地震数据去噪的研究与分析

Research and Analysis of Seismic Data Denoising

小波分解利用信号和噪声在时频域内小波系数差异,重构信号从而去除噪声;SVD算法利用信号和噪声奇异值的差异,选取合适的空间维数重构信号从而去除噪声。通过模型验证对比了两种算法的去噪效果和优缺点,发现小波算法对水平同相轴和倾斜同相轴具有相同的去噪效果,但运算时间较长;SVD算法对水平同相轴去噪效果很好,对于倾斜同相轴几乎不能去噪,但运算时间较短。

The wavelet decomposition uses the difference of signal and noise in the time-frequency domain wavelet coeficient,then reconstructs the signal wiping off noise; SVD algorithm uses the difference of the signal and noise singular value, selecting the appropriate dimensions to reconstruct the signal wiping off noise. Through the comparison of two kinds of model validation algorithm for denoising effect, we know that wavelet algorithm for level phase axis and dip phase axis has the same denoising effect, but the operation time is long. The SVD algorithm for level event denoising effect is very good and the operation time is short,but it does not fit for dip event axis denoising.