中文摘要：

经验模态分解算法（EMD）是一种基于有效波和噪声尺度差异进行波场分离的随机噪声压制方法，但由于实际地震数据波场复杂，导致模态混叠较严重，仅凭该方法进行去噪很难达到理想效果．本文基于EMD算法对信号多尺度的分解特性，结合Hausdorff维数约束条件，提出一种用于地震随机噪声衰减的新方法．首先对地震数据进行EMD自适应分解，得到一系列具有不同尺度的、分形自相似性的固有模态分量（IMF）；在此基础上，基于有效信号和随机噪声的Hausdorff维数差异，识别混有随机噪声的IMF分量，对该分量进行相关的阈值滤波处理，从而实现有效信号和随机噪声的有效分离．文中从仿真信号试验出发，到模型地震数据和实际地震数据的测试处理，同时与传统的EMD处理结果相对比．结果表明，本文方法对地震随机噪声的衰减有更佳的压制效果．

英文摘要：

Empirical mode decomposition （EMD） is a noise suppression algorithm by using wave field separation, which is based on the scale differences between effective signal and noise. However, because the complexity of the real seismic wave field can result in serious aliasing modes, it is not ideal and effective to denoise using this method alone. Based on the multi-scale decomposition characteristics of the EMD algorithm for signal, combining with Hausdorff dimension constraints, we propose a new method for seismic random noise attenuation. Firstly, we apply EMD algorithm adaptive decomposition of seismic data to obtain a series of IMF components with different scales. On this basis, based on the difference of Hausdorff dimension between effective signals and random noise, we identify IMF component mixed with random noise. Then we use the threshold correlation filtering process to separate the valid signal and random noise effectively. This method includes three steps, i.e. simulation signal experiment, the seismic model data processing and real seismic data processing. Compared with traditional EMD method, this new method of seismic random noise attenuation has a better suppression effect.