The regular reconstruction of seismic data is one of the most important and fundamental aspects of seismic data pro-cessing. In this paper we first analyzed the characteristics of the seismic data in the frequency-wave number domain. We found that an insufficient regular or irregular sampling of the seismic data may result in a dispersion of the spectral energy in the wave number domain,which illustrates the importance of regularized reconstruction.We classified the existing regularized reconstruction methods according to the types of the seismic data, and then analyzed the merits, drawbacks and application conditions of the methods based on their realization principle. We subsequently proposed a selection strategy of the regular-ized reconstruction methods under different conditions. Finally,we carried out a model testing of the reconstruction method using the proposed selection strategy. Our results demonstrate that the anti-aliasing reconstruction is the key of seismic data reconstruction. The regular under-sampled data can be reconstructed by methods based on the prediction-error filtering theo-ry,various mathematical transforms,and the combination of various mathematical transforms with sparse inversion. The ir-regular under-sampled data can be reconstructed by methods based on various mathematical transforms,the rank-reduced the-ory,the prediction-error filtering theory and their combinations with the sparse inversion or compressed sensing theory. A comparison among the different methods validates the robustness and feasibility of the proposed selection strategy of the regu-larized reconstruction methods.%地震数据规则化重构是地震资料处理十分重要的基础性工作.首先分析地震数据在频率波数域的表现特征,研究发现规则或非规则采样不足均会造成频谱能量在波数域发散,从频谱分析角度阐述规则化重构的重要性;然后根据地震数据的类型对现有的各种规则化重构方法进行分类,依据其实现原理分析各种方法的优缺点及适用条件,总结不同条件下规则化重构方法的选择策略;最后根据总结策略对重构方法进行模型测试.结果表明,地震数据重构的重点为抗假频重建,当规则采样不足时可以采用基于预测误差滤波理论、各种数学变换及其与稀疏反演组合的重构方法,当非规则采样不足时采用基于各种数学变换、矩阵降秩理论和预测滤波理论以及三者分别与稀疏反演或压缩感知理论相结合的方法,不同方法对比效果进一步验证了总结的规则化重构选择策略的正确性和可行性.
展开▼
