摘要
野外地震数据所处地质环境复杂,受干扰信息影响严重,信噪比低、分辨率低等,不利于地质数据解释及后期地质勘探,单一的降噪算法很难有效去除各种干扰信号。本文提出一种改进的Mallat算法对地震数据进行降噪处理,利用Mallat算法将地震信号分解到不同尺度频带空间上,再选取合理的阈值对不同频带空间上的地震数据进行降噪处理,然后利用ICA分离方法在不同尺度频带空间上提取有效信息,通过Mallat算法将有效信号重构回原始地震信息。用构建的Mallat算法对实际金属地震数据进行降噪处理,结果表明,改进的Mallat算法降噪效果明显,能较好地去除地震数据的各类干扰,降噪后的地震剖面纹理清晰、分辨率高。
With the field seismic data from complex geological environment disturbed by the influence of interferential information severely,and with low signal to noise ratio and low resolution,there are many disadvantages of geological data interpretation. A single algorithm of noise reduction is difficult to remove a great variety of noise effectively. So this paper proposes a new noise reduction named improved Mallat algorithm,an effective combination of Mallat and ICA. In this method,the seismic signals are decomposed into different scales on the frequency bands by Mallat. Seismic data on different frequency bands will process noise reduction through threshold method. Then we can regain the effective data on different frequency bands through ICA. Eventually the effective original geological data can be restructured by Mallat. The proposed improved Mallat algorithm can deal with the actual seismic data. The results show that improved Mallat algorithm can effectively eliminate most of noise. The texture of the seismic profile and the hyperbolic of the seismic profile morphology is clear by the noise reduction of the improved Mallat.
作者
程鲁
秦飞龙
张津
王茜
周详全
柳炳利
郭科
CHENG Lu;QIN Fei-long;ZHANG Jin;WANG Xi;ZHOU Xiang-quan;LIU Bing-li;GUO Ke(Key Laboratory of Mathematical Geology in Sichuan, Chengdu University of Technology, Chengdu 610059 , China;Department of Information and Computing Science, Chengdu Technological University, Chengdu 610031, China;Chengdu Hanlin School, Chengdu 611534, China)
出处
《桂林理工大学学报》
CAS
北大核心
2017年第4期602-607,共6页
Journal of Guilin University of Technology
基金
中国地质调查局项目(1212010916040)
四川省科技厅项目(18ZDYF)
关键词
地震数据
降噪
改进的Mallat算法
ICA算法
seismic data
noise re du c t ion
improved Ma l la t
independect component analysis （IC A ）
