摘要
This paper discusses the principle and procedures of the second-generation wavelet transform and its application to the denoising of seismic data. Based on lifting steps, it is a flexible wavelet construction method using linear and nonlinear spatial prediction and operators to implement the wavelet transform and to make it reversible. The lifting scheme transform -includes three steps： split, predict, and update. Deslauriers-Dubuc （4, 2） wavelet transforms are used to process both synthetic and real data in our second-generation wavelet transform. The processing results show that random noise is effectively suppressed and the signal to noise ratio improves remarkably. The lifting wavelet transform is an efficient algorithm.
本文讨论了第二代小波变换的基本原理和变换过程,并将第二代小波变换引入到地震资料去噪处理中,基于提升法的小波变换是一种柔性的小波构造方法,它使用线性、非线性或空间变化的预测和更新算子,并能确保变换的可逆性。通过对模拟数据和实际资料的处理,证明了的它对地震信号去噪具有很好的效果。离散信号的小波去噪可分为三步:小波分解,系数缩减(切除噪声部分),信号重建。目前常用的小波去噪的方法有硬阈值法和软阈值法,这里采用软阈值法去噪。本文的提升变换采用的是Deslauriers-Dubuc(4,2)小波,基于以上变换方法,分别对含噪的模拟数据及实际地震数据进行3级可逆提升变换,对每一级上的细节信号按上述的软域值法进行处理,削减小波系数中的噪声部分,从而实现了信号去噪,结果证明去除随机噪声的效果是令人满意的。
