应用同态滤波提取子波的改进方法

1991年 30卷 第No. 3期
阅读:91
查看详情
AN IMPROVEMENT ON WAVELET EXTRACTION USING HOMOMORPHIC FILTERING
地质矿产部北京计算中心
Beijing Computer Centre, MGMG
本方法建立在地层反射系数序列是变化的,相邻道子波基本不变的基础上的。我们改进了以往在剖面上随机选取时窗的方式,有目的地在信噪比和分辨率较高和同相轴较一致、连续性较好的目的层附近选取一叠加段。然后再将时窗强行随机化,即对所选的同一叠加段中的各时窗起始时间点,人为加上一个随机数,使各时窗的起始点服从随机分布,相应时窗内所包含的反射系数序列也随之重新分布。这样就满足了反射系数序列为变化的,地震子波保持基本不变的条件。最后将这些时窗的对数谱进行叠加,使子波这个不变部分的对数谱能量增加,而反射系数序列这个变化部分的能量极大地减弱,从而达到提取于波的目的。为了提高所提取子波的精度,本方法还采用了对数谱平均、子波的拾取与整形等处理方法,取得了好的效果。
The technique recommended in the paper is developed on the ground of that the stratigraphic reflection coefficient sequence is subject to change while the wavelets of its neighboring trace should stay the same mostly. The attempt that we made is to give an update on the conventional way of practice, to select window on section arbitrarily. The way we are doing now is, with definite purpose in mind, to pick up a CDP-stacked segment near the target where S/N ratio and resolution are higher; events shown are quite lined up and the continuity is preferable. And then to let the window be compul-sorily randomized. That means to add an arbitrary number to the starling time of each window of the same CDP-stacked segment picked up, so that to enable the starting time to serve the stochastic distribution. And the reflection coefficient sequence involved in the rest corresponding windows will perform the redistribution accordingly. This does have met the requirements which include the reflection coefficient sequence should be changeable but the seismic wavelets stay constant basically. Last, to stack up log spectrum of these windows for increasing the wavelet energy and decreasing the energy of reflection coefficient sequence. Thus,wavelet could be extracted. For ensuring the accuracy of wavelets to be extracted, the alternative processing techniques, such as log spectrum averaging, wavelet pickup and shaping, etc. have been used. And satisfactory results are achieved.
子波提取; 同态滤波; 反射系数序列; 对数谱; 复赛谱;
Wavelet Extraction; Ilomomorphic Filtering; Reflection Coefficient Sequence; Log Spectrum; Cepstrum;