【摘要】 在地震记录中,随机噪声严重影响了有效信号的提取,为此必须进行消噪处理。这里首先使用小波包变换对不同频段的信号进行精细分离,有效信号和噪声经小波包分解后,其小波包系数将表现出不同特性,然后根据这种不同特性进行去噪处理,对小波包分析法处理后的剩余地震信号再进行KL(Karhunen-Loeve)变换,提取相关有效信号,最后对提取的有效信号进行中值滤波处理,进一步去除剩余噪声。经合成地震剖面和实际地震剖面处理实验证明,小波包分析、KL变换和中值滤波联合去噪方法,能有效地消除较强的随机噪声,提高地震剖面信噪比和分辨率。更多还原

【Abstract】 In the seismic records,random noises seriously affect the extraction of the significant signals,so that the denoising processing is necessary.This paper first uses wavelet package transform to separate different frequency signals.And then after significant signals and noises are decomposed by the wavelet package,their wavelet package coefficients represent different characteristic.Then denoising processing is carried out based on these characters,and the KL transform is applied to extract correlative significant signals.Finally,median filter is used to eliminate noises included by residual seismic signals which have been processed by wavelet package analysis method and KL transform.The synthetic and real seismic-section processing tests show that the proposed method in this paper,which combines wavelet package analysis,KL transform and median filter,can efficiently eliminate strong energy random noises and improve seismic section signal-to-noise ratio and seismic resolution.更多还原