f-x-y域最小平方反演预测滤波

在对f-x域预测滤波方法深入研究的基础上,本文提出了f-x-y域最小平方反演预测滤波方法。该去噪方法的基本思想是假设信号中也存在不可预测部分,利用信号中的不可预测部分和随机噪音的联合分布建立目标函数,进而实现频率空间的预测滤波。理论模型试算和实际资料处理表明,该方法较f-x-y域预测滤波能更好的提高地震资料的信噪比,保持地震同相轴的连续性。

Prediction Filtering by Least-square Inversion in f-x-y Domain

The prediction filtering method by least-square inversion in f-x-y domain has been put forward on the basis of detailed analysis for noise elimination using prediction filtering in f-x domain. The method makes the assumption that the seismic signal comprises predictable proportion and unpredictable proportion. The transcendental information about predicting operator is introduced in the function. We establish the function using the joint distribution of the unpredictable proportion in signal and the noise. Thus the prediction filtering is carried out in f-x-y domain. With this technique, we can not only increase the signal/noise ratio but also maintain the continuity of effective events. The validity of this technique has been proved by applying it to the theoretical modeling and real seismic data.