【摘要】 地震资料中随机噪声的压制一直是人们关注的热点。传统方法难以平衡噪声的去除与有效信号的保护，且执行效率低。为此，提出了基于双通道卷积神经网络的随机噪声压制方法。首先，该网络是一个双通道网络，即由两个结构不同的子网络组成，目的是在压制噪声过程中提取到互补有效信息；其次，在下通道子网络中引入空洞卷积增大感受野，充分捕捉到地震资料中的邻域信息，从而更充分地保留细节信息；最后，借鉴残差学习的思想并使用Swish激活函数，提高了网络的降噪性能。模型和实际资料的实验结果表明，所提方法在有效地压制随机噪声的同时能够保留更丰富的纹理细节信息。更多还原

【Abstract】 Random noise suppression has always been a focus in seismic data processing,and the traditional methods can hardly balance the removal of noise and the preservation of effective signals.Therefore,a random noise suppression method based on a dual-channel convolutional neural network(SDCCNN)is proposed in this paper.SDC-CNN is composed of two different sub-networks aiming to extract complementary information in the process of noise suppression.In the lower channel of the network,dilated convolution is introduced to expand the receptive field,which can fully capture the neighborhood information in seismic data and better preserve the details.Moreover,the idea of residual learning and the Swish activation function are adopted to improve the denoising performance of the network.The synthetic and field data experiments all demonstrate that the proposed method can effectively suppress random noise while preserving more texture details.更多还原