Efficient Intelligent Denoising Method of Transformation Domain Hybrid Technique for Geophysical Data

As an efficient nondestructive exploration technique, geophysical technologies have been widely used in engineering construction, resources exploration and medical imaging. Geophysical data is observed in long period time domain with transmitting high frequency wave to underground to extract the properties and shapes from the interpretation of signal anomalous. In regard to more complicate and unknown environment and record the signal in a wide bandwidth, geophysical signals unavoidably are mixed with many types of multiple level energy noise. In order to obtain the objective signals for accuracy data interpretation, it is indispensable to denoise and evaluate the useful geophysical data from the noisy data. By selecting different wavelets and truncating the wavelets coefficients in different scales, the peak signal to noise ratio (PSNR) of objective weak signal is improved. Meanwhile, with achieving the directional suitability of curvelet transformation, a efficient algorithm of curvelet transformation can remove the clutter and noise of noisy geophysical data. However, since the multi-scale analysis with non-intelligent, the denoising scheme of transformation domain hybrid technique is proposed by integrating the window high-order correlation statistical. The intelligent scheme of this new method is statistical objective signal in wavelet domain, the noise coefficients in remaining scales can be detected using curvelet transformation because of mode mixing with the signal’s components. The validation and demonstration of this new intelligent denoising scheme are tested in some synthetic and field data.