Wavelet frame based seismic attributes extraction using a filtering scheme

Extraction of instantaneous attributes is important for seismic data processing and interpretation. However, the instantaneous attributes extracted by the conventional Hubert transform method are sensitive to noise that inevitably lies in field seismic data. We propose a robust approach to extract instantaneous attributes in wavelet domain. In the proposed approach, we apply a superfamily of analytic wavelets with some desirable properties-the generalized Morse wavelets-in the proposed approach. Based on the proposed discretization, the wavelet family can constitute a tight frame. For signal in noise, we implement a filtering scheme to determine the distribution of the effective signal in the transformed domain before calculating the instantaneous attributes. In this filtering scheme, a percentage thresholding strategy is manipulated. Compared with the conventional method based on Hubert transform, the synthetic trace and real data examples show higher precision and anti-noise performance of the proposed approach, even for signals contaminated by strong noise.