基于匹配追踪算法的时频滤波去噪方法

Time-frequency filtering de-noise method based on matching pursuit algorithm

(中国石油大学（北京）油气资源与探测国家重点实验室，北京102249)

State Key Laboratory of Petroleum Resource and Prospecting, China University of Petroleum, Beijing 102249, China

地震资料信噪比是影响地震资料质量的关键因素之一，目前的去噪方法大多难以保证在去噪的同时不损伤有效波。为此，提出了基于匹配追踪算法的时频滤波方法，该方法采用“减去”去噪方式，可以有效地去除噪声，且不损伤有效波。匹配追踪算法的基本原理是，将任意信号分解为波形的线性延展，而这些波形是从函数的冗余代码中选出，可以最佳匹配信号的结构。用选出的波形中代表有效信号的部分对信号进行重构，可以达到无损去噪的目的。分别利用匹配追踪时频滤波法和F-X滤波法对含有随机噪声的仿真信号进行了去噪处理，结果表明，匹配追踪时频滤波法可以较好地去除仿真信号中的随机噪声，降低对高频信号的损伤。

S/N ratio is one of the key factors affecting the quality of seismic data. The existing de-noise methods are difficult in removing noise without damaging effective signals. Therefore, the time-frequency filtering de-noise method based on matching pursuit algorithm was proposed. Adopting the “subtracting way”, this method can effectively remove the noise without damaging the effective signals. Matching pursuit algorithm can decompose any signal into a linear expansion of waveforms selected from a redundant dictionary of functions to best match the signal structures. The waveform representing the effective signals is used to reconstruct the signals to remove noise without damaging other signals. Matching pursuit time-frequency filtering method and F-X filtering method were separately used to attenuate the random noise in artificial signals, showing that the former can effectively remove the random noise in artificial signals and decrease the damage of high-frequency signals.