while preserving the useful signal.