CN104644142B - A kind of signal processing algorithm of non-contact vital sign monitoring - Google Patents

Abstract

The present invention provides a kind of signal processing algorithm of non-contact vital sign monitor system, including：Anyway contact transformation is carried out to the I roads obtained by non-contact vital sign patient monitor, Q roads signal and tries to achieve quadrature demodulated signal, heartbeat and breath signal are separated using Local Integral average Empirical Mode Decomposition Algorithm, breathing, the Hilbert conversion sections of heartbeat signal.The present invention is separated using the average Empirical Mode Decomposition Algorithm of Local Integral to heartbeat and breath signal, and the Hilbert of breathing, heartbeat signal is converted, solve under the unconspicuous situation of fluctuating in thoracic cavity caused by heartbeat, the frequency and waveform of heartbeat and breathing are successively obtained according to the height of frequency, computational accuracy is high, practical application is strong, facilitates accurate, real-time medical worker, the continuous breathing for grasping patient and Heartbeat State.

Description

A kind of signal processing algorithm of non-contact vital sign monitoring

Technical field

The invention belongs to field of radar, a kind of signal transacting side of non-contact vital sign monitor system is specifically designed Method.

Background technology

Heartbeat and breathing be the important physical trait condition information of human body.Non-contact vital sign letter used at present Number processing method is Fourier conversion.With, it is necessary to using substantial amounts of data, and being breathed and heartbeat signal during this method Relatively, the fluctuating in the thoracic cavity as caused by heartbeat is less obvious for frequency, is easily flooded, makes by the higher hamonic wave of breath signal It is not easy to detect with traditional FFT methods.In addition, the frequency spectrum obtained by after FFT is merely capable of representing institute in signal The frequency of presence, but can not illustrate that a frequency content of signal appears in which of signal at moment, letter can not be represented Which number converted at moment.

The content of the invention

The present invention relies on non-contact vital sign monitor system to be realized.

The system is by the way of modern radar technology and biomedical technology are blended, based on Doppler's Cleaning Principle, Obtain the sign information of measured target, including breathing, heartbeat etc. in real time by way of wireless measurement；Mainly include radar signal Emitting portion, general radar signal simulation receiving portion and signal process part, wherein, general radar signal simulation receiving portion difference It is connected with radar signal part and signal process part.

Emitting portion is used to send continuous wave signal to measured target by transmitting antenna and is radiated at the thoracic cavity of target, even Continuous ripple signal obtains echo-signal after the chest cavity movement as caused by the periodic signals such as breathing and heartbeat is modulated to it；Return Ripple signal turns into quadrature digital signal of the two-way comprising breathing and heartbeat message by simulation receiving portion demodulation, is all the way I roads Signal, another road is Q roads signal；I roads and Q roads signal by signal process part carry out vital sign parameter signals extraction and Frequency detecting.

The technical solution for realizing the object of the invention is：

Step 1, non-contact vital sign monitor system analog end sends orthogonal two-way heartbeat and breathing numeral mixing Signal, is all the way I roads signal, and another road is Q roads signal, and carrying out arc tangent processing to I roads and Q roads obtains quadrature demodulated signal；

Step 2, quadrature demodulated signal is separated using Local Integral average Empirical Mode Decomposition Algorithm, successively by the heart Jump signal and breath signal is separated from quadrature demodulated signal；

Step 3, Hilbert conversion is carried out to the breathing isolated in step 2 and heartbeat signal, obtains breathing and heartbeat letter Number instantaneous frequency.

The present invention realizes goal of the invention using following technological means：

(1) separating step of breathing and heartbeat signal is as follows in step 2：

Step 2.1, initialized, make r₀(t)=x (t), h₀(t)=r_ii-1(t), wherein x (t) believes for quadrature demodulation Number；

Step 2.2, ii=1, jj=1 are made；

Step 2.3, h is found out_jj-1(t) whole Local Extremum (t_k,x_k), wherein k is the index value of extreme value, k=0,1, 2 ..., t_kIt is the time corresponding to k-th of extreme point, x_kIt is the amplitude of k-th of extreme point；

Step 2.4, according to extreme point (t_k,x_k), to be divided into one section by signal decomposition between two extreme points, each section is all It is monotonic increase or successively decreases, wherein kth section is signal in time t ∈ (t_k,t_k+1) on signal, be designated as C_k；

Step 2.5, each section of C is obtained using the method for integral mean_kLocal mean value point

Step 2.6, the local mean value point to being calculatedCubic spline interpolation fitting is carried out, its average bag is tried to achieve Network curve m_jj-1(t)；

Step 2.7, from signal h_jj-1(t) average envelope m is subtracted in_jj-1(t) h, is obtained_jj(t)=h_jj-1(t)-m_jj-1(t)；

Step 2.8, h is judged_jj(t) stop condition S whether is met_D, the h if meeting_jj(t) it is intrinsic for isolate one Modular function, now C_ii(t)=h_jj(t) step 2.8, is gone to；Otherwise, jj=jj+1, goes to step 2.2, until isolating one Intrinsic mode functions circulation terminates, and goes to step 2.9；

Wherein ε ∈ [0.2,0.3]

Step 2.9, from r_ii-1(t) intrinsic mode functions obtained by subtracting in step 2.7, obtain r_ii(t), r_ii(t)= r_ii-1(t)-C_ii(t)；

Step 2.10, r is judged_ii(t) whether the number of extreme point is less than two, is decomposed if extreme point number is less than two Terminate；If extreme point number is more than or equal to two, ii=ii+1 goes to step 2.2.

(2) detailed process of step 3 is as follows：

Step 3.1, the intrinsic mode functions C to being obtained in step 2.7_ii(t) Hilbert conversion is carried out, wherein Hilbert becomes It is changed toWherein H represents that Hilbert is converted, and p represents Cauchy's principal value；

Step 3.2, analytical function z is generated by the intrinsic mode functions isolated_i(t),

z_ii(t)=C_ii(t)+iH[C_ii(t)]=a_ii(t)e^iφ(t)

The imaginary part of l representative functions, a_ii(t)=[C² _ii(t)+H²(C_ii(t))]^1/2It is magnitude function,It is phase function；

Step 3.3, it is that can obtain instantaneous frequency to phase function derivation

The present invention compared with prior art, with advantages below：The present invention utilizes the average empirical mode decomposition of Local Integral (LIM-EMD) algorithm is separated to heartbeat and breath signal, and the Hilbert of breathing, heartbeat signal is converted, and is solved in heartbeat Under the unconspicuous situation of fluctuating in caused thoracic cavity, the frequency and ripple of heartbeat and breathing are successively obtained according to the height of frequency Shape, computational accuracy is high, and practical application is strong, facilitates accurate, real-time medical worker, the continuous breathing for grasping patient and heartbeat shape Condition.

With reference to Figure of description, the present invention will be further described.

Brief description of the drawings

Fig. 1 is the signal processing flow block diagram of the non-contact vital sign monitor system of the present invention；

Fig. 2 is actual I, Q two paths of signals and composite signal waveform of the invention；Fig. 2 (a) is I roads signal waveforms, Fig. 2 (b) it is Q roads signal waveforms, Fig. 2 (c) is quadrature demodulated signal oscillogram；

Fig. 3 is the spectrogram after traditional FFT methods processing；

Fitting respiratory waveform, Fig. 4 (b) heartbeat signal spectrograms of the Fig. 4 (a) for the present invention；Fig. 4 (c) heartbeat signals are instantaneous Frequency；

Real-time heart beat waveform, Fig. 5 (b) real-time heart beat signal spectrum figures of the Fig. 5 (a) for the present invention；Fig. 5 (c) real-time heart beats Signal transient frequency.

Embodiment

With reference to Fig. 1, a kind of signal processing algorithm of non-contact vital sign monitoring, it is characterised in that including following step Suddenly：

Step 1, non-contact vital sign monitor system analog end sends orthogonal two-way heartbeat and breathing numeral mixing Signal, is all the way I roads signal (in-phase branch), and another road is Q roads signal (quadrature branch), and I roads and Q roads are carried out at arc tangent Reason obtains quadrature demodulated signal；

Step 2, quadrature demodulated signal is separated using Local Integral average Empirical Mode Decomposition Algorithm, successively by the heart Jump signal and breath signal is separated from quadrature demodulated signal；

Step 3, Hilbert conversion is carried out to the breathing isolated in step 2 and heartbeat signal, obtains breathing and heartbeat letter Number instantaneous frequency.

Feature in terms of heartbeat and breath signal separate section are the time scale of basis signal in itself is entered to original signal Row is decomposed, and primary signal is resolved into limited intrinsic mode functions, and the frequency of heartbeat and breathing is successively obtained according to the height of frequency The separating step of breathing and heartbeat signal is as follows in rate and waveform, step 2：

Step 2.1, initialized, make r₀(t)=x (t), h₀(t)=r_ii-1(t), wherein x (t) believes for quadrature demodulation Number；

Step 2.2, ii=1, jj=1 are made；

Step 2.3, h is found out_jj-1(t) whole Local Extremum (t_k,x_k), wherein k is the index value of extreme value, k=0,1, 2 ..., t_kIt is the time corresponding to k-th of extreme point, x_kIt is the amplitude of k-th of extreme point；

Step 2.4, according to extreme point (t_k,x_k), to be divided into one section by signal decomposition between two extreme points, each section is all It is monotonic increase or successively decreases, wherein kth section is signal in time t ∈ (t_k,t_k+1) on signal, be designated as C_k；

Step 2.5, each section of C is obtained using the method for integral mean_kLocal mean value point

Step 2.6, the local mean value point to being calculatedCubic spline interpolation fitting is carried out, its average bag is tried to achieve Network curve m_jj-1(t)；

Step 2.7, from signal h_jj-1(t) average envelope m is subtracted in_jj-1(t) h, is obtained_jj(t)=h_jj-1(t)-m_jj-1(t)；

Step 2.8, h is judged_jj(t) stop condition S whether is met_D, the h if meeting_jj(t) it is intrinsic for isolate one Modular function, now C_ii(t)=h_jj(t) step 2.8, is gone to；Otherwise, jj=jj+1, goes to step 2.2, until isolating one Intrinsic mode functions circulation terminates, and goes to step 2.9；

Wherein ε ∈ [0.2,0.3]

Step 2.9, from r_ii-1(t) intrinsic mode functions obtained by subtracting in step 2.7, obtain r_ii(t), r_ii(t)= r_ii-1(t)-C_ii(t)；

Step 2.10, r is judged_ii(t) whether the number of extreme point is less than two, is decomposed if extreme point number is less than two Terminate；If extreme point number is more than or equal to two, ii=ii+1 goes to step 2.2.

Breathing, heartbeat signal Hilbert conversion sections be by breathe, heartbeat signal each time point frequency change table Show to come, the detailed process of step 3 is as follows：

Step 3.1, the intrinsic mode functions C to being obtained in step 2.7_ii(t) Hilbert conversion is carried out, wherein Hilbert becomes It is changed toWherein H represents that Hilbert is converted, and p represents Cauchy's principal value；

Step 3.2, analytical function z is generated by the intrinsic mode functions isolated_i(t),

z_ii(t)=C_ii(t)+iH[C_ii(t)]=a_ii(t)e^iφ(t)

The imaginary part of l representative functions, a_ii(t)=[C² _ii(t)+H²(C_ii(t))]^1/2It is magnitude function,It is phase function；

Step 3.3, it is that can obtain instantaneous frequency to phase function derivation

With reference to Fig. 2, the data that non-contact vital sign patient monitor is collected carry out contact transformation anyway and obtained just Hand over demodulated signal as shown in Fig. 2 bottom figures, specific signal separation process flow chart is as shown in Figure 1.

With reference to Fig. 3, quadrature demodulated signal x (t) carries out frequency diagram obtained by FFT as shown in fig. 3, it was found that heartbeat is believed Number flooded by the triple-frequency harmonics of breath signal, the heartbeat signal in quadrature demodulated signal now can not be identified with this method, wherein Data used are that respiratory rate is 18 beats/min (being 0.3Hz) in Fig. 3, and palmic rate is 66 beats/min and (is 1.1Hz)。

With reference to Fig. 4, Fig. 5, quadrature demodulated signal x (t) carries out signal separation process, can first divided from quadrature demodulated signal Heartbeat signal is separated out, breath signal is then isolated again.Middle graph in heartbeat signal spectrogram such as Fig. 4 is obtained after step 2 Shown in shape, palmic rate is intermediate pattern institute in 1.003Hz (60.18 beats/min), obtained breath signal spectrogram such as Fig. 5 Show, respiratory rate is 0.3Hz (i.e. 18 beats/min).

With reference to Fig. 4, Fig. 5, the breathing isolated to the first step and heartbeat signal carry out Hilbert conversion, obtain heartbeat letter Number, the instantaneous frequency of breath signal, last secondary picture in such as Fig. 4 and Fig. 5 can represent conversion of the signal frequency with the time Situation.

Claims (2)

1. a kind of signal processing algorithm of non-contact vital sign monitoring, it is characterised in that comprise the following steps：

Step 1, non-contact vital sign monitor system analog end sends orthogonal two-way heartbeat and breathing digital mixing signal, It is I roads signal all the way, another road is Q roads signal, carrying out arc tangent processing to I roads and Q roads obtains quadrature demodulated signal；

Step 2, quadrature demodulated signal is separated using Local Integral average Empirical Mode Decomposition Algorithm, successively believed heartbeat Number and breath signal separated from quadrature demodulated signal；

Step 3, Hilbert conversion is carried out to the breathing isolated in step 2 and heartbeat signal, obtains breathing and heartbeat signal Instantaneous frequency.

Wherein, the separating step of breathing and heartbeat signal is as follows in step 2：

Step 2.1, initialized, make r₀(t)=x (t), h₀(t)=r_ii-1(t), wherein x (t) is quadrature demodulated signal；

Step 2.2, ii=1, jj=1 are made；

Step 2.3, h is found out_jj-1(t) whole Local Extremum (t_k,x_k), wherein k is the index value of extreme value, k=0,1, 2 ..., t_kIt is the time corresponding to k-th of extreme point, x_kIt is the amplitude of k-th of extreme point；

Step 2.4, according to extreme point (t_k,x_k), to be divided into one section by signal decomposition between two extreme points, each section is all single Increasing or decreasing is adjusted, wherein kth section is signal in time t ∈ (t_k,t_k+1) on signal, be designated as C_k；

Step 2.5, each section of C is obtained using the method for integral mean_kLocal mean value point

$\left\{\begin{array}{c}{\stackrel{\‾}{x}}_k=\frac{1}{t_{k+1}-t_k}{\∫}_{t_k}^{t_{k+1}}x\left(t\right)dt\\ {\stackrel{\‾}{t}}_k=\frac{{\∫}_{t_k}^{t_{k+1}}t|x\left(t\right)-{\stackrel{\‾}{x}}_k{|}^{2}dt}{{\∫}_{t_k}^{t_{k+1}}|x\left(t\right)-{\stackrel{\‾}{x}}_k{|}^{2}dt}\end{array}\right.$

Step 2.6, the local mean value point to being calculatedCubic spline interpolation fitting is carried out, its average envelope is tried to achieve bent Line m_jj-1(t)；

Step 2.7, from signal h_jj-1(t) average envelope m is subtracted in_jj-1(t) h, is obtained_jj(t)=h_jj-1(t)-m_jj-1(t)；

Step 2.8, h is judged_jj(t) stop condition S whether is met_D, the h if meeting_jj(t) an eigen mode letter to isolate Count, now C_ii(t)=h_jj(t) step 2.9, is gone to；Otherwise, jj=jj+1, goes to step 2.2, intrinsic up to isolating one Modular function circulation terminates, and goes to step 2.9；

Wherein ε ∈ [0.2,0.3]

T is the duration of signal；

Step 2.9, from r_ii-1(t) intrinsic mode functions obtained by subtracting in step 2.8, obtain r_ii(t), r_ii(t)=r_ii-1 (t)-C_ii(t)；

Step 2.10, r is judged_ii(t) whether the number of extreme point is less than two, and knot is decomposed if extreme point number is less than two Beam；If extreme point number is more than or equal to two, ii=ii+1 goes to step 2.2.

2. a kind of signal processing algorithm of non-contact vital sign monitor system according to claim 1, its feature exists In the detailed process of step 3 is as follows：

Step 3.1, the intrinsic mode functions C to being obtained in step 2.8_ii(t) Hilbert conversion is carried out, wherein Hilbert is transformed toWherein H represents that Hilbert is converted, and p represents Cauchy's principal value；

Step 3.2, analytical function z is generated by the intrinsic mode functions isolated_i(t),

$z_{ii}\left(t\right)=C_{ii}\left(t\right)+iH\[C_{ii}\left(t\right)\]=a_{ii}\left(t\right)e^{{\mathrm{i\φ}}_{ii}\left(t\right)}$

The imaginary part of i representative functions, a_ii(t)=[C² _ii(t)+H²(C_ii(t))]^1/2It is magnitude function, It is phase function；

Step 3.3, it is that can obtain instantaneous frequency to phase function derivation