中文摘要：

多方向矢量中值滤波在压制噪声的同时能够有效地保护多分量地震信号的矢量特性.将多方向矢量中值滤波扩展 应用于多分量地震数据去噪处理,把多分量地震数据当作一个矢量波场进行处理,每个时间采样点均为一个矢量.基于地震数据沿同相轴变化缓慢的特点,在矢量空间中,以当前滤波点为中心沿不同试验方向截取相邻道矢量信号片段,根据这些矢量信号片段之间的相关性自适应地选取最佳矢量中值滤波窗.扩展后的方法在压制噪声的同时能够有效保护多分量地震数据各分量之间的相对振幅信息.理论模型和实际数据测试结果表明,多方向矢量中值滤波应用于多分量地震数据 去噪处理是有效的.

英文摘要：

The multi-directional vector median filtering （MD-VMF） can attenuate random noise of multi-component seismic data,and effectively protect the vector characteristics of multi-component seismic wavefields simultaneously. We deployed a vector-median filtering （VMF） ,along lines with different dips to relax the flattening requirement,to suppress noise contained in seismic data represented by multi-dimensional, multi-component vector fields. Although the VMF can be applied to each component of a vector field individually, the MD-VMF is applied to al l components simultaneously. The MD-VMF treats multi-component seismic data as a vector wavefield rather than a set of scalar fields and every time samples is a vector. Ac-cording to the characteristics of seismic signals, we implement the MD-VMF along local events. Then, the optimal local events are selected according to the cross-correlation between the vector signals？ which are windowed from neighboring seismic traces with a sliding time window along each trial direction The MD-VMF can attenuate noise of multi-component seismic data while preserving the relative amplitude information effectively than the single-channel filtering. We demonstrated the effectiveness of the MD-VMF on simulated data and real multi-component seismic data.