小波包节点域和空间域倾角扫描高阶相关去噪技术
|
摘要
小波包去噪就是根据信号和噪声在不同节点、不同层上的差异,在节点域利用其他数学方法(如相关、阈值化处理等)对含噪信号的系数进行缩幅、置零、相关等处理,最后重构原信号。空间域相关是指对经节点域高阶相关后的分量剖面的相邻道进行相关。由于剖面上相邻道的信号具有较强的相关性,而随机噪声的相关性较弱,可利用该特征增强信号,抑制或消除噪声。基于小波包方法、高阶相关方法的特点以及小波包方法在高频段分辨率高的优势,本文提出了小波包节点域和空间域倾角扫描高阶(3阶)相关去噪技术。首先对原始地震记录剖面进行4层小波包分解,得到16个分量剖面,并对这些剖面的每道在节点域对相邻频段的小波包系数进行高阶相关;然后在空间域对同一分量剖面相邻道的小波包系数进行倾角扫描高阶相关;在完成所有分量剖面的空间域相关后,对这些分量剖面进行重构,最终得到去噪后的地震记录剖面。数值实验和实际资料处理结果表明,本文方法的去噪效果明显优于小波尺度域和空间域高阶相关去噪方法,也明显优于常规小波包节点域2阶相关去噪方法。经本文方法去噪后,高斯白噪声得到了明显的压制或消除。
Base on the difference in different nodes and different layers between signal and noise,Wavelet packet denoise method utilizes mathematical algorithms (such as correlation,threshold processing and so on) to process (such as shortening amplitude,nulling and correlating and so on) the coefficients of the noisy signals,and at last the original signals are reconstructed. The correlation in space domain is referred to as the correlation for the nearby traces of the component section after high order correlation in node field,as the signals in the nearby traces on seismic section have strong correlativity and the correlativity of the random noise is weaker,the correlativity difference characteristics between signal and noise can be used to enhance signal and suppress or even remove the noise. Base on the characteristics of wavelet packet method and high order correlation method,as well as the high resolution advantage of the wavelet packet method in high frequency band,the dip scanning high order (third order) correlation denoise technique in wavelet packet node field and space domain was introduced in this paper,at first the original seismic sections were decomposed in 4 layer wavelet packet,16 component sections were obtained,for each channel of the section high order correlation was conducted for nearby frequency band wavelet packet in node field,anf then the dip scanning high order correlation were conducted for wavelet packet coefficients of the nearby traces on the component section in space domain,secondly after the correlation for all component sections was carried out,the component sections were reconstructed,at last denoised seismic sections were obtained.The numerical experiment and field data processing results show that the denoised results of the technique is much better than the results by wavelet scale domain method and high order correlation denoise method,also much better than wavelet packet node field second order correlation denoise method,after applying the technique to seismic data,the Gaussian white noise was obvoiusly suppressed and removed.
Base on the difference in different nodes and different layers between signal and noise,Wavelet packet denoise method utilizes mathematical algorithms (such as correlation,threshold processing and so on) to process (such as shortening amplitude,nulling and correlating and so on) the coefficients of the noisy signals,and at last the original signals are reconstructed. The correlation in space domain is referred to as the correlation for the nearby traces of the component section after high order correlation in node field,as the signals in the nearby traces on seismic section have strong correlativity and the correlativity of the random noise is weaker,the correlativity difference characteristics between signal and noise can be used to enhance signal and suppress or even remove the noise. Base on the characteristics of wavelet packet method and high order correlation method,as well as the high resolution advantage of the wavelet packet method in high frequency band,the dip scanning high order (third order) correlation denoise technique in wavelet packet node field and space domain was introduced in this paper,at first the original seismic sections were decomposed in 4 layer wavelet packet,16 component sections were obtained,for each channel of the section high order correlation was conducted for nearby frequency band wavelet packet in node field,anf then the dip scanning high order correlation were conducted for wavelet packet coefficients of the nearby traces on the component section in space domain,secondly after the correlation for all component sections was carried out,the component sections were reconstructed,at last denoised seismic sections were obtained.The numerical experiment and field data processing results show that the denoised results of the technique is much better than the results by wavelet scale domain method and high order correlation denoise method,also much better than wavelet packet node field second order correlation denoise method,after applying the technique to seismic data,the Gaussian white noise was obvoiusly suppressed and removed.
引文
[1]樊计昌,李松林,刘明军.利用小波包变换提取地震波高频信息.石油地球物理勘探,2006,41(2):144~149
[2]Donoho D L.De-noising by soft thresholding.IEEETramon IT,1995,41(3):613~627
[3]Xu Y,Weaver J B,Healy D Met al.Wavelet trans-formdomain filters:a spatially selective noise filtra-tiontechnique.IEEE Trans Image Processing,1994,3(6):747~758
[4]Mallat S and Hwang W L.Singularity detection and processing with wavelets.IEEE Trans Inform Theo-ry,1992,38(2):617~643
[5]Witkin A.Scale-space filtering.Bundy A.Proc8th Int Joint Conf Art Intell.Karlsruhe,West Germany:William Kauf mann,1983,1019~1022
[6]白英奎,申铉国,孟宪江.空域相关滤噪法的改进及用于红外光谱信号处理.激光与红外,2004,34(1):37~39
[7]潘泉,张磊,孟晋丽等.小波滤波方法及应用.北京:清华大学出版社,2005,63~72
[8]Pan Q and Zhang L.Two de-noising methods by wavelets transform.IEEE Trans Signal Processing,1999,47(12):3401~3406
[9]赵瑞珍.基于小波系数区域相关性的阈值滤波算法.西安电子科技大学学报,2001,28(3):324~327
[10]Zhang L and Paul B.De-noising by spatial correlation thresholding.IEEE Transactions On Circuits and System for Video Technology,2003,13(6):535~538
[11]段晨东,姜洪开,何正嘉.一种基于信号相关性检测的自适应小波变换及应用.西安交通大学学报,2004,38(7):674~677
[12]满蔚仕,高静怀.垂直地震剖面资料中的随机噪声衰减.西安交通大学学报,2006,40(12):1419~1423
[13]章珂,刘贵忠,邹大文等.二进小波变换方法的地震信号分时分频去噪处理.地球物理学报,1996,39(2):265~271
[14]强琳,刘贵忠.利用小波变换多尺度传播特性实现地震信号去噪.西安交通大学学报,1998,32(1):29~32
[15]付燕.基于小波变换的地震信号分时分频相关去噪.煤田地质与勘探,2002,30(6):52~54
[16]王振国,汪恩华.小波包相关阈值去噪.石油物探,2002,41(4):400~404
[17]张三宗,徐义贤.地震记录小波域高阶相关叠加技术.地球物理学报,2006,49(2):554~560
[18]徐长发,李国宽.实用小波方法.武汉:华中科技大学出版社,2001,99~100
[19]樊计昌,刘明军,王夫运,海燕.浅析小波最大分解层.科技导报,2008,26(10):40~42
[20]樊计昌,王夫运,刘明军,海燕.小波尺度域和空间域倾角扫描高阶相关去噪技术.石油地球物理勘探,2008,43(5):514~517
[2]Donoho D L.De-noising by soft thresholding.IEEETramon IT,1995,41(3):613~627
[3]Xu Y,Weaver J B,Healy D Met al.Wavelet trans-formdomain filters:a spatially selective noise filtra-tiontechnique.IEEE Trans Image Processing,1994,3(6):747~758
[4]Mallat S and Hwang W L.Singularity detection and processing with wavelets.IEEE Trans Inform Theo-ry,1992,38(2):617~643
[5]Witkin A.Scale-space filtering.Bundy A.Proc8th Int Joint Conf Art Intell.Karlsruhe,West Germany:William Kauf mann,1983,1019~1022
[6]白英奎,申铉国,孟宪江.空域相关滤噪法的改进及用于红外光谱信号处理.激光与红外,2004,34(1):37~39
[7]潘泉,张磊,孟晋丽等.小波滤波方法及应用.北京:清华大学出版社,2005,63~72
[8]Pan Q and Zhang L.Two de-noising methods by wavelets transform.IEEE Trans Signal Processing,1999,47(12):3401~3406
[9]赵瑞珍.基于小波系数区域相关性的阈值滤波算法.西安电子科技大学学报,2001,28(3):324~327
[10]Zhang L and Paul B.De-noising by spatial correlation thresholding.IEEE Transactions On Circuits and System for Video Technology,2003,13(6):535~538
[11]段晨东,姜洪开,何正嘉.一种基于信号相关性检测的自适应小波变换及应用.西安交通大学学报,2004,38(7):674~677
[12]满蔚仕,高静怀.垂直地震剖面资料中的随机噪声衰减.西安交通大学学报,2006,40(12):1419~1423
[13]章珂,刘贵忠,邹大文等.二进小波变换方法的地震信号分时分频去噪处理.地球物理学报,1996,39(2):265~271
[14]强琳,刘贵忠.利用小波变换多尺度传播特性实现地震信号去噪.西安交通大学学报,1998,32(1):29~32
[15]付燕.基于小波变换的地震信号分时分频相关去噪.煤田地质与勘探,2002,30(6):52~54
[16]王振国,汪恩华.小波包相关阈值去噪.石油物探,2002,41(4):400~404
[17]张三宗,徐义贤.地震记录小波域高阶相关叠加技术.地球物理学报,2006,49(2):554~560
[18]徐长发,李国宽.实用小波方法.武汉:华中科技大学出版社,2001,99~100
[19]樊计昌,刘明军,王夫运,海燕.浅析小波最大分解层.科技导报,2008,26(10):40~42
[20]樊计昌,王夫运,刘明军,海燕.小波尺度域和空间域倾角扫描高阶相关去噪技术.石油地球物理勘探,2008,43(5):514~517
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心