基于联合稀疏表示的共偏移距道集随机噪声压制方法

Random noise attenuation based on joint sparse representation in common offset gathers

受多解性和单道信号处理方法制约,传统基于稀疏表示的一维随机噪声压制方法面临着单道数据处理方法没有考虑有效信号的空间相关性,去噪的同时会损害有效波,以及稀疏表示算法具有多解性,相邻地震道处理结果差异大,难以适应信号空间变化的问题。叠前共偏移距道集中各波形均为水平同相轴,具有相同的双程旅行时间,各道信号具有相同的支撑。在该道集中利用联合稀疏表示进行随机噪声压制处理,能够兼顾信号的道间相干性和空间变化,降低算法的多解性,参与计算的各道在同一条件下获得最优稀疏表示,因此处理结果具有较好的一致性。数值模拟和实际资料试算结果表明,该方法不仅可以实现随机噪声的压制,而且可以很好地保持有效信号,具有良好的应用效果。

Under the restriction of multiplicity and single-channel signal processing methods,the traditional onedimensional random noise attenuation method based on sparse representation faces two problems:Firstly,the single-channel signal processing method does not consider the spatial correlation of the effective signals,thus the de-noise performance degrades significantly and may damage effective waves.Secondly,the sparse representation algorithm suffers from multiplicity,and the processing results of adjacent seismic traces are quite different,which makes it difficult to adapt to the spatial variation.The waveforms in the pre-stack common offset gathers are characterized by horizontally events,and have the same two-way travel time.Therefore,each trace signal in a gather shares the universal support.In this gather,joint sparse representation was used to suppress random noise,which can balance the correlation between channels and spatial variations of signals so as to reduce the multi-solution of the algorithm.The optimal sparse representation was obtained under the same conditions for each channel participated in the calculation,so the processing result had good consistency.The results of numerical simulation and practical data trial show that the proposed method can suppress random noise and maintain the effective signal,and has a good application effect.