Processing Step-Based Adaptive Seismic Denoising Method of Transformation Domain Hybrid Technique

To deal with remaining noise in a single domain, a transformation domain hybrid technique that employs a processing step-based adaptive denoising scheme is proposed to identify and remove noise while preserving the target seismic signals for practical engineering applications. The adaptive algorithm uses a higher order statistic to stack the wavelet-based coefficients of the seismic signals in the preprocessing step. It then whitens the noise and recovers the processed signals in the wavelet coefficient environment without the need to choose an arbitrary threshold. Automatic thresholding is employed to identify the high-correlation signal and stack this signal using the higher order statistic. The complex block thresholding filter captures the features of the remaining mode elements by iteratively using the wavelet and curvelet basis functions in the curvelet domain in the postprocessing step, with the goal of removing the noise with disordered correlations in the rotational directions. Seismic records of karst topography and hidden sinkhole detections, which are used for the evaluation of foundation stability in constructional engineering, are analyzed to determine the signal-to-noise ratio and fidelity of this scheme. A comparison of the performances between different denoising methods in the single domain is presented using synthetic seismic waveforms that are contaminated with different types of noise to validate the efficiency of this hybrid denoising scheme. Processing field data demonstrates the effectiveness of this hybrid denoising scheme compared with electrical resistivity imaging.