王凌群, 李冰冰, 林君, 謝賓, 王琦, 程宇奇, 朱凱光
地球信息探測儀器教育部重點(diǎn)實(shí)驗(yàn)室,吉林大學(xué)儀器科學(xué)與電氣工程學(xué)院, 長春 130026
?
航空電磁數(shù)據(jù)主成分濾波重構(gòu)的噪聲去除方法
王凌群, 李冰冰, 林君, 謝賓, 王琦, 程宇奇, 朱凱光*
地球信息探測儀器教育部重點(diǎn)實(shí)驗(yàn)室,吉林大學(xué)儀器科學(xué)與電氣工程學(xué)院, 長春 130026
主成分分析方法利用低階主成分重構(gòu)航空電磁數(shù)據(jù),解決了航空電磁探測中噪聲與數(shù)據(jù)在頻譜重疊情況下的噪聲壓制問題,但是參與重構(gòu)的低階主成分仍包含高頻空間噪聲,影響數(shù)據(jù)成像精度.本文提出的主成分濾波重構(gòu)去噪方法,根據(jù)自適應(yīng)窗寬平滑算法,設(shè)計(jì)了主成分低通濾波器組,對參與重構(gòu)的低階主成分進(jìn)行測線濾波,再將濾波后的低階主成分重構(gòu)為電磁信號,不僅可以去除低階主成分中的高頻空間噪聲,而且去除了高階主成分包含的不相關(guān)噪聲.仿真數(shù)據(jù)的去噪結(jié)果表明,主成分濾波重構(gòu)獲得較高的信噪比,較常規(guī)測線濾波與主成分重構(gòu)分別提高了10.96 dB和2.52 dB;電導(dǎo)率深度成像結(jié)果證明了主成分濾波重構(gòu)方法能夠提高地下深部異常體的識別能力.最后通過實(shí)測數(shù)據(jù)的成像結(jié)果進(jìn)一步驗(yàn)證了本文研究的主成分濾波重構(gòu)去噪方法的有效性.
時(shí)間域航空電磁; 主成分; 濾波重構(gòu); 自適應(yīng)窗寬; 空間噪聲
時(shí)間域航空電磁探測具有勘查速度快、探測范圍廣等優(yōu)勢,在我國具有廣泛的應(yīng)用前景.但是其機(jī)載的飛行探測方式,能夠引起發(fā)射線圈、接收線圈晃動,同時(shí)飛行速度、飛行姿態(tài)等變化都能引起探測系統(tǒng)裝置參數(shù)的不穩(wěn)定,而且空中飛行探測過程中氣壓、溫度等變化也能夠引入系統(tǒng)電子參數(shù)改變,這些系統(tǒng)裝置參數(shù)以及電子參數(shù)的變化都引入系統(tǒng)噪聲,嚴(yán)重影響數(shù)據(jù)質(zhì)量及成像精度,制約航空電磁探測系統(tǒng)對地下深部異常體及小異常體的反演解釋,減小勘探深度.
常規(guī)數(shù)據(jù)處理,如時(shí)域疊加、頻域?yàn)V波(Macnae et al.,1984;Buselli et al.,1998;Ridsdill and Dentith,1999;李楠,2009;呂東偉,2011)可以消除大部分隨機(jī)噪聲,但對于與信號頻譜重疊的噪聲卻有很大的局限性(Lane et al,1998,2000).針對這一問題,20世紀(jì)90年代末期,國外學(xué)者利用數(shù)據(jù)的相關(guān)性,采用主成分分析法(Principal Component Analysis,PCA)研究數(shù)據(jù)的去除噪聲方法.Jones 和 Levy(1987)利用相關(guān)性對數(shù)據(jù)進(jìn)行降維壓縮,去除了地震數(shù)據(jù)中的不相關(guān)噪聲,提高了信噪比;Green(1998)利用主成分對圖像信息進(jìn)行降維壓縮,提高了圖像質(zhì)量;主成分分析通過調(diào)整噪聲的奇異值分解,去除了射線光譜中的噪聲(Minty and Hovgaard,2002);Qian和Fowler(2007)實(shí)現(xiàn)了光譜去相關(guān)以及光譜降維,提高了信息的保存性;主成分分析法不僅可以壓制電磁數(shù)據(jù)中的噪聲,還被運(yùn)用到反演計(jì)算中(Kass and Li,2007;Kass et al.,2010);2011年,PCA被應(yīng)用去除心電信號中的偽影和噪聲,有效地提取有用的心電信號(Chawla,2011).近年來,我國學(xué)者也開始將主成分分析方法引入到數(shù)據(jù)處理中,利用PCA去除高光譜遙感數(shù)據(jù)噪聲(常威威等,2009);利用主成分分析法成功地壓制了乘性噪聲(姚莉麗等,2011);Zhu等(2012)將主成分引入到航空電磁數(shù)據(jù)的反演中,在噪聲數(shù)據(jù)的反演解釋中,取得了優(yōu)于其他反演方法的結(jié)果;2013年,主成分分析法應(yīng)用于航空電磁數(shù)據(jù)的噪聲處理中,解決了由于噪聲與數(shù)據(jù)的時(shí)頻特性重疊而帶來的去噪困難的問題(朱凱光等,2013).
上述主成分分析去噪方法,均是將數(shù)據(jù)轉(zhuǎn)化成一組按方差由大到小排列的不相關(guān)的主成分,采用少量低階主成分重構(gòu)原數(shù)據(jù),從而去除高階主成分包含的不相關(guān)噪聲.但是,這些參與重構(gòu)的低階主成分中仍然包含一定程度噪聲(朱凱光等,2013,圖2),對數(shù)據(jù)的精細(xì)處理與信息提取影響很大.
針對這一問題,本文提出基于主成分濾波重構(gòu)的航空電磁數(shù)據(jù)去噪方法,首先分析了參與重構(gòu)測線數(shù)據(jù)的低階主成分中的高頻空間噪聲特點(diǎn),研究了自適應(yīng)窗寬的低通濾波器,設(shè)計(jì)了主成分低通濾波器組,并詳細(xì)給出了主成分濾波重構(gòu)算法,最后通過仿真數(shù)據(jù)和實(shí)測數(shù)據(jù)的主成分濾波重構(gòu)去噪實(shí)例,結(jié)合電導(dǎo)率深度成像結(jié)果對比,驗(yàn)證了本文去噪算法的有效性.
文獻(xiàn)(朱凱光等,2013)給出的主成分噪聲去除結(jié)果表明,主成分重構(gòu)去噪方法雖然去除了高階主成分中的不相關(guān)噪聲,但是參與重構(gòu)的低階主成分仍將其包含的高頻空間噪聲引入測線數(shù)據(jù).
2.1 主成分濾波的去噪原理
基于主成分重構(gòu)去噪方法,本文首先對參與重構(gòu)的低階主成分沿測線低通濾波,再進(jìn)行航空電磁數(shù)據(jù)重構(gòu),不僅能夠去除高階主成分中包含的不相關(guān)噪聲,而且能有效地壓制低階主成分中的高頻空間噪聲.
設(shè)航空電磁探測某m道n個測點(diǎn)的測線數(shù)據(jù)為X(m×n),利用文獻(xiàn)(朱凱光等,2013)中的式(1)—(3),計(jì)算其自相關(guān)陣的特征向量矩陣R(m×m)、特征值矩陣λ(m×1),得到測線數(shù)據(jù)n個測點(diǎn)的m個主成分Ψ(m×n).根據(jù)式(1)計(jì)算前p個主成分的累計(jì)貢獻(xiàn)率δp為
(1)
式中,λk為第k階主成分對應(yīng)的特征值.當(dāng)前p個主成分的累計(jì)貢獻(xiàn)率高于95%時(shí),選p為重構(gòu)電磁信號的低階主成分個數(shù).
(2)
那么,濾波后的第k階主成分可表示為
(3)
同樣,根據(jù)式(2)和式(3)可得到低通濾波器組濾波后的前p個低階主成分的剖面數(shù)據(jù),則濾波后的主成分矩陣為
(4)
2.2 主成分濾波器設(shè)計(jì)
(5)
(6)
根據(jù)各測點(diǎn)的自適應(yīng)窗寬Wk(j),得到第k階主成分濾波器的單位沖激響應(yīng)函數(shù)hk(j)為(鄭君里等,2000)
(7)
式中j=1,2,…,n.
但是對剖面曲線起始和尾部測點(diǎn)進(jìn)行濾波時(shí),為避免濾波點(diǎn)數(shù)不足,窗寬采用下式計(jì)算:
Wk(j)=
(8)
本文采用自適應(yīng)窗寬平滑算法設(shè)計(jì)的主成分低通濾波器組,能夠根據(jù)信號的局部特性自適應(yīng)地改變?yōu)V波器頻帶,比常規(guī)固定帶寬的低通濾波器具有明顯優(yōu)勢,不僅能夠有效去除主成分剖面的高頻空間噪聲,還可以保證主成分異常的幅度.但是,最佳濾波參數(shù)的確定尤為重要,濾波過度導(dǎo)致剖面弱異常被濾掉,而欠濾波會導(dǎo)致噪聲過大而形成深部假異常.因此,不僅要根據(jù)實(shí)測數(shù)據(jù)處理經(jīng)驗(yàn)、當(dāng)?shù)氐刭|(zhì)條件,還要根據(jù)電磁數(shù)據(jù)剖面異常形態(tài)甚至是各測點(diǎn)的off-time段衰減曲線特征,反復(fù)調(diào)整才能確定最佳濾波參數(shù).
3.1 時(shí)間域航空電磁探測系統(tǒng)的仿真數(shù)據(jù)
為評價(jià)主成分濾波重構(gòu)去噪效果,特別是晚期道的噪聲水平,本文設(shè)計(jì)了含有深部異常體的準(zhǔn)二維大地模型,如圖1所示,在電導(dǎo)率為0.02 S·m-1的均勻半空間模型400 m深處,有一個厚50 m、長1 km的異常體,測點(diǎn)間隔5 m,測線共有1000個測點(diǎn).
時(shí)間域航空電磁探測系統(tǒng)為中心回線方式,發(fā)射線圈半徑為7.5 m,飛行高度為25 m,發(fā)射波形為基頻25 Hz的正負(fù)方波,電流強(qiáng)度歸一化為1 A (實(shí)際發(fā)射電流一般為300~500 A),經(jīng)正演計(jì)算(朱凱光等,2010;殷長春等,2013),得到各測點(diǎn)的17道(0.2~10.76 ms)off-time段電磁響應(yīng),剖面曲線如圖2中藍(lán)色線所示.為仿真野外實(shí)測數(shù)據(jù),在正演數(shù)據(jù)中加入含基頻的復(fù)雜噪聲,形成的剖面曲線如圖2中紅色線所示.
由圖2可以看到,正演數(shù)據(jù)(藍(lán)色線)的剖面曲線在第400~600測點(diǎn)處存在明顯的晚期異常,與理論模型的深部異常體一致;而含噪數(shù)據(jù)(紅色線)晚期道的異常幾乎被噪聲淹沒,無法有效地反映地下深部異常體.
3.2 主成分濾波去噪結(jié)果分析與對比
采用文獻(xiàn)(朱凱光等,2013)中的式(1)—(3)分別計(jì)算正演數(shù)據(jù)和含噪數(shù)據(jù)的17個主成分,前兩個低階主成分的累計(jì)貢獻(xiàn)率約為99%,因此選擇第1與第2階主成分重構(gòu)電磁數(shù)據(jù).正演數(shù)據(jù)與含噪數(shù)據(jù)的第1與第2階主成分如圖3中的藍(lán)色線和紅色線所示.
可以看到,含噪數(shù)據(jù)主成分的剖面數(shù)據(jù)(紅色線)包含高頻空間噪聲,如第1階主成分噪聲幅度的峰峰值約為0.005 nT·s-1.針對主成分剖面中的噪聲,本文采用自適應(yīng)窗寬濾波器,對參與重構(gòu)的第1、第2階主成分分別進(jìn)行測線濾波.設(shè)計(jì)濾波器的最大窗寬為51,最小窗寬為3,各主成分剖面數(shù)據(jù)的濾波結(jié)果如圖3中的綠色線所示.濾波后主成分剖面噪聲明顯減小,與正演數(shù)據(jù)主成分剖面基本一致.
圖1 準(zhǔn)二維大地模型示意圖Fig.1 Sketch of a pseudo-2D earth model
為對比去噪效果,本文分別采用常規(guī)測線濾波、主成分重構(gòu)和主成分濾波重構(gòu)對仿真數(shù)據(jù)進(jìn)行去噪處理,圖4給出了三種方法去噪后電磁數(shù)據(jù)的后四道剖面曲線.
對比圖4a、4b和4c可以看到,經(jīng)常規(guī)測線濾波處理后的剖面曲線(圖4a)的晚期道數(shù)據(jù)無明顯異常,無法反映大地模型的深部異常體;主成分重構(gòu)去噪處理(圖4b)的后四道剖面數(shù)據(jù)能夠基本恢復(fù)原剖面數(shù)據(jù),但仍有高頻殘余噪聲,影響晚期道異常的
圖2 準(zhǔn)二維大地模型的17道電磁數(shù)據(jù)剖面曲線Fig.2 A 17-channel profile of the pseudo-2D earth model
圖3 準(zhǔn)二維大地模型的主成分剖面曲線(a) 第1階主成分剖面曲線; (b) 第2階主成分剖面曲線.Fig.3 Profiles of PCs for the pseudo-2D earth model(a) Profile of PC1; (b) Profile of PC2.
圖4 準(zhǔn)二維大地模型的后四道剖面曲線去噪結(jié)果對比(a) 常規(guī)測線濾波結(jié)果; (b) 主成分重構(gòu)結(jié)果; (c) 主成分濾波重構(gòu)結(jié)果.Fig.4 Comparison of the last 4-channel profiles denoised by different methods for the pseudo-2D earth model(a) By filtered traditional profile;(b) PC reconstruction;(c) By filtered PC reconstruction.
圖5 去噪后的電導(dǎo)率深度成像結(jié)果對比圖(a) 常規(guī)測線濾波結(jié)果; (b) 主成分重構(gòu)結(jié)果; (c) 主成分濾波重構(gòu)結(jié)果.Fig.5 CDI comparison for the pseudo-2D model with data processed by three denoising methods(a) By filtered traditional profile;(b) PC reconstruction;(c) By filtered PC reconstruction.
圖6 河南野外航空電磁探測的17道剖面曲線Fig.6 Seventeen-channel profiles of field data from airborne time domain electromagnetic survey in Henan Province
圖7 實(shí)測數(shù)據(jù)去噪后電磁數(shù)據(jù)剖面曲線(a) 主成分重構(gòu); (b) 主成分濾波重構(gòu).Fig.7 Profiles after noise removal for survey data in Henan Province(a) PC reconstruction; (b) By filtered PC reconstruction.
圖8 實(shí)測數(shù)據(jù)去噪后電導(dǎo)率深度成像對比圖(a) 常規(guī)低通測線濾波; (b) 主成分重構(gòu); (c) 主成分濾波重構(gòu).Fig.8 CDI comparison for field data denoised by different methods(a) By filtered traditional profile;(b) PC reconstruction;(c) By filtered PC reconstruction.
識別;而主成分濾波重構(gòu)處理(圖4c)后的晚期道數(shù)據(jù)不僅可以清晰地看到異常位置,而且保證了異常的幅值,噪聲幅值明顯減小,信噪比較常規(guī)測線濾波提高了10.9633 dB,較主成分重構(gòu)提高了2.5234 dB.3.3 電導(dǎo)率深度成像結(jié)果對比
圖5分別給出了三種方法處理后的剖面數(shù)據(jù)的電導(dǎo)率深度成像(Conductivity-Depth Imaging,CDI)結(jié)果,常規(guī)測線濾波處理的數(shù)據(jù)CDI結(jié)果(圖5a)不能有效反映深部異常體位置,異常體沿測線出現(xiàn)橫向擴(kuò)散,甚至有部分區(qū)域出現(xiàn)偽異常;主成分重構(gòu)去噪的CDI結(jié)果(圖5b),能夠較清晰地看到深部異常體,但同深度由于殘余噪聲沿測線仍出現(xiàn)小部分異常擴(kuò)散;而主成分濾波重構(gòu)去噪的CDI結(jié)果(圖5c)效果明顯,與理論模型接近,表明該方法有較強(qiáng)壓制噪聲能力,較其他兩種方法具有更好的成像精度.可見,本文提出的主成分濾波重構(gòu)算法能夠提高航空電磁探測對深部異常體的識別能力.
4.1 實(shí)測剖面數(shù)據(jù)的去噪結(jié)果對比
2012年4月,國土資源部航空物探遙感中心與吉林大學(xué)合作研制開發(fā)的我國首套完整吊艙式時(shí)間域直升機(jī)航空電磁探測系統(tǒng),在河南省某地勘查測量.該地區(qū)地質(zhì)結(jié)構(gòu)顯高阻特性,電阻率約為3000~8000 Ωm,電磁響應(yīng)幅值較小.本文以河南野外實(shí)驗(yàn)中某條測線的17道(0.2~10.76 ms)剖面數(shù)據(jù)為例,常規(guī)測線濾波處理的剖面數(shù)據(jù)如圖6所示,在測點(diǎn)1100附近有幅值約為200 nT·s-1的異常.
對該測線的剖面數(shù)據(jù)進(jìn)行主成分重構(gòu)與主成分濾波重構(gòu)的去噪處理,結(jié)果分別如圖7a和圖7b所示.對比圖6和圖7,可以看到本文研究的主成分濾波重構(gòu)算法取得了最好的去噪結(jié)果(圖7b),去除了測線上的高頻空間噪聲,有助于對深部異常體的識別.
4.2 電導(dǎo)率深度成像結(jié)果對比
對三種方法去噪后的電磁數(shù)據(jù)(圖6、圖7a和圖7b)分別進(jìn)行電導(dǎo)率深度成像,結(jié)果分別如圖8a、圖8b和圖8c所示.可以看到,三種數(shù)據(jù)去噪方法都能明顯顯出第1100測點(diǎn)處的異常體,但是常規(guī)測線濾波處理的數(shù)據(jù)成像結(jié)果(圖8a)還顯示,在地下深部450 m處存在幾乎連成一層的低阻異常,無法準(zhǔn)確的圈定地下異常體位置;主成分重構(gòu)處理的結(jié)果也出現(xiàn)了不同程度的低阻異常擴(kuò)散現(xiàn)象,與當(dāng)?shù)氐母咦璧刭|(zhì)情況不符,是測線高頻空間噪聲所致;但是主成分濾波重構(gòu)的CDI結(jié)果(圖8c)顯示,本文研究算法有效地壓制了測線高頻噪聲,并保持測線異常幅度,獲得清晰的電導(dǎo)率成像結(jié)果.
(1)本文將航空電磁數(shù)據(jù)噪聲處理變換到主成分域中進(jìn)行,經(jīng)主成分分解、去噪、再重構(gòu),既濾除了低階主成分中的高頻空間噪聲,又去除高階主成分中的不相關(guān)噪聲.主成分濾波重構(gòu)去噪方法不僅可以提高數(shù)據(jù)晚期道的信噪比,而且增強(qiáng)了航空電磁探測系統(tǒng)對深部異常體的分辨能力,并成為時(shí)域、頻域噪聲處理方法的重要補(bǔ)充,為地球物理探測數(shù)據(jù)的變換域分解與合成以及噪聲處理等方面提供了新思路.
(2)本文采用自適應(yīng)窗寬濾波算法,設(shè)計(jì)的低通濾波器組能夠根據(jù)航空電磁數(shù)據(jù)各階主成分剖面數(shù)據(jù)的局部變化特征,自適應(yīng)地改變?yōu)V波器的帶寬,不僅可以有效地濾除主成分的高頻空間噪聲,而且有效地保持了異常的幅值,具有優(yōu)于常規(guī)低通濾波器的濾波性能.該濾波算法也可有效地用于測線電磁數(shù)據(jù)的濾波.
致謝 衷心感謝吉林大學(xué)地球信息探測儀器教育部重點(diǎn)實(shí)驗(yàn)室為本文提供的吊艙式時(shí)間域直升機(jī)航空電磁探測系統(tǒng)的實(shí)測數(shù)據(jù),感謝吉林大學(xué)時(shí)間域航空電磁組的全體成員對此文的幫助和指導(dǎo).
Buselli G, Hwang H S, Pik J P. 1998. AEM noise reduction with remote referencing.ExplorationGeophysics, 29(2): 71-76.
Chang W W, Guo L, Liu K, et al. 2009. Denoising of hyperspectral data based on contourlet transform and principal component analysis.JournalofElectronics&InformationTechnology(in Chinese), 31(12): 2892-2896.
Chawla M P S. 2011. PCA and ICA processing methods for removal of artifacts and noise in electrocardiograms: A survey and comparison.AppliedSoftComputing, 11(2): 2216-2226.
Green A. 1998. The use of multivariate statistical techniques for the analysis and display of AEM data.ExplorationGeophysics, 29(2): 77-82.
Jones I F, Levy S. 1987. Signal-to-noise ratio enhancement in multichannel seismic data via the karhunen-loéve transform.GeophysicalProspecting, 35(1): 12-32.
Kass M A, Li Y G. 2007. Use of principal component analysis in the de-noising and signal-separation of transient electromagnetic data. The 3rdInternational Conference on Environmental and Engineering Geophysics (ICEEG). Wuhan, China. Kass M A, Li Y G, Krahenbuhl R, et al. 2010. Enhancement of TEM data and noise characterization by principal component analysis. Douglas Oldenburg: Department of Geophysics Colorado School of Mines.Lane R, Green A, Golding C, et al. 2000. An example of 3D conductivity mapping using the TEMPEST airborne electromagnetic system.ExplorationGeophysics, 31(2): 162-172. Lane R, Plunkett C, Price A, et al. 1998. Streamed data—a source of insight and improvement for time domain airborne EM.ExplorationGeophysics, 29(2): 16-23.
Li N. 2009. Research on Airborne time-domain electromagnetic data preprocessing [Master thesis] (in Chinese). Changchun: Jilin University.
Lü D W. 2011. Methods study of helicopter-borne towed bird time domain electromagnetic data processing [Master thesis] (in Chinese). Chengdu: Chengdu University of Technology. Macnae J C, Ltagne Y, West G F. 1984. Noise processing techniques for time-domain EM systems.Geophysics, 49(7): 934-948.Minty B, Hovgaard J. 2002. Reducing noise in gamma-ray spectrometry using spectral component analysis.ExplorationGeophysics, 33(4): 172-176. O′Connell M D. 2010. Adaptive width filters for GEOTEM data. Canada, Ottawa: Consultant to Fugro Airborne Surveys.
Qian D, Fowler J E. 2007. Hyperspectral image compression using
JPEG2000 and principal component analysis.IEEEGeoscienceandRemoteSensingLetters, 4(2): 201-205.
Ridsdill-Smith T A, Dentith M C. 1999. The wavelet transform in aeromagnetic processing.Geophysics, 64(4): 1003-1013.
Yao L L, Feng X C, Li Y F. 2011. Principal component analysis method for muitiplicative noise removal.ActaPhotonicaSinica(in Chinese), 40(7): 1031-1035.
Yin C C, Huang W, Ben F. 2013. The full-time electromagnetic modeling for time-domain airborne electromagnetic system.ChineseJournalofGeophysics(in Chinese), 53(3): 743-750, doi: 10.6038/cjg20130928.
Zheng J L, Ying Q H, Yang W L. 2000. Signals and Systems (in Chinese). Beijing: Higher Education Press.
Zhu K G, Lin J, Han Y H, et al. 2010. Research on conductivity depth imaging of time domain helicopter-borne electromagnetic data based on neural network.ChineseJournalofGeophysics(in Chinese), 53(3): 743-750, doi: 10.3969/j.issn.0001-5733.2010.03.030.
Zhu K G, Ma M Y, Che H W, et al. 2012. PC-based artificial neural network inversion for airborne time-domain electromagnetic data.AppliedGeophysics, 9(1): 1-8. Zhu K G, Wang L Q, Xie B, et al. 2013. Noise removal for airborne electromagnetic data based on principal component analysis.TheChineseJournalofNonferrousMetals(in Chinese), 23(9): 2430-2435.
附中文參考文獻(xiàn)常威威, 郭雷, 劉坤等. 2009. 基于Contourlet 變換和主成分分析的高光譜數(shù)據(jù)噪聲消除方法. 電子與信息學(xué)報(bào), 31(12): 2892-2896. 李楠. 2009. 時(shí)間域航空電磁數(shù)據(jù)預(yù)處理技術(shù)研究[碩士論文]. 長春: 吉林大學(xué).
呂東偉. 2011. 吊艙式時(shí)間域直升機(jī)航空電磁數(shù)據(jù)處理方法研究[碩士論文]. 成都: 成都理工大學(xué).
姚莉麗, 馮象初, 李亞峰. 2011. 去除乘性噪音的主成分分析算法. 光子學(xué)報(bào), 40(7): 1031-1035.
殷長春, 黃威, 賁放. 2013. 時(shí)間域航空電磁系統(tǒng)瞬變?nèi)珪r(shí)響應(yīng)正演模擬. 地球物理學(xué)報(bào), 56(9): 3153-3162, doi: 10.6038/cjg20130928. 鄭君里, 應(yīng)啟珩, 楊為理. 2000. 信號與系統(tǒng). 北京: 高等教育出版社. 朱凱光, 林君, 韓悅慧等. 2010. 基于神經(jīng)網(wǎng)絡(luò)的時(shí)間域直升機(jī)電磁數(shù)據(jù)電導(dǎo)率深度成像. 地球物理學(xué)報(bào), 53(3): 743-750, doi: 10.3969/j.issn.0001-5733.2010.03.030.
朱凱光, 王凌群, 謝賓等. 2013. 基于主成分分析的航空電磁數(shù)據(jù)噪聲去除方法. 中國有色金屬學(xué)報(bào), 23(9): 2430-2435.
(本文編輯 張正峰)
Noise removal based on reconstruction of filtered principal components
WANG Ling-Qun, LI Bing-Bing, LIN Jun, XIE Bin, WANG Qi, CHENG Yu-Qi, ZHU Kai-Guang*
KeyLaboratoryofGeo-ExplorationInstrumentation,MinistryofEducation,JiLinUniversity,Changchun130026,China
Airborne electromagnetic data has complex noise due to the flight environment, changes of system parameters and device parameters. The traditional time-frequency approach is difficult to remove noise. PCA (principal component analysis) can remove the noise which overlaps with the signal frequency spectrum. However the low-order components still contain high frequency spatial noise. To solve this problem, this work proposes the reconstruction of filtered principal components which can not only remove the high-frequency spatial noise of low-order components, but also remove uncorrelated noise of high-order components.This approach uses filtered principal components to reconstruct electromagnetic data. Firstly, it computes eigenvectors and eigenvalues matrix of the correlation matrix and the principal component profiles which are uncorrelated. The low-order principal components associated with the big eigenvalues reflect the correlated electromagnetic signals, while high-order principal components associated with the small eigenvalues are corresponding to the uncorrelated noise. It operates by first filtering the principal component profiles with the widest filter. Then according to local variation of each smoothed principal component profile, a group of low-pass filters is designed. Local variation of each smoothed principal component profile is converted linearly into adaptive smoothing filter window width of the measuring point. So each principal component profile is filtered by the corresponding filter. Finally, filtered low-order principal components are used to reconstruct electromagnetic signal.This paper designed a pseudo-2D earth model with a deep target. The noise added to it drowned the target of later channels and the CDI (conductivity depth imaging) cannot reflect the deep target. After PCA processing, principal component profiles of noise-contaminated data have high frequency spatial noise compared with those of the forward data. The peak to peak of noise amplitude of the first principal component is about 0.005 nT/s. With an adaptive window wide filter, noise of the principal component profile is significantly reduced and consistent with the forward data mostly. The data is processed by three de-noising methods. The later channels profiles by the traditional profile filter have no obvious target and CDI cannot reflect the deep target either. The later channels profiles by principal component reconstruction still contain high frequency noise, impacting identification of the deep target. But data by filtered principal component reconstruction not only eliminates high frequency noise, but also ensures the magnitude of the target. It shows the more accurate position and shape. SNR is improved by 10.96 dB and 2.52 dB relative to the other two methods.This work deniosed one survey line measured in Henan Province. The CDI results of other two methods exhibit low-resistance target diffusion phenomenon to different degrees and cannot show exact location of the target for this survey line. But the CDI by filtered principal component reconstruction indicates that the algorithm in this paper can suppress high frequency noise effectively, consistent with real local condition.ATEM data is converted into the principal component domain. A group of low-pass filters is able to change the filter bandwidth adaptively according to local variation of each principal component profile. It can not only filter out high frequency spatial noise, but also maintain the target amplitude. At the same time, it also improves the SNR of latest channel data and enhances the ability to identify the deep target after filtered principal component reconstruction. Filtered principal component reconstruction is an important complement to noise processing of time and frequency domain, providing a new line of thought for decomposition and synthesis of the transform domain and de-noising processing of geophysical data.
Time-domain airborne electromagnetism; Principal component analysis; Reconstruction after filtering; Adaptive width algorithm; Spatial noise
國家高技術(shù)研究發(fā)展計(jì)劃項(xiàng)目(2013AA063904),國家自然科學(xué)基金項(xiàng)目(41274076)和國家重大科研裝備研制項(xiàng)目(ZDYZ2012-1-03)聯(lián)合資助.
王凌群,女,1988年生,吉林大學(xué)博士研究生,研究方向?yàn)楹娇针姶艛?shù)據(jù)處理與噪聲壓制. E-mail:399127431@qq.com
*通訊作者朱凱光,女,1970年生,吉林大學(xué)教授,博士生導(dǎo)師,研究方向?yàn)殡姶盘綔y技術(shù)與信號處理. E-mail:zhukaiguang@jlu.edu.cn
10.6038/cjg20150815.
10.6038/cjg20150815
P631
2014-01-20,2014-10-28收修定稿
王凌群, 李冰冰, 林君等. 2015. 航空電磁數(shù)據(jù)主成分濾波重構(gòu)的噪聲去除方法.地球物理學(xué)報(bào),58(8):2803-2811,
Wang L Q, Li B B, Lin J, et al. 2015. Noise removal based on reconstruction of filtered principal components.ChineseJ.Geophys. (in Chinese),58(8):2803-2811,doi:10.6038/cjg20150815.