WO2023113246A1 - Device and method for extracting number of breathes of driver using motion correction - Google Patents

Abstract

The present invention relates to a device and method for extracting the number of breathes of a driver using motion correction. The present invention provides a method for extracting the number of breathes, the method comprising the steps of: receiving a signal transmitted via a radar and then reflected off a target; fast Fourier transforming an intermediate frequency signal generated by using a transmission signal and a reception signal of the radar; calculating a motion indicator of the target at the current point by using a deviation between a fast Fourier transformation signal of the current point and a fast Fourier transformation signal of a preceding point; correcting a phase signal, included in the fast Fourier transformation signal of the current point, by using the motion indicator; and extracting the number of breathes of the target with respect to the current point by using the corrected phase signal. According to the present invention, a respiratory signal can accurately be extracted and the number of breathes of the driver can accurately be estimated, by correcting a distortion, in the respiratory signal, which occurs when the driver moves.

Description

Device and method for extracting driver's respiratory rate through motion compensation

The present invention relates to an apparatus and method for extracting a driver's respiratory rate through motion compensation, and more particularly, to a motion correction capable of more accurately extracting a driver's respiratory rate by correcting a distorted breathing signal generated in a driver's movement section. It relates to a driver's respiratory rate extraction device and method through

As driving occupies a large part of daily life, the incidence of traffic accidents due to various factors is increasing. Driver breathing abnormalities such as apnea or hyperventilation and drowsy driving are one of the major causes of traffic accidents every year. According to a recent study, about 20% of traffic accidents are caused by drowsiness, which accounts for a large proportion of all traffic accidents. Large-scale accidents can be prevented if the driver's breathing abnormality and drowsy driving can be detected in advance and measured in real time. In this regard, a highly accurate driver biometric monitoring study is required.

Existing driver bio-signal monitoring methods monitor factors outside the body by detecting driver's eyelid movements, head nods, electromyography (EMG) and electrocardiography (ECG) through cameras and motion sensors inside the vehicle. However, the existing method has difficulty in directly recognizing the driver's movement in a low-light environment such as night driving. In addition, the method of using the EMG and ECG sensors has problems such as requiring physical contact between the driver's skin and the sensor or occupying a large space inside the vehicle for installing the sensor.

Bio-signal monitoring using FMCW (Frequency Modulated Continuous Wave) radar is a non-contact method using electromagnetic waves and has the advantage of low power consumption and small packaging, thus solving the problems of the existing method. Existing methods for monitoring the driver's vital signs estimate the respiratory rate in a situation where the driver's condition is stable and the driver's movement is not considered. However, in a real driving environment, it is very difficult to accurately monitor vital signs due to external factors such as a driver's movement while driving.

FMCW radar is a telecommunication equipment using electromagnetic waves. In this case, the driver's movement causes irregular fluctuations in the radar signal, which reduces the accuracy of bio-signal monitoring using the FMCW radar. Therefore, in order to increase the accuracy of signal extraction, it is necessary to correct the breathing signal reflecting the driver's movement in the moving situation.

The background technology of the present invention is disclosed in Korean Patent Publication No. 10-2021-0001217 (published on January 7, 2021).

An object of the present invention is to provide an apparatus and method for extracting a driver's respiration rate through motion compensation capable of accurately extracting a driver's respiration rate by correcting distortion of a respiration signal due to movement in a driver's movement situation.

The present invention is a method for extracting a driver's breathing rate performed by a driver's breathing rate extraction device, comprising the steps of receiving a signal transmitted through a radar and reflected from a target, and generating a signal using a transmission signal and a reception signal of the radar. Fast Fourier transforming an intermediate frequency signal and extracting a phase signal from the fast Fourier transform signal, and using the deviation between the fast Fourier transform signal at the current time and the fast Fourier transform signal at the previous time, the motion index of the target at the current time is calculated. Calculating, correcting the phase signal extracted from the fast Fourier transform signal at the current time using the motion index, and extracting the respiratory rate of the target at the current time using the corrected phase signal. It provides a driver's respiratory rate extraction method comprising a.

를 아래의 수학식에 의해 산출할 수 있다.In addition, the calculating of the motion index may include the motion index at the current time point.

can be calculated by the equation below.

는 현재 시점의 고속 푸리에 변환 신호,

는 직전 시점의 고속 푸리에 변환 신호를 나타낸다.Here, k is a distance index corresponding to the distance resolution index of the radar, t is a current time point, t-1 is a previous time point,

is the fast Fourier transform signal at the current time,

represents the fast Fourier transform signal at the previous point in time.

In addition, the step of correcting the phase signal may include calculating a motion quantification value corresponding to the motion index of the current view using a sigmoid function that quantifies the motion index to a value between 0 and 1; and The method may include calculating a phase signal correction value at the current view using a motion quantification value at the current view, the phase signal at the current view, and a pre-corrected phase signal at a previous view.

을 아래의 시그모이드 함수를 이용하여 연산할 수 있다.In addition, in the step of correcting the phase signal, the motion quantification value

can be calculated using the sigmoid function below.

는 상기 움직임 지표,

는 상기 움직임 지표의 임계값을 나타낸다. here,

is the motion index,

represents the threshold value of the motion index.

In addition, in the sigmoid function, the output corresponding to the motion quantification value converges to 0 as the motion index increases above the threshold value, and the output converges to 1 as the motion index value decreases below the threshold value, and the output corresponding to the threshold value If equal, the output can be set to a value of 0.5.

을 결정할 수 있다.In addition, the step of correcting the phase signal is a phase signal correction value at the current time using the following equation.

can determine

는 상기 현재 시점의 움직임 정량화 값,

는 상기 현재 시점의 위상 신호,

는 상기 기 보정된 직전 시점의 위상 신호를 나타낸다. here,

is the motion quantification value at the current time point,

Is the phase signal at the current time,

denotes a phase signal at a time immediately before the pre-correction.

In addition, in the step of extracting the respiration rate, a window may be applied for each hour to the phase signal correction value calculated according to time, and the respiration rate of the target may be extracted for each time from a frequency component analyzed within the window.

In addition, the present invention, a signal acquisition unit for receiving a signal transmitted through a radar and reflected from a target, and a fast Fourier transform of an intermediate frequency signal generated using a transmission signal and a reception signal of the radar, and a fast Fourier transform signal A signal processing unit extracting a phase signal from the current view, a calculation unit calculating a motion index of the target at the current view using a deviation between the fast Fourier transform signal at the current view and the fast Fourier transform signal at the previous view, and the fast Fourier transform at the current view. Provided is a driver's breathing rate extraction device including a correction unit for correcting a phase signal extracted from a signal using the motion index, and an extraction unit for extracting a target's breathing rate for a current time point using the corrected phase signal. .

According to the present invention, the respiration signal can be accurately extracted by correcting the distortion of the respiration signal due to the driver's movement in the driver's movement situation, and the driver's respiration rate can be accurately estimated from the extracted respiration signal.

1 is a diagram showing the configuration of a driver's respiratory rate extraction device according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating a method for extracting a driver's respiratory rate using the apparatus of FIG. 1 .

3 is a diagram showing an example of a driver's breathing signal measured by FMCW radar.

4 is a diagram showing the accuracy of a motion index for detecting a driver's motion according to an embodiment of the present invention.

5 is a diagram illustrating a relationship function between a motion index and a motion quantification value according to an embodiment of the present invention.

6 is a diagram showing an example of correcting a distorted phase signal in a driver's movement section according to an embodiment of the present invention.

7 is a diagram showing the result of calculating the number of breaths per minute of a driver using a corrected phase signal according to an embodiment of the present invention.

8 is a diagram showing the specifications of an FMCW radar for an embodiment of the present invention.

9 is a diagram showing an in-vehicle test environment according to an embodiment of the present invention.

10 is a diagram showing respiration signals and respiration rate estimation results for driving situations (situations 1 and 2) in which the driver does not move in an embodiment of the present invention.

11 is a diagram showing respiration signals and respiration rate estimation results for driving situations (situations 3 and 4) accompanied by a driver's motion in an embodiment of the present invention.

12 is a diagram showing respiration signals and respiration rate estimation results for driving situations (situations 5, 6, and 7) accompanied by a driver's motion in an embodiment of the present invention.

Then, with reference to the accompanying drawings, an embodiment of the present invention will be described in detail so that those skilled in the art can easily practice it. However, the present invention may be implemented in many different forms and is not limited to the embodiments described herein. And in order to clearly explain the present invention in the drawings, parts irrelevant to the description are omitted, and similar reference numerals are attached to similar parts throughout the specification.

Throughout the specification, when a part is said to be "connected" to another part, this includes not only the case where it is "directly connected" but also the case where it is "electrically connected" with another element interposed therebetween. . In addition, when a certain component is said to "include", this means that it may further include other components without excluding other components unless otherwise stated.

The present invention relates to a method for extracting a driver's respiratory rate through motion correction, and proposes a method for more accurately extracting a driver's respiratory rate by correcting a distorted breathing signal according to a driver's movement.

1 is a diagram showing the configuration of a driver's respiratory rate extraction device according to an embodiment of the present invention.

As shown in FIG. 1, the respiratory rate extraction device 100 includes a signal acquisition unit 110, a signal processing unit 120, a calculation unit 130, a correction unit 140 and an extraction unit 150, and an FMCW radar. (10) may be further included. Here, the operation of each unit 110 to 150 and the flow of data between each unit may be controlled by a controller (not shown).

Radar is divided into several types according to the modulation scheme of the transmission and reception signals. In the embodiment of the present invention, the use of a frequency modulated continuous wave (FMCW) radar using component modulation in terms of frequency is exemplified. The FMCW radar 10 transmits a chirp signal using a linearly increasing frequency.

The signal acquisition unit 110 may receive and acquire a signal reflected from a target after transmission through the FMCW radar 10 . The FMCW radar 10 may transmit a linearly modulated frequency signal, receive a signal reflected from a target, and transmit the signal to the signal processing unit 120 . Here, the FMCW radar 10 may be embedded in a seat of a vehicle or installed in the front of the vehicle.

The signal processing unit 120 may generate an intermediate frequency (IF) signal using the transmitted signal and the received signal of the FMCW radar 10, perform Fast Fourier Transform (FFT), and obtain the fast Fourier transform signal from the signal. Phase signals can be extracted. The signal processing unit 120 may transmit the fast Fourier transform signal generated by time to the calculation unit 130 and may transmit the phase signal extracted by time to the correction unit 140 .

The calculator 130 may calculate a motion index of the target at the current time point by using a deviation between the Fast Fourier Transform signal at the current time point and the Fast Fourier Transform signal at the previous time point. The calculation unit 130 may calculate the movement index of the target according to time in this way and transfer it to the correction unit 140 .

The correction unit 140 may correct the phase signal included in the fast Fourier transform signal at the current time by using the motion index calculated by the operation unit 130 . The correction unit 140 may acquire the phase signal correction value every hour and transmit it to the extraction unit 150 .

The extraction unit 150 extracts the respiratory rate of the target at the current time point using the phase signal corrected by the correction unit 140 . Here, the extraction unit 150 may estimate the respiratory rate of the target for each hour based on the phase signal correction value obtained through the correction unit 140 every hour.

In order to accurately monitor vital signals, it is essential to correct distortion of radar signals caused by environmental factors. By correcting the driver's breathing signal, it is possible to accurately estimate the driver's respiratory rate.

Next, a method for extracting a driver's respiratory rate according to an embodiment of the present invention will be described in detail.

FIG. 2 is a diagram illustrating a method for extracting a driver's respiratory rate using the apparatus of FIG. 1 .

First, the signal acquisition unit 110 acquires the received signal reflected from the target after being transmitted from the FMCW radar 10 over time (S210).

Thereafter, the signal processing unit 120 generates an intermediate frequency signal (IF signal) using the transmission signal and the reception signal of the FMCW radar 10 and performs fast Fourier transform (S220).

를 추출할 수 있다. Here, the signal processing unit 120 mixes the radar transmission signal and the received signal through a mixer and passes the mixed signal through a low-pass filter to obtain an intermediate frequency signal as shown in Equation 1 below.

can be extracted.

는

로 정의된 chirp 신호의 선형 주파수 증가율,

는 FMCW 레이더(10)의 대역폭,

는 처프(chirp) 지속시간,

는 송신 및 수신 신호 사이의 지연 시간,

는 반송파의 주파수,

은 레이더 신호의 크기를 나타낸다.here,

Is

The linear frequency increase rate of the chirp signal, defined as

is the bandwidth of the FMCW radar 10,

is the chirp duration,

is the delay time between the transmitted and received signals,

is the carrier frequency,

represents the magnitude of the radar signal.

이므로, 수학식 1은 다음의 수학식 2와 같이 근사화될 수 있다.In addition, the transmitted and received radar signals operate within microseconds (μs). thus,

, Equation 1 can be approximated as Equation 2 below.

로 전환되며,

는

이고, c는 빛의 속도이다.The FMCW radar 10 estimates the distance to the target using the frequency and phase of the intermediate frequency signal. When the distance between the radar and the object changes, the time delay between the transmitted and received signals is affected by the phase component due to the change in the position of the object.

is converted to

Is

and c is the speed of light.

는 현재 레이더 신호의 위상 성분을 나타낸다.In addition, the intermediate frequency signal of Equation 2 is converted to Equation 3 below,

represents the phase component of the current radar signal.

Here, n is the sampling index of the chirp and d represents the distance from the target.

The result of fast Fourier transform of the intermediate frequency signal of Equation 3 is shown in Equation 4 below.

Here, N is the number of chirp samples, and k represents the range index corresponding to the range resolution index of the FMCW radar. The range index can be determined within the radar measurement range, and the measurement range can be predefined by product standards or preset by the user.

In general, biosignals such as respiration or heartbeat are detected by extracting a phase change from a signal obtained by fast Fourier transforming an intermediate frequency signal over time.

를 이용하여 아래 수학식 5와 같이 각 시간 별 위상 신호를 추출할 수 있고, 추출한 위상 신호를 보정부(140)로 전달할 수 있다. The signal processing unit 120 is a fast Fourier transform signal

A phase signal for each time can be extracted using Equation 5 below, and the extracted phase signal can be transmitted to the correction unit 140.

는 타겟이 존재한다고 가정한 대상 거리 범위(관심 거리 범위)에 해당한다. here,

Corresponds to the target distance range (interest distance range) assumed that the target exists.

3 is a diagram showing an example of a driver's breathing signal measured by FMCW radar.

3(a) is an image of the driver's IF signal (intermediate frequency signal), and the color bar on the right represents the amplitude of the IF signal. Time on the horizontal axis is the radar scan time, and n on the vertical axis represents the sampling time of the chirp.

The section where the driver's movement occurs is indicated as Movement at the top of the figure. The displacement difference per hour of the radar signal due to the driver's breathing may form an irregular waveform. It can be seen that the irregular distortion of the breathing signal occurred in the movement section, and it can be seen that the driver's breathing signal is regular in other parts.

Figure 4(b) shows the result of comparing the respiration signal derived from the IF signal and the reference respiration signal (Reference) derived from the reference sensor. For comparison with the reference respiration signal, the respiration signal was derived from the phase component of the radar signal obtained after fast Fourier transform processing of the IF signal. Here, the phase component was derived from a radar signal within a specific range. The specific range refers to the distance between the radar and the driver and was extracted using the Magnitude-Phase Coherency (MPC) technique. A baseline breathing signal was acquired through a pressure sensor worn on the abdomen.

As shown in (b) of FIG. 3 , it can be seen that the breathing signal (Radar) observed by the radar is greatly distorted compared to the reference breathing signal (Reference) in the section where the driver's movement occurs. Therefore, in order to accurately detect the driver's breathing, it is necessary to correct the driver's moving breathing signal. To this end, the embodiment of the present invention calculates the driver's movement index every hour and corrects the phase signal therefrom.

Referring back to FIG. 2 , after step S220, the calculation unit 130 calculates a motion index of the target at the current time point using the deviation between the fast Fourier transform signal at the current time point and the previous time point (S230).

를 아래의 수학식 6에 의해 산출할 수 있다.Here, the calculation unit 130 is a motion index at the current time.

Can be calculated by Equation 6 below.

는 현재 시점의 고속 푸리에 변환 신호,

는 직전 시점의 고속 푸리에 변환 신호를 나타낸다.Here, k is the distance index corresponding to the range resolution index of the radar, t is the current time point, t-1 is the previous time point,

is the fast Fourier transform signal at the current time,

represents the fast Fourier transform signal at the previous point in time.

로 정의하였다. In the embodiment of the present invention, it is assumed that the breathing signal in the movement section is a distorted signal, and in order to correct the radar signal irregularly distorted by the movement, the difference in magnitude of the radar signal between the current and the previous one is used as a movement indicator.

defined as

는 물체가 존재하는 특정 범위에 대해 시간상 직전 신호와의 크기 차이를 평균화하여 파생된 것을 알 수 있다.these movement indicators

It can be seen that is derived by averaging the magnitude difference from the previous signal in time for a specific range in which the object exists.

4 is a diagram showing the accuracy of a motion index for detecting a driver's motion according to an embodiment of the present invention.

Figure 4 (a) shows the driver's breathing signal and the driver's movement index value derived from a driving situation in which the driver's movement exists, and Figure 4 (b) shows an acceleration sensor worn on the driver's chest, right arm, and right foot. Indicates the driver's body acceleration value measured by .

The radar signal was measured through a radar installed inside the driver's car seat. In order to check the detection reliability of the motion index derived from the radar signal, the motion index derived from the radar signal according to an embodiment of the present invention is compared with the actual driver's acceleration value detected through the acceleration sensor.

Looking at the result of FIG. 4(b), it is possible to check the section in which the driver moved within the measurement time based on the variability of the measured acceleration value. In the section where the driver's movement occurred, irregular distortion occurred in the driver's breathing signal, as shown in the movement section (Movement) shown at the top of FIG. 4(a).

In addition, movements of arms and legs as well as the driver's chest, which is a direction in which the radar signal is transmitted, cause minute displacement differences in the driver's body, which may distort the signal. Accordingly, when (a) of FIG. 4 is compared with (b), it can be seen that the distortion of the breathing signal is caused not only by the driver's body but also by movements of the limbs necessary for driving. As shown in FIG. 4 , it was confirmed that the movement index reflects the driver's movement by matching the section where the driver's movement occurred with the variation of the movement index.

Referring back to FIG. 2 , after step S230, the correction unit 140 corrects the phase signal included in the fast Fourier transform signal at the current time by using the motion index (S240).

를 수학식 6의 움직임 지표

를 이용하여 보정함으로써 현재 시점에 대한 위상 신호 보정 값

을 획득할 수 있다.The correction unit 140 is the current phase signal shown in Equation 5.

The motion index of Equation 6

The phase signal correction value for the current point in time by correcting using

can be obtained.

에 대응하는 움직임 정량화 값

을 아래 수학식 7에 나타낸 시그모이드 함수를 이용하여 연산할 수 있다.To this end, first, the correction unit 140 is a motion index at the current time point.

Motion quantification value corresponding to

Can be calculated using the sigmoid function shown in Equation 7 below.

는 움직임 지표이고,

는 움직임 지표의 임계값을 나타내고,

는 기울기(경사도) 계수를 나타낸다. 이때, 임계값은 현재까지 전체 측정 시간 동안의 움직임 지표의 평균 값으로 설정될 수 있다. here,

is the movement index,

represents the threshold of the motion index,

denotes the slope (slope) coefficient. In this case, the threshold may be set as an average value of motion indicators for the entire measurement time up to now.

는 0과 1 사이의 값으로 정량화될 수 있다. According to the sigmoid function of Equation 7, the motion index

can be quantified as a value between 0 and 1.

5 is a diagram illustrating a relationship function between a motion index and a motion quantification value according to an embodiment of the present invention.

가 임계값 이상으로 높아질수록 움직임 정량화 값에 해당한 출력

이 0에 수렴하고 임계값 미만으로 낮아질수록 출력이 1에 수렴하고 임계값과 동일한 경우에는 출력이 0.5의 값으로 설정되는 것을 알 수 있다. As shown in FIG. 5, according to the function of Equation 7, the motion index

Output corresponding to the motion quantification value as the value increases above the threshold

It can be seen that the output converges to 1 as it converges to 0 and decreases below the threshold value, and the output is set to a value of 0.5 when it is equal to the threshold value.

The embodiment of the present invention corrects the distorted radar signal using the motion quantification value in order to detect breathing more accurately than the existing breathing detection method in the driver's motion situation.

과 현재 시점의 위상 신호

및 기 보정된 직전 시점의 위상 신호

를 이용하여 현재 시점의 위상 신호 보정값

을 아래의 수학식 8과 같이 연산할 수 있다. Afterwards, the correction unit 140 calculates the motion quantification value at the current point in time.

and the current phase signal

and the pre-corrected phase signal at the immediately preceding point in time

The phase signal correction value at the current time using

Can be calculated as in Equation 8 below.

는 현재 시점의 움직임 정량화 값,

는 현재 시점의 위상 신호,

는 기 보정된 직전 시점의 위상 신호를 나타낸다. here,

is the motion quantification value at the current time point,

is the current phase signal,

denotes a phase signal at the immediately preceding point of time that has been pre-corrected.

가 임계값 아래로 감소하면

가 1로 수렴하여 현재 신호의 위상 값

을 현재의 최종 위상 값으로 결정할 수 있고, 반대로

가 임계값 이상으로 증가하면

는 0으로 수렴하여 시간상 직전의 최종 보상된 위상 값

을 현재의 최종 위상 값으로 결정할 수 있다. According to this,

decreases below the threshold

converges to 1 and the phase value of the current signal

can be determined as the current final phase value, and conversely

increases above the threshold

converges to 0 and the final compensated phase value just before time

can be determined as the current final phase value.

Referring back to FIG. 2 , after step S240 , the extraction unit 150 extracts the respiratory rate of the target at the current time using the corrected phase signal as above (S250).

에 대해 매시간 별로 윈도우를 적용하고 윈도우 내에서 분석되는 주파수 성분으로부터 타겟의 호흡수를 시간 별로 추출할 수 있다.Here, the extraction unit 150 calculates the phase signal correction value over time by Equation 8.

It is possible to apply a window for each hour and extract the target's respiration rate for each hour from the frequency components analyzed within the window.

Here, of course, the window means a time window, and the target's respiration rate at the current time point is estimated by applying a window of a set time length to past data including the current time point for each hour and analyzing frequency components within the window. Of course, a sliding window method can be applied for real-time respiratory rate detection. In this case, a fast Fourier transform (FFT) may be used for frequency analysis.

에 윈도우를 적용하고 FFT 처리하여 주파수 성분을 분석한다.For example, the extractor 150 uses Equation 9 below to final correct the phase signal.

A window is applied to and FFT processing is performed to analyze the frequency component.

는

를 FFT 처리한 신호, M은 윈도우 길이,

는 시간,

은 주파수를 의미한다. here,

Is

A signal processed by FFT, M is the window length,

time,

means frequency.

In addition, by analyzing the frequency component from the FFT processing result, the respiratory rate of the target can be extracted.

)를 이전 시점(t-1)에서의 윈도우로부터 결정된 호흡 주파수 값(

)과 개별 비교한 후에 주파수 편차가 최소(min)인 피크 주파수를 현재 시점(t)에서의 호흡 주파수 값(

)으로 결정한다. Here, as shown in Equation 10 below, the extractor 150 performs a fast Fourier transform on the data on the window applied at the current point in time t and detects a plurality of peak frequencies (

) to the respiratory frequency value determined from the window at the previous time point (t-1) (

) and individual comparisons, the peak frequency with the minimum frequency deviation (min) is the respiratory frequency value (

) to determine

)을 다음의 수학식 11에 적용하여 현재 시점(t)에서의 타겟의 분당 호흡수(RR; Respiratory rate)를 계산할 수 있다.Then, the respiratory frequency value determined at the current time point (t) (

) can be applied to the following Equation 11 to calculate the respiratory rate (RR) of the target at the current time point (t).

집합에서 수학식 10를 통하여 이전 호흡 주파수인

와 가장 가까운 peak를 탐색하여 현재 시점의 호흡 주파수 값

을 결정하고, 추출한 호흡 주파수 값을 수학식 11을 통하여 분당 호흡수(RR)로 계산하여 호흡수를 최종 출력한다.It is assumed that human breathing does not suddenly increase or decrease. Therefore, having a peak value through FFT

In the set, the previous breathing frequency through Equation 10

Respiratory frequency value at the current time by searching for the peak closest to

is determined, and the extracted respiratory frequency value is calculated as the respiratory rate per minute (RR) through Equation 11 to finally output the respiratory rate.

6 is a diagram showing an example of correcting a distorted phase signal in a driver's movement section according to an embodiment of the present invention. 6 shows a process of correcting a phase signal distorted by motion and a change in a corresponding frequency component.

는 보정 전 위상 신호,

는 수학식 7에 의한 움직임 정량화 값,

는 수학식 8에 의한 보정 후 위상 신호를 나타낸다. 보정 전 위상 신호를 보면 운전자의 움직임 발생 구간에서 신호가 크게 왜곡된 것을 확인할 수 있다.First, (a) of FIG. 6 corresponds to a correction process of a distorted signal,

is the phase signal before correction,

Is the motion quantification value by Equation 7,

Represents the phase signal after correction by Equation 8. Looking at the phase signal before correction, it can be seen that the signal is greatly distorted in the driver's motion occurrence section.

는 0으로 수렴하고 이때 위상 성분은 이를 반영한 수학식 8을 통하여 보정될 수 있다. 보정 후 위상 신호를 보면 운전자의 움직임 발생 구간에서 위상 신호가 보정됨을 확인할 수 있다.In the case of an embodiment of the present invention, the phase signal may be corrected using the motion quantization value. due to driver movement

converges to 0, and at this time, the phase component can be corrected through Equation 8 reflecting this. Looking at the phase signal after correction, it can be confirmed that the phase signal is corrected in the driver's movement occurrence section.

6(b) shows frequency components of the driver observed at a point in time when the driver is not in a dynamic state in order to check the correction accuracy of the distortion signal. When the frequency extracted from the corrected driver's breathing signal is compared with the frequency of the non-corrected signal, it can be confirmed that it matches the frequency (Reference) of the actual breathing signal. As such, it can be confirmed that the driver's breathing signal distorted by the motion is corrected through the frequency component comparison.

7 is a diagram showing the result of calculating the number of breaths per minute of a driver using a corrected phase signal according to an embodiment of the present invention. The driver's respiratory rate was estimated by peak tracking using the frequency components of the driver's vital signs.

Figure 7 (a) shows the frequency components of the original phase signal and the distorted phase signal, (c) shows the result of estimating the respiratory rate from the distorted phase signal. As a result, the respiratory rate estimated in the movement including the driver's movement is less accurate than the reference respiratory rate. Here, the reference respiratory rate corresponds to the correct respiratory rate measured by the sensor.

Unlike the previous result, (b) of FIG. 7 shows the frequency component extracted from the phase signal corrected by the method proposed in Equation 8. The frequency value of the corrected phase signal is strongly confirmed at 0.35 to 0.4 Hz, and it can be seen that it is similar to the original signal. Figure 7 (d) shows the respiratory rate estimated using the peak tracking technique to verify the accuracy of the signal correction. It can be seen that the respiration rate estimated from the corrected phase signal coincides with the actual respiration rate and the estimation accuracy is high.

Hereinafter, performance test results of the driver's respiratory rate estimation technique according to an embodiment of the present invention will be described.

8 is a diagram showing the specifications of an FMCW radar for an embodiment of the present invention. All experimental procedures and recordings were performed using the FMCW radar (Bitsensing INC, Korea) shown in FIG. 8, and the proposed algorithm was implemented using the parameters shown in FIG. A radar of this specification can detect objects in front within a range of 0-3.187 m.

9 is a diagram showing an in-vehicle test environment according to an embodiment of the present invention. FIG. 9 shows an experimental environment for detecting driver's biological signals equipped with a car seat for an actual vehicle and equipment for realizing an actual driving situation. A baseline respiratory rate for comparison was measured using a respiratory monitoring belt sensor (Neulog Inc, Israel).

(a) of FIG. 9 shows the attachment position of the corresponding sensor, and the motion of the driver was measured using a motion capture device, Perception Neuron Studio (Noitom Inc, China). At this time, among 14 motion sensors attachable to the human body, acceleration values obtained from three sensors worn on the chest, right wrist, and right foot were used. The radar is installed inside the car seat so as to face the center of the driver's chest as shown in (b) of FIG. 7 .

In order to verify the accuracy of the algorithm proposed in the present invention, the driver's vital signals were monitored for the seven driving situations in Table 1 according to the driver's movement.

vehicle driving situation driver movement details has exist doesn't exist go straight 1 √ Normal breathing for 1 minute go straight 2 √ Normal and rapid breathing for 30 seconds each turn left √ 1. High intensity movements 2. Long movement times turn right √ 1. High intensity movements 2. Long movement times Combined Driving 1 √ √ Sudden stop while driving Combined Driving 2 √ √ Resume straight driving after stopping while driving lane change √ 1. Low intensity movements 2. Short movement times

Each experiment consisted of 1 minute in length, and was conducted while viewing driving images for each situation on the front display window shown in (a) of FIG.

10 is a diagram showing respiration signals and respiration rate estimation results for driving situations (situations 1 and 2) in which the driver does not move.

10 (a) and (b) are respiration signal measurement data for situations 1 and 2 (going straight 1 and 2) in Table 1. Since the driver does not move, the measured respiration signal is regular and the signal correction index The value is 1. Figure 10 (c), (d) shows the respiration rate estimation results for the previous two situations, the respiration rate estimated from the frequency component of the respiration signal using the peak value tracking for the two situations coincides with the reference respiration rate can confirm that

11 is a diagram showing respiration signals and respiration rate estimation results for driving situations (situations 3 and 4) accompanied by a driver's motion in an embodiment of the present invention.

(a), (b), (c), and (d) of FIG. 11 show the result of estimating the driver's breathing signal and respiratory rate for driving conditions of a direction change, that is, a left turn and a right turn, respectively. Here, the irregular distortion of the driver's breathing signal can be corrected by Equation 8 in the direction change situation. From the results of (c) and (d) of FIG. 11, it can be seen that the respiratory rate estimated from the driver's respiratory signal corrected according to the embodiment of the present invention has a high correlation with the reference respiratory rate measured by the actual sensor. there is.

12 is a diagram showing respiration signals and respiration rate estimation results for driving situations (situations 5, 6, and 7) accompanied by a driver's motion in an embodiment of the present invention.

12 shows the result of estimating the driver's respiratory signal and respiratory rate in a driving situation including complex movements such as deceleration, acceleration, and straight ahead and lane change. As shown in (a) of FIG. 12, a sudden stop causes a large movement of the driver's body. Therefore, in this case, the variability of the motion index is higher than that of other driving situations. In (b) of FIG. 12 , in a situation in which driving is resumed after being stopped while driving in a straight line, the movement index is relatively small because the driver's movement is smaller than the situation in which the driver suddenly stops. In the two driving situations mentioned above, the distorted breathing signal of the driver is corrected by detecting the section in which the driver is moving.

(c) of FIG. 12 shows that the breathing signal during a lane change is similar to a direction change situation, but is related to a relatively small movement of the driver's body. Although the distortion of the breathing signal is caused by the body movement, relatively little change in the movement index is observed because there is almost no change in the body movement when changing lanes.

From the results of (d), (e) and (f) of FIG. 12, it can be seen that in all cases, the respiratory rate estimated from the respiratory signal corrected by the present invention has a high correlation with the reference respiratory rate. In addition, these results show that the respiratory rate estimation technique proposed in the present invention accurately estimates the respiratory rate by correcting irregularly distorted signals in a moving driving situation.

The following describes the accuracy analysis results for the respiratory rate estimation technique according to an embodiment of the present invention. The average error and accuracy of the respiratory rate estimated from the corrected respiratory signals according to the driving situation of each driver were calculated. The accuracy represents the absolute difference between the reference respiratory rate and the respiratory rate estimated from the estimated respiratory signal, and the average error was estimated by the standard deviation of the respiratory rate.

Driving
situation target
number Error in estimated respiratory rate compared to baseline respiratory rate before correction after calibration P-value turn left One 2.778 ±5.201 0.374 ±0.489 0.0003 2 1.415 ±1.542 1.077 ± 1.292 3 1.210 ±1.588 1.091 ±1.380 mean ± standard deviation 1.801 ±2.777 0.847 ±0.987 turn right One 2.097 ± 1.688 1.369 ± 1.282 0.0106 2 1.234 ±1.767 0.936 ± 1.492 3 2.134 ±2.095 1.488 ±1.267 mean ± standard deviation 1.822 ±1.849 1.264 ± 1.347 complex
driving 1 One 1.541 ±2.645 1.062 ±1.338 0.0372 2 1.888 ±1.933 1.526 ±1.698 3 3.654 ±2.921 2.961 ±2.026 mean ± standard deviation 2.361 ±2.500 1.850 ±1.687 complex
driving 2 One 1.260 ±1.940 0.546 ±0.753 0.0075 2 1.426 ±2.468 1.185 ±1.935 3 1.240 ±1.947 0.694 ±0.843 mean ± standard deviation 1.309 ±2.188 0.808±1.177 lane
change One 2.536 ±2.188 1.012 ±1.047 0.0022 2 0.884 ±1.849 0.485 ±0.779 3 0.587 ±0.973 0.522 ±0.657 mean ± standard deviation 1.336 ±1.647 0.673 ±0.828

From the results of Table 2, it can be seen that the respiratory rate estimated from the corrected respiratory signal has a smaller error compared to the reference respiratory rate than the respiratory rate estimated from the respiratory signal before correction. The p-value is an index representing the confidence interval of the result, and when p-value < 0.05, it means that the two groups compared at a confidence level of about 95% remain significantly different. Here, the two groups compared represent the results of the respiratory rate extracted from the pre-correction signal and the respiratory rate extracted from the corrected signal, respectively. Since the p-value for each situation is less than 0.05, it can be seen that the difference between the two compared groups is significant at the 95% confidence level.

As described above, according to the present invention, the respiration signal can be accurately extracted by correcting the distortion of the respiration signal due to the driver's movement in the driver's movement situation, and the driver's respiration rate can be accurately estimated from the extracted respiration signal. In addition, the proposed signal correction technique can be usefully applied to monitoring bio-signals in all fields where motion exists during monitoring.

Although the present invention has been described with reference to the embodiments shown in the drawings, this is merely exemplary, and those skilled in the art will understand that various modifications and equivalent other embodiments are possible therefrom. Therefore, the true technical scope of protection of the present invention should be determined by the technical spirit of the appended claims.

Claims (14)

In the driver's breathing rate extraction method performed by the driver's breathing rate extraction device, Receiving a signal transmitted through a radar and then reflected from a target; fast Fourier transforming an intermediate frequency signal generated by using a transmission signal and a reception signal of the radar and extracting a phase signal from the fast Fourier transform signal; Calculating a motion index of a target at a current time point using a deviation between a Fast Fourier Transform signal at a current time point and a Fast Fourier Transform signal at a previous time point; correcting a phase signal extracted from the fast Fourier transform signal at the current point in time using the motion index; and and extracting a respiratory rate of a target at a current time point using the corrected phase signal. The method of claim 1, The step of calculating the motion index, Movement index at the current time above

Driver respiration rate extraction method calculating by the following equation:

Here, k is a distance index corresponding to the distance resolution index of the radar, t is a current time point, t-1 is a previous time point,

is the fast Fourier transform signal at the current time,

represents the fast Fourier transform signal at the previous point in time. The method of claim 1, The step of correcting the phase signal, calculating a motion quantification value corresponding to the current motion index using a sigmoid function that quantifies the motion index to a value between 0 and 1; and and calculating a phase signal correction value at the current time point using the motion quantification value at the current time point, the phase signal at the current time point, and the pre-corrected phase signal at the immediately preceding time point. The method of claim 3, The step of correcting the phase signal, The motion quantification value

Driver respiration rate extraction method using the sigmoid function below:

here,

is the motion index,

represents the threshold value of the motion index. The method of claim 3, The sigmoid function is As the motion index increases above the threshold, the output corresponding to the motion quantification value converges to 0, and as the motion index decreases below the threshold, the output converges to 1. When the motion index is equal to the threshold, the output becomes a value of 0.5 Driver respiration rate extraction method to be set. The method of claim 5, The step of correcting the phase signal, Phase signal correction value at the current time using the equation below

Driver respiration rate extraction method to determine:

here,

is the motion quantification value at the current time point,

Is the phase signal at the current time,

denotes a phase signal at a time immediately before the pre-correction. The method of claim 1, The step of extracting the respiratory rate, A method of extracting a driver's respiratory rate by applying a window every hour to a phase signal correction value calculated according to time and extracting the respiratory rate of the target by time from a frequency component analyzed within the window. a signal acquisition unit for receiving a signal transmitted through a radar and then reflected from a target; a signal processing unit for fast Fourier transforming an intermediate frequency signal generated using the radar transmission signal and reception signal and extracting a phase signal from the fast Fourier transform signal; a calculation unit calculating a motion index of a target at a current point of view using a deviation between a Fast Fourier Transform signal at a current point of view and a Fast Fourier Transform signal at a previous point in time; a correction unit correcting the phase signal extracted from the fast Fourier transform signal at the current time point using the motion index; and An apparatus for extracting a respiratory rate of a driver including an extraction unit for extracting a respiratory rate of a target at a current time point using the corrected phase signal. The method of claim 8, The calculation unit, Movement index at the current time above

Respiratory rate extraction device that calculates by the equation below:

Here, k is a distance index corresponding to the distance resolution index of the radar, t is a current time point, t-1 is a previous time point,

is the fast Fourier transform signal at the current time,

represents the fast Fourier transform signal at the previous point in time. The method of claim 8, The correction unit, Calculate a motion quantification value corresponding to the motion index at the current time by using a sigmoid function that quantifies the motion index to a value between 0 and 1, and then, Driver breathing rate extraction device for calculating a phase signal correction value at the current time point using the motion quantification value at the current time point, the phase signal at the current time point, and the pre-corrected phase signal at the immediately preceding time point. The method of claim 10, The correction unit, The motion quantification value

Driver respiration rate extraction device that calculates using the sigmoid function below:

here,

is the motion index,

represents the threshold value of the motion index. The method of claim 10, The sigmoid function is, As the motion index increases above the threshold value, the output corresponding to the motion quantification value converges to 0, and as the motion index decreases below the threshold value, the output converges to 1. When the motion index is equal to the threshold value, the output value is 0.5. Operator respiration rate extraction device to be set. The method of claim 12, The correction unit, Phase signal correction value at the current time using the equation below

Operator respiration rate extraction device to determine:

here,

is the motion quantification value at the current time point,

Is the phase signal at the current time,

denotes a phase signal at a time immediately before the pre-correction. The method of claim 8, The extraction part, A driver's respiratory rate extraction device for applying a window every hour to a phase signal correction value calculated according to time and extracting the target's respiratory rate for each hour from a frequency component analyzed within the window.

Images