CN105286845A - Movement noise elimination method suitable for wearable heart rate measurement device - Google Patents

Abstract

The invention discloses a motion noise elimination method suitable for a wearable heart rate measuring device, so as to improve the accuracy of the heart rate measurement value of the wearable heart rate measuring device. In this method, the wearable heart rate measurement device collects multiple photoplethysmogram signals and motion acceleration signals of the user in the same time period; The spectral peak positions of the spectrum are relatively aligned, and multiple photoplethysmography signal spectra with motion noise removed can be obtained by spectral subtraction; finally, the position of the heart rate frequency point is accurately located according to the spectral peak tracking mechanism. The invention effectively eliminates the motion noise in the heart rate, solves the situation of no peak, multiple peaks and target spectrum peaks being lost in the spectrum of multiple photoplethysmography signals after spectral subtraction, and realizes the wearable device-based Accurate measurement of real-time heart rate.

Description

A Motion Noise Cancellation Approach for Wearable Heart Rate Measurement Devices

technical field

The invention relates to the field of information processing, in particular to a motion noise elimination method suitable for wearable heart rate measurement equipment.

Background technique

Heart rate refers to the number of times the human heart beats per minute, subject to the first sound. In the detection of human body parameters, heart rate is an important physiological indicator, which provides a reference for medical diagnosis. At the same time, the heart rate can also be used as an objective evaluation index of the physiological load of human exercise. With the rise of wearable smart devices such as smart watches, smart wristbands, and smart bracelets, and people's emphasis on health, the method of monitoring heart rate based on photoplethysmography signals has been widely accepted by industry and academia. focus on.

Because the photoplethysmography signal often contains motion noise, it is difficult to accurately measure the heart rate, and corresponding denoising technology is required. At present, many techniques for removing motion noise have been proposed, such as independent component analysis (ICA), wavelet denoising method, adaptive filter denoising method (ANC), classical mode decomposition method (EMD), etc. are widely used. However, the above algorithm is mainly aimed at moderate or non-violent movements, such as hand movement, walking, jogging (speed less than 8km/h). In the case of very strong motion noise, the effect of the above algorithm is not satisfactory.

The present invention proposes a method for eliminating strong motion noise. In this method, the frequency positions of the motion noise in multiple photoplethysmogram signal spectra are aligned with the frequency positions of the motion acceleration signal spectrum, and the spectral peaks of motion noise can be easily subtracted from multiple original photoplethysmography signal spectra by using spectral subtraction , to obtain multiple clean photoplethysmography signal spectra. At the same time, this method proposes a spectral peak tracking mechanism, which can deal with the cases of no peak, multiple peaks and target spectral peaks being lost in the spectrum of multiple photoplethysmography signals after spectral subtraction. The invention effectively eliminates motion noise in the heart rate, and realizes accurate measurement and calculation of the real-time heart rate based on the wearable device.

Contents of the invention

The technical problem to be solved by the present invention is how to provide an effective method for removing motion noise when the motion noise is very strong, so as to obtain an accurate real-time heart rate value.

In order to solve the above technical problems, the present invention provides a motion noise elimination method suitable for wearable heart rate measurement equipment, including two parts of spectral subtraction and spectral peak tracking mechanism, characterized in that:

The wearable heart rate measuring device collects a plurality of photoplethysmography signals and motion acceleration signals in the same time period at the user's wrist; then, uses the spectral subtraction to remove motion noise in the plurality of photoplethysmography signals ;Finally, according to the spectrum peak tracking mechanism, accurately locate the position of the heart rate frequency point.

The method comprises the steps of:

The wearable heart rate measuring device collects a plurality of photoplethysmography signals and motion acceleration signals of the user within the same time period; performs down-sampling processing on the above-mentioned multiple photoplethysmography signals and motion acceleration signals; and then downloads The above-mentioned signal after sampling is subjected to a band-pass filtering operation.

The spectral subtraction method effectively removes the motion noise signal according to the strong correlation performance between the motion acceleration signal and the motion noise signal in the plurality of photoplethysmography signals, and obtains a plurality of pure photoplethysmography signal spectra; Each sub-stage of the spectrum peak tracking mechanism processes the above-mentioned multiple pure photoplethysmography signal spectrums to locate the position of the user's heart rate frequency point.

Preferably, the wearable heart rate measuring device is embedded with multiple photoplethysmography sensors and three-axis accelerometers; the multiple photoplethysmography sensors collect multiple photoplethysmography signals of the user; the three The axial accelerometer collects the motion acceleration signal of the user in the same time period.

Preferably, the motion noise in the plurality of photoplethysmography signals has more of the same frequency as the synchronous motion acceleration signal, so that the frequency position of the motion noise in the frequency spectrum of the plurality of photoplethysmography signals is the same as that of the synchronous motion acceleration signal. The frequency positions of the motion acceleration signal spectrum are aligned, and the spectral peaks of the motion noise can be easily subtracted from the multiple photoplethysmography signal spectrums by using the spectrum subtraction method to obtain multiple clean photoplethysmography signal spectrums.

Preferably, in order to ensure that the spectral subtraction is effective, the plurality of photoplethysmographic signal spectra and the motion acceleration signal spectrum need to undergo an energy normalization operation before performing the spectral subtraction.

Preferably, the spectral peak tracking mechanism is mainly based on the following two principles: first, in most cases, the maximum spectral peak position in the spectrum of the multiple photoplethysmographic signals corresponds to the heart rate spectral peak position; second, The center rates in most of the overlapping continuous windows are very close; the spectrum peak tracking mechanism consists of three sub-stages of initialization, spectrum peak selection and spectrum peak discovery.

Compared with the prior art, the technical solution provided by the present invention effectively eliminates the motion noise in the heart rate, and solves the problems of no peak, multiple peaks and target spectral peaks in the spectrum of multiple photoplethysmography signals after spectral subtraction The situation of being followed. Therefore, the accuracy of the heart rate measurement value of the wearable heart rate measuring device is improved, and the real-time heart rate measurement and calculation based on the wearable device are realized.

Description of drawings

1 is a schematic flow chart of a motion noise elimination method according to an embodiment of the present invention;

Fig. 2 is a schematic flow chart of spectral subtraction according to an embodiment of the present invention.

detailed description

The implementation of the present invention will be described in detail below in conjunction with the accompanying drawings and examples, so as to fully understand and implement the implementation process of how to apply technical means to solve technical problems and achieve corresponding technical effects in the present invention.

In the technical solution of the present invention, the frequency positions of the motion noise in the multiple photoplethysmogram signal spectra are aligned with the frequency positions of the motion acceleration signal spectrum, and the motion acceleration signal spectrum can be easily extracted from the original photoplethysmography signal spectrum. The spectral peaks of motion noise are subtracted to obtain multiple clean photoplethysmographic signal spectra. At the same time, this method proposes a spectral peak tracking mechanism, which can deal with the cases of no peak, multiple peaks and target spectral peaks being lost in the spectrum of multiple photoplethysmography signals after spectral subtraction. This technical solution effectively eliminates motion noise in heart rate, and realizes accurate measurement and calculation of real-time heart rate based on wearable devices.

Embodiment 1. Motion noise elimination method of wearable heart rate measuring device

FIG. 1 is a schematic flowchart of the motion noise elimination method of this embodiment, and FIG. 2 is a schematic flowchart of the spectral subtraction method of this embodiment.

The present embodiment shown in Fig. 1 is the overall process of the motion noise elimination method of the wearable heart rate measuring device, which mainly includes the following steps:

Step S210, the wearable heart rate measuring device uses two photoplethysmography sensors distributed in different positions to collect two channels of photoplethysmography signals (hereinafter referred to as PPG ₁ and PPG ₂ ), and uses a three-axis accelerometer to collect simultaneous The motion acceleration signals of the three channels in the time period.

In step S220, the initial sampling frequency of the original signal is 125 Hz, and in order to reduce the amount of calculation, it is necessary to down-sample the original signal to a sampling frequency of 25 Hz.

In step S230, the above-mentioned down-sampled signal needs to be filtered by a second-order Butterworth filter with a passband of 0.4 Hz-4 Hz, so as to eliminate motion noise and other noise interference outside a certain frequency range.

Step S240, the frequency position of the motion noise in the two photoplethysmography signal spectra is aligned with the frequency position of the motion acceleration signal spectrum, and the two photoplethysmography signal spectra with motion noise removed can be obtained by spectral subtraction, that is, the clean PPG ₁ , PPG ₂ signal spectrum.

In this step, typically, the specific steps of spectral subtraction are as shown in Figure 2:

Step S310, for each frequency point f _i (i=1, . . . , N), select the largest spectral coefficient from the motion acceleration signal spectra of the three channels, which is defined as C _i .

In step S320, C _i is subtracted from the spectral coefficients of the signal spectrums of PPG ₁ and PPG ₂ at each frequency point f _i (i=1, ..., N), and after the above processing, the signal spectrum of PPG ₁ is within 0≤f The maximum value of the spectral coefficient within the range of _i ≤ 199 is defined as p _max1 , and the maximum value of the spectral coefficient of the PPG ₂ signal spectrum within the range of 0 ≤ _{f i} ≤ 199 is defined as p _max2 .

Step S330, PPG ₁ signal spectrum in the range of 0 ≤ _{f i} ≤ 199 with a spectral coefficient less than p _max1 /4 is set to 0, PPG ₂ signal spectrum in the range of 0 ≤ _{f i} ≤ 199 with a spectral coefficient less than p _max2 /4 Both are set to 0.

In step S340, the clean photoplethysmography signal spectrum of two different channels is obtained after the above operations, and the spectrum peak tracking mechanism is started at this time, so as to accurately locate the position of the heart rate frequency point.

In order to better describe the spectral subtraction method, the following points are explained:

First, the frequency point f _i (i=1, ..., N) of the digital signal spectrum starts from 0, and the relationship between it and the position index i is shown in formula (1),

f _i =i-1(1)

Second, the relationship between the frequency point fi of the digital signal spectrum and the frequency _f of the analog signal is shown in formula (2),

$\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{\mathrm{<span\; class="notranslate">\; N</span>}}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; N</span>}}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

Among them, f _s is the sampling frequency, and N is the sampling point;

Third, the highest heart rate recorded by humans is 230 beats/minute, and in most cases (including strenuous exercise) the heart rate is lower than 180 beats/minute; f _s =25Hz is set in this embodiment, and N=1024, so the described Spectrum subtraction only analyzes the two photoplethysmography signal spectra within the range of 0≤f _i ≤199;

Fourth, in order to ensure that the spectral subtraction is effective, it should be noted that the two photoplethysmographic signal spectra and the motion acceleration signal spectrum need to pass the energy normalization operation before the spectral subtraction processing.

Step S250, after the above operations, two clean photoplethysmography signal spectra are obtained, and then the user's heart rate frequency point is located using the spectrum peak tracking mechanism.

In this step, typically, the peak tracking mechanism consists of three parts:

1) Initialization: The user needs to minimize hand movements in the first few seconds to ensure the accuracy of the initial heart rate frequency point position. In this embodiment, the position of the largest peak in the PPG ₁ signal spectrum is selected as the corresponding heart rate peak Location;

2) Spectral peak selection: use the spectral peak position corresponding to the center rate of the previous time window to find the corresponding spectral peak of the two photoplethysmography signal spectrum center rates in the current time window;

In practical applications, there will be some extreme cases. For example, no peak or multiple peaks appear in the spectrum of the two photoplethysmography signals after spectral subtraction. Among them, no peak means that both photoplethysmography signal spectrums have no spectral peak corresponding to heart rate or only one photoplethysmographic signal spectrum contains a spectral peak corresponding to heart rate; multi-peak refers to two photoplethysmography signal spectrums There are multiple spectral peaks near the spectral peak position corresponding to the spectral center rate or there are multiple spectral peaks near the spectral peak position corresponding to the spectral central rate of only one photoplethysmography signal.

3) Spectrum peak discovery: This stage can effectively prevent the situation that the spectral peak corresponding to the heart rate is lost due to abnormal conditions in multiple continuous time windows.

The spectral peak tracking mechanism in this embodiment can effectively deal with no peaks, multiple peaks, and target spectral peaks being lost in the spectrum of the two photoplethysmography signals after spectral subtraction.

In step S260, after the above steps are processed, the user's real-time heart rate value can be output.

In this embodiment, the wearable heart rate measurement device is embedded with two photoplethysmography sensors and a three-axis accelerometer, and collects two photoplethysmography signals and motion acceleration signals within the same time period at the wrist of the user; The peak position of the motion noise in the two photoplethysmogram signal spectra is aligned with the peak position of the motion acceleration signal spectrum, and the two photoplethysmography signal spectra with motion noise removed can be obtained by using spectral subtraction; finally, according to the spectral peak The tracking mechanism can precisely locate the position of the heart rate frequency point. This method effectively eliminates the motion noise in the heart rate, and solves the situation of no peak, multiple peaks and target spectral peaks being lost in the spectrum of the two photoplethysmography signals after spectral subtraction. Therefore, the accuracy of the heart rate measurement value of the wearable heart rate measuring device is improved, and the real-time heart rate measurement and calculation based on the wearable device are realized.

Although the embodiments disclosed in the present invention are as above, the above content is only an embodiment adopted for easy understanding of the present invention, and is not intended to limit the present invention. Under the premise of not departing from the spirit and scope disclosed in the present invention, any modifications and changes can be made in the form and details of the implementation, but the patent protection scope of the present invention must still be defined by the appended claims range prevails.

Claims (5)

1. be applicable to a motion artifacts removing method for wearable heart rate measuring device, comprise spectrum-subtraction and spectrum peak follow-up mechanism two parts, it is characterized in that:

Described wearable heart rate measuring device gathers with the multiple photoplethysmographic signal in the time period and acceleration of motion signal at user's wrist place; Then, described spectrum-subtraction is utilized to remove motion artifacts in multiple photoplethysmographic signal; Finally, heart rate frequency point position is accurately located according to described spectrum peak follow-up mechanism;

The method comprises the steps:

Described wearable heart rate measuring device gathers the multiple photoplethysmographic signal of user at one time in section and acceleration of motion signal; Down-sampling process is carried out to above-mentioned multiple photoplethysmographic signal and acceleration of motion signal; Then the above-mentioned signal after down-sampling is carried out bandpass filtering operation;

Described spectrum-subtraction removes described motion artifacts signal effectively according to the strong correlation performance of motion artifacts signal in described acceleration of motion signal and described multiple photoplethysmographic signal, obtains multiple pure photoplethysmographic signal frequency spectrum; Each sub stage of described spectrum peak follow-up mechanism processes above-mentioned multiple pure photoplethysmographic signal frequency spectrum, the heart rate frequency point position of consumer positioning.

2. the motion artifacts removing method being applicable to wearable heart rate measuring device according to claim 1, is characterized in that:

The embedded multiple photoplethysmographic sensor of described wearable heart rate measuring device and three axis accelerometer; Multiple photoplethysmographic signal of described multiple photoplethysmographic sensor acquisition user; Described three axis accelerometer gather user at the same time between acceleration of motion signal in section.

3. the motion artifacts removing method being applicable to wearable heart rate measuring device according to claim 1, is characterized in that:

In described multiple photoplethysmographic signal, motion artifacts has more identical frequency with synchronous described acceleration of motion signal, the frequency location of motion artifacts in described multiple photoplethysmographic signal frequency spectrum is alignd with the frequency location of described acceleration of motion signal spectrum, utilize described spectrum-subtraction easily can deduct the spectrum peak of motion artifacts from described multiple photoplethysmographic signal frequency spectrum, obtain multiple clean photoplethysmographic signal frequency spectrum.

4. the motion artifacts removing method being applicable to wearable heart rate measuring device according to claim 3, is characterized in that:

For guaranteeing that described spectrum-subtraction is effective, described multiple photoplethysmographic signal frequency spectrum and described acceleration of motion signal spectrum needed to be operated by energy normalized before carrying out described spectrum-subtraction.

5. the motion artifacts removing method being applicable to wearable heart rate measuring device according to claim 1, is characterized in that:

Described spectrum peak follow-up mechanism is mainly based on following two principles: the first, and in described multiple photoplethysmographic signal frequency spectrum, maximum spectrum peak position is corresponding with heart rate spectrum peak position in most cases; The second, the continuous window center rate overlapping in major part is very close; Described spectrum peak follow-up mechanism is selected and composes peak to find that three sub stages form by initializing, composing peak.