国产日韩欧美一区二区三区三州_亚洲少妇熟女av_久久久久亚洲av国产精品_波多野结衣网站一区二区_亚洲欧美色片在线91_国产亚洲精品精品国产优播av_日本一区二区三区波多野结衣 _久久国产av不卡

?

HHT和HMM在血細(xì)胞信號(hào)識(shí)別中的應(yīng)用

2017-01-12 09:14尹璀陶凌龍偉
現(xiàn)代電子技術(shù) 2016年23期
關(guān)鍵詞:經(jīng)驗(yàn)?zāi)B(tài)分解特征提取

尹璀++陶凌++龍偉

摘 要: 針對(duì)血細(xì)胞信號(hào)具有多形態(tài)、非線性、非平穩(wěn)的特點(diǎn),提出將希爾伯特黃變換(HHT)和隱馬爾可夫模型(HMM)相結(jié)合的血細(xì)胞信號(hào)識(shí)別方法。該方法采用HHT對(duì)血細(xì)胞信號(hào)進(jìn)行分析,選取經(jīng)過(guò)經(jīng)驗(yàn)?zāi)B(tài)分解得到的各本質(zhì)模態(tài)函數(shù)中相關(guān)性較大的分量,以這些分量的能量矩作為信號(hào)的特征量,由HMM訓(xùn)練得到正常人和病患者的模型參數(shù)并用做分類識(shí)別。實(shí)驗(yàn)結(jié)果表明,該方法可以較好地識(shí)別正常人和病患者的血細(xì)胞信號(hào),綜合準(zhǔn)確率達(dá)89.13%。

關(guān)鍵詞: 信號(hào)檢測(cè)與分析; 希爾伯特黃變換; 經(jīng)驗(yàn)?zāi)B(tài)分解; 隱馬爾科夫模型; 特征提??; 血細(xì)胞信號(hào)分析

中圖分類號(hào): TN911.7?34; TP 391.4 文獻(xiàn)標(biāo)識(shí)碼: A 文章編號(hào): 1004?373X(2016)23?0058?05

Application of HHT and HMM in blood cell signal recognition

YIN Cui, TAO Ling, LONG Wei

(School of Information Engineering, Nanchang University, Nanchang 330031, China)

Abstract: For the multi?form, nonlinear and non?stationary characteristics of the blood cell signal, the blood cell signal re?cognition method based on Hilbert Huang transform (HHT) and hidden Markov model (HMM) is proposed. The HHT is used in the method to analyze the blood cell signal. The strong dependency components in each intrinsic mode function obtained with empirical mode decomposition are selected, and their energy moments are taken as the signal feature value to achieve the model parameters of healthy people and patient by HMM training for classification and recognition. The experimental results indicate this method can recognize the blood cells signals of the healthy people and patient, and the synthetical accuracy rate can reach up to 89.13%.

Keywords: signal detection and analysis; Hilbert Huang transform; empirical mode decomposition; hidden Markov mo?del; feature extraction; blood cell signal analysis

0 引 言

作為一種非常普遍的檢測(cè)方法,血細(xì)胞分析在臨床疾病的診斷及健康體檢等方面發(fā)揮著重要的作用[1]。一般地,疾病會(huì)引起血液中紅細(xì)胞、白細(xì)胞、血小板等血細(xì)胞數(shù)量變化,因此通過(guò)對(duì)血細(xì)胞的分類識(shí)別有助于臨床上判斷人體健康與否。目前國(guó)內(nèi)外的血細(xì)胞分析儀大多采用庫(kù)爾特原理采集原始信號(hào),基本原理是懸浮在電解液中的血細(xì)胞隨電解液通過(guò)小孔管時(shí),會(huì)導(dǎo)致小孔管內(nèi)外兩電極間電阻發(fā)生瞬時(shí)變化,產(chǎn)生電位脈沖[2]。血細(xì)胞的大小和數(shù)目會(huì)引起脈沖信號(hào)的大小和次數(shù)的變化。針對(duì)血細(xì)胞信號(hào)多形態(tài)、非線性和非平穩(wěn)的特點(diǎn),一般需采用短時(shí)傅里葉變換(Short?time Fourier Transform,STFT),小波變換(Wavelet Transform,WT)或Wigner?Ville分布等時(shí)頻分析的方法,但這些方法分析非平穩(wěn)信號(hào)均有各自的不足,短時(shí)傅里葉變換容易受到窗函數(shù)的影響,小波變換的結(jié)果在很大程度上取決于小波的選擇,Wigner?Ville分布易受到交叉相干擾[3]。本研究采用由Huang提出的希爾伯特黃變換(Hilbert Huang Transform,HHT)方法提取血細(xì)胞信號(hào)特征向量,以避免上述種種不足[4],并采用隱馬爾可夫模型對(duì)血細(xì)胞的特征向量進(jìn)行分類識(shí)別,實(shí)現(xiàn)對(duì)正常人和病患者的血細(xì)胞信號(hào)分類識(shí)別。

1 希爾伯特黃變換

HHT由兩大主要部分組成:第一部分為經(jīng)驗(yàn)?zāi)B(tài)分解(Empirical Mode Decomposition,EMD),它是由Huang所創(chuàng)立的信號(hào)篩選方法[5];第二部分為Hilbert譜分析(Hilbert Spectrum Analysis,HSA)。EMD是HHT的核心算法,不同于其他的時(shí)頻分析方法,HHT具有直接獲得、自適應(yīng)性等特點(diǎn),分解的過(guò)程是基于原數(shù)據(jù)獲得的后驗(yàn)基函數(shù),這一點(diǎn)與小波變換有著很大的不同,同時(shí)也是相對(duì)于小波變換的巨大優(yōu)勢(shì),因?yàn)椴煌男〔ɑ瘮?shù)對(duì)小波分析結(jié)果的好壞有很大的影響,由于對(duì)并不需要固定的基函數(shù),針對(duì)不同的待處理信號(hào)可以自適應(yīng)地分解出有限的多個(gè)本質(zhì)模態(tài)函數(shù)(Intrinsic Mode Function,IMF)。這些IMF在經(jīng)過(guò)Hilbert變換后得到瞬時(shí)頻率,最終輸出的Hilbert譜即為所有IMF瞬時(shí)頻率的集合。

5 結(jié) 語(yǔ)

本文結(jié)合HHT和HMM的優(yōu)點(diǎn),將非平穩(wěn)的血細(xì)胞信號(hào)經(jīng)驗(yàn)?zāi)B(tài)分解為多個(gè)IMF分量,選取其中與原信號(hào)相關(guān)系數(shù)較大的分量作為有效分量,分析提取出其關(guān)于時(shí)間能量的特征向量。而HMM模型又反映出特征向量的隱藏狀態(tài)是正常人或病患者之間的聯(lián)系,通過(guò)實(shí)驗(yàn)分析,獲得了很有效的識(shí)別結(jié)果。

參考文獻(xiàn)

[1] 曾斌,鄭會(huì)豐,林奕麗,等.早期糖尿病腎病患者血細(xì)胞形態(tài)變化臨床分析[J].中華全科醫(yī)學(xué),2011,9(2):179?180.

[2] GARRIDO M, ARUNDELL M, VALENCIA, et al. High?speed particle detection in a micro?Coulter counter with two?dimensional adjustable aperture [J]. Biosensors and bioelectronics, 2008, 24(2): 290?296.

[3] HAMDI S E, LE DUFF A, SIMON L, et al. Acoustic emission pattern recognition approach based on Hilbert?Huang transform for structural health monitoring in polymer?composite materials [J]. Applied acoustics, 2013, 74(5): 746?757.

[4] LAW L S, KIM J H, LIEW W Y H, et al. An approach based on wavelet packet decomposition and Hilbert?Huang transform (WPD?HHT) for spindle bearings condition monitoring [J]. Mechanical systems and signal processing, 2012, 33(2): 197?211.

[5] HUANG N E, SHEN Z, LONG S R, et al. The empirical mode decomposition method and the Hilbert spectrum for nonlinear and non?stationary time series analysis [J]. Journal of self?assemble and molecular electronics, 1998, 454: 903?995.

[6] BOUCHIKHI A, BOUDRAA A O. Multicomponent AM?FM signals analysis based on EMD?B?splines ESA [J]. Signal proces?sing, 2012, 92(9): 2214?2228.

[7] 董紅生,張愛(ài)華,邱天爽,等.分頻段Hilbert譜熵的心率變異信號(hào)分析方法[J].生物醫(yī)學(xué)工程學(xué)雜志,2011,28(2):248?254.

[8] LIN L, CHU F L. Feature extraction of AE characteristics in offshore structure model using Hilbert?Huang transform [J]. Measurement, 2011, 44(1): 46?54.

[9] ZHANG Xiaofei, TAO Ling, DENG Juan, et al. Blood cell signal classification via Hilbert?Huang transform combined with information entropy [J]. Journal of information & computational science, 2014, 12(15): 5603?5612.

[10] LUPU C, DAN S. Another look at some new Cauchy?Schwarz type inner product inequalities [J]. Applied mathematics and computation, 2014, 231: 463?477.

[11] 張小薊,張歆,孫進(jìn)才.基于IMF能量熵的目標(biāo)特征提取與分類方法[J].計(jì)算機(jī)工程與應(yīng)用,2008,44(4):68?69.

[12] BEH J, HAN D K, DURASIWANMI R, et al. Hidden Markov model on a unit hypersphere space for gesture trajectory recognition [J]. Pattern recognition letters, 2014, 36(15): 144?153.

[13] 龔勛,馮毅雄,譚建榮,等.基于空間結(jié)構(gòu)隱markov模型的故障診斷[J].計(jì)算機(jī)集成制造系統(tǒng),2012,18(1):132?140.

[14] 朱嘉瑜,高鷹.基于改進(jìn)粒子群算法的隱馬爾可夫模型訓(xùn)練[J].計(jì)算機(jī)工程與設(shè)計(jì),2010,31(1):157?160.

猜你喜歡
經(jīng)驗(yàn)?zāi)B(tài)分解特征提取
基于Gazebo仿真環(huán)境的ORB特征提取與比對(duì)的研究
基于Daubechies(dbN)的飛行器音頻特征提取
一種基于LBP 特征提取和稀疏表示的肝病識(shí)別算法
基于聯(lián)合采用EMD與高通低通濾波的信號(hào)分析
基于DSP的直線特征提取算法
礦山動(dòng)態(tài)輕軌衡系統(tǒng)稱重信號(hào)處理的研究
基于EMD的運(yùn)動(dòng)想象腦電特征提取與識(shí)別
基于MED和循環(huán)域解調(diào)的多故障特征提取
Walsh變換在滾動(dòng)軸承早期故障特征提取中的應(yīng)用