摘要
传统的二维随机噪声压制方法应用于三维地震数据的随机噪声压制时,去噪效果往往不理想,为此提出基于稀疏冗余表示的压制三维地震数据随机噪声的方法。该方法在贝叶斯框架下,通过正交匹配追踪(OMP)和K-奇异值分解(K-SVD)不断迭代更新三维稀疏矩阵和三维超完备离散余弦变换(DCT)字典,利用三维超完备DCT字典作为三维地震数据的稀疏冗余表示,使三维地震数据中随机噪声得到压制。三维模拟数据和实际地震数据试算表明:与常规f-x反褶积法和K-L变换法相比,该方法既提高了三维地震数据体的信噪比,又有效地保护了地震反射信号,而且水平切片的连续性和平滑性很好,构造复杂区域的分辨率也得到提高。
Conventional methods to suppress random noise work very well for 2Dseismic data,but not for 3Dseismic data.Therefore we present in this paper a new method to remove random noise from3 Dseismic data,which is driven by sparse and redundant representation algorithm.Under Bayesian framework,this method uses 3D overcomplete DCT dictionary to sparsely and redundantly represent seismic data.Using orthogonal matching pursuit(OMP)and K-singular value decomposition(K-SVD)continuously to update 3Dsparse matrix and 3D overcomplete DCT dictionary,random noise of 3Dseismic data can be significantly suppressed.We apply this proposed method to both3 Dtheoretical and real seismic data together with conventional f-x deconvolution method and K-L transform method.Application results show that the proposed method can improve signal to noiseratio and protect seismic signals,and slice continuity and smoothness,and the resolution of complicated structures are also improved.
出处
《石油地球物理勘探》
EI
CSCD
北大核心
2015年第4期600-606,1-2,共7页
Oil Geophysical Prospecting
基金
国家"973"项目(2014CB239201
2013CB228604)
国家油气重大专项(2011ZX05014-001-010HZ)资助
