阅读说明：本技术 一种基于cmor小波的非接触式心率测量方法 (Non-contact heart rate measurement method based on CMOR wavelet ) 是由 李晓媛 武鹏 刘允 刘强 陈亚民 陈锋 李守豪 王娟娟 于 2019-10-30 设计创作，主要内容包括：本发明公开了一种基于CMOR小波的非接触式心率测量方法,首先以自然光为光源,录制一段60s的AVI格式的被试者正脸视频,对录制的视频进行检测,并将所检测的人脸区域用多边形框出、截取,设为重构人脸图像,然后选取眼睛、鼻子、嘴巴和脸部中心点为追踪特征点；将重构人脸图像转换到YCbCr空间进行皮肤检测；对录制的视频中除正脸外的其他帧依次记录Cg通道绿色浓度偏差；将Cg通道绿色浓度偏差值作为原始PPG信号进行带通滤波处理,滤除通带外的噪声信号,得到目标心率信号；将目标心率信号通过CMOR小波分析得到小波时频图。本发明可以准确、实时的测量心率,克服了以往波峰间隔算法误差较大以及傅里叶变换只能提取一段时间段内占主要成分心率参数的缺陷。(The invention discloses a CMOR wavelet-based non-contact heart rate measurement method, which comprises the steps of firstly taking natural light as a light source, recording a 60s AVI format front face video of a tested person, detecting the recorded video, framing and intercepting a detected face area by using a polygon, setting the framed and intercepted face area as a reconstructed face image, and then selecting eyes, a nose, a mouth and a face central point as tracking characteristic points; converting the reconstructed face image into an YCbCr space for skin detection; sequentially recording the Cg channel green concentration deviation of other frames except the front face in the recorded video; taking the Cg channel green concentration deviation value as an original PPG signal to perform band-pass filtering processing, and filtering noise signals outside a pass band to obtain a target heart rate signal; and analyzing the target heart rate signal by CMOR wavelet to obtain a wavelet time-frequency diagram. The method can accurately measure the heart rate in real time, and overcomes the defects that the conventional peak interval algorithm has large error and the Fourier transform can only extract the heart rate parameter which occupies the main component within a period of time.)

1. A non-contact heart rate measuring method based on CMOR wavelet is characterized in that: the method comprises the following steps:

firstly, recording a 60s AVI format human face video of a human subject by taking natural light as a light source, wherein the reading frame rate of a camera is 20fps, and the resolution is 800 multiplied by 600 pixels;

secondly, detecting the recorded face video of the tested person, framing and intercepting the detected face area by using a polygon, setting the face area as a reconstructed face image, and then selecting the center points of eyes, a nose, a mouth and a face as tracking feature points;

thirdly, converting the reconstructed face image into a YCbCr space for skin detection, wherein a conversion formula for converting RGB into the YCgCr color space is as follows;

according to the Asian skin color characteristics, the threshold values of the three channels of Y, Cb and Cr are shown as follows:

fourthly, sequentially recording the Cg channel green concentration deviation of other frames except the front face in the recorded video according to the methods of the second step and the third step;

fifthly, performing band-pass filtering processing on the Cg channel green concentration deviation value serving as an original PPG signal, wherein the pass band frequency is set to be 40/60Hz and 200/60Hz, and corresponds to the heart rate of 40-200 bpm; filtering noise signals outside the passband to obtain target heart rate signals;

and sixthly, analyzing the target heart rate signal through CMOR wavelet to obtain a wavelet time-frequency diagram.

2. The CMOR wavelet based non-contact heart rate measurement method of claim 1, wherein: during the recording of the video in said first step, the head position remains substantially unchanged, except for small-amplitude facial expressions and normal blinking movements.

3. The CMOR wavelet based non-contact heart rate measurement method of claim 1, wherein: in the third step, in order to extract the PPG signal with a high signal-to-noise ratio, the Cg channel, that is, the green concentration offset component, needs to be calculated, and the formula is as follows:

。

4. the CMOR wavelet based non-contact heart rate measurement method of claim 1, wherein: in the sixth step, in the wavelet transform, sub-waveletsPost-conjugation to PPG signalAnd (3) carrying out convolution to obtain a wavelet transformation coefficient:

in the formulaRepresenting conjugate, sub-waveletsBy mother or fundamental waveletsThe stretching and the translation are obtained as shown in the formula:

。

Technical Field

The invention relates to a non-contact heart rate measurement technology, in particular to a non-contact heart rate measurement method based on CMOR wavelets.

Background

With the continuous development of society, the living standard of people is gradually improved, so that the health problems are more and more noticed by people. The heart rate is an important parameter for measuring the human health index and can reflect the health level of the human body. In recent years, cardiovascular diseases, hypertension, coronary heart disease, diabetes and the like continue to come, so that the health of people is greatly damaged, and the psychology of people is greatly disturbed, therefore, the prevention and the control of diseases are very important for our lives, the occurrence of various cardiovascular diseases can be effectively prevented by regularly measuring the heart rate, and the heart rate measurement method has irreplaceable effects on the health of people.

The existing heart rate measuring equipment mainly comprises a heart rate measuring instrument, a heart rate belt, a heart rate bracelet, an arterial sphygmomanometer and the like. However, these devices need to be in contact with a human body to measure the heart rate, the operation process is complex, some devices need to be installed by special persons, discomfort can be brought to a measurer when the devices are in contact with the skin for a long time, the devices are inconvenient, the patients suffering from skin burn can be increased, and particularly, certain difficulty exists in contact type measurement in heart rate measurement of newborns.

Some recent researches find that the heart rate can be extracted by analyzing skin color change in a face signal in a video, the method can measure the heart rate of a human body in a non-contact mode, and the network camera has the characteristic of low price, can meet the economical efficiency and practicability of measurement at the same time, and is particularly more suitable for measuring skin damage, newborns and the like under the condition that the skin damage, the newborns and the like are not easy to perceive. However, the existing non-contact heart rate measurement technology has the problem of large measurement error, and the reason is as follows: firstly, the signal intensity extracted through the change of the face color is very weak, the signal to noise ratio is low, the signal is easily interfered by the external environment (for example, the human face cannot be accurately tracked due to the change of the ambient light and the overlarge head motion amplitude), the quality of the target heart rate signal obtained by adopting a blind source analysis algorithm is poor, and the part of the signal of the target heart rate signal after the blind source separation cannot be determined; secondly, Fourier transform is carried out on the extracted heart rate signals, and only heart rate parameters which occupy main components can be observed for a period of time, so that the real-time detection requirement cannot be met; thirdly, the heart rate parameters are obtained by calculating a peak interval algorithm, but due to the existence of noise interference, the conditions of peak deviation, missing detection, multiple detection and the like can occur, and the accuracy of heart rate parameter measurement is seriously influenced.

Disclosure of Invention

The invention aims to provide a non-contact heart rate measuring method based on CMOR wavelet, which can accurately measure heart rate parameters of a testee in real time.

In order to achieve the purpose, the invention can adopt the following technical scheme:

the non-contact heart rate measuring method based on the CMOR wavelet comprises the following steps:

firstly, recording a 60s AVI format human face video of a human subject by taking natural light as a light source, wherein the reading frame rate of a camera is 20fps, and the resolution is 800 multiplied by 600 pixels;

secondly, detecting the recorded face video of the tested person, framing and intercepting the detected face area by using a polygon, setting the face area as a reconstructed face image, and then selecting the center points of eyes, a nose, a mouth and a face as tracking feature points;

thirdly, converting the reconstructed face image into a YCbCr space for skin detection, wherein a conversion formula for converting RGB into the YCgCr color space is as follows;

according to the Asian skin color characteristics, the threshold values of the three channels of Y, Cb and Cr are shown as follows:

；

fourthly, sequentially recording the Cg channel green concentration deviation of other frames except the front face in the recorded video according to the methods of the second step and the third step;

fifthly, performing band-pass filtering processing on the Cg channel green concentration deviation value serving as an original PPG signal, wherein the pass band frequency is set to be 40/60Hz and 200/60Hz, and corresponds to the heart rate of 40-200 bpm; filtering noise signals outside the passband to obtain target heart rate signals;

and sixthly, analyzing the target heart rate signal through CMOR wavelet to obtain a wavelet time-frequency diagram.

Wherein during the recording of the video in the first step the head position remains substantially unchanged except for small-amplitude facial expressions and normal blinking movements.

In order to extract a PPG signal with a high signal-to-noise ratio in the third step, the Cg channel, that is, the green concentration offset component, needs to be calculated, and the formula is as follows:

。

in the sixth step, in the wavelet transform, sub-waveletsPost-conjugation to PPG signalAnd (3) carrying out convolution to obtain a wavelet transformation coefficient:

in the formulaRepresenting conjugate, sub-waveletsBy mother or fundamental waveletsThe stretching and the translation are obtained as shown in the formula:

。

a clear heart rate energy spectrogram can be observed through a wavelet time-frequency diagram, and then the energy spectrogram can be converted into a clear visible heart rate curve diagram through Matlab programming.

The invention has the advantages that the heart rate can be accurately measured in real time, and the change of the skin color of the face of the testee can be accurately extracted by converting the acquired RGB image into the YCgCrCg color space; heart rate parameters changing along with time are extracted through a CMOR wavelet energy spectrum algorithm, and the defects that the error of a conventional peak interval algorithm is large and the heart rate parameters occupying the main components in a period of time can only be extracted through Fourier transform are overcome; specifically, the innovation of the invention is embodied in the following three aspects:

1. the innovation of the signal source is that the signal intensity extracted through the change of the face color is very weak, the signal-to-noise ratio is low, and the signal source is easy to be interfered by the external environment, the application combines the biological characteristic that hemoglobin on the surface of the skin has better absorption to the spectrum of the wavelength band of 510 ~ 590nm, and adopts a specific matrix group to convert the acquired signals of R, G, B three channels into YCgCrCg color space to be used as the signal source, thereby being capable of accurately extracting the weak skin color conversion information of the face, overcoming the influence of the environment light on the skin color of the face, and being capable of accurately tracking the face when the head moves left and right, and back and forth;

2. innovation of heart rate extraction algorithm: different from high-intensity heart rate signals extracted by contact equipment, the signals extracted by a common camera are weak and have different amplitudes, the heart rate signals continuously transformed along with time are extracted by improved wavelet transformation, and heart rate parameters of a testee can be accurately measured in real time;

3. innovation of real-time processing: the non-contact heart rate measurement in the past is that a section of video is recorded through a camera at first, then corresponding heart rate parameter is obtained by analyzing the video, and a real-time processing algorithm is designed through analyzing a camera read-in image mechanism, so that the requirement on computer hardware is low, and real-time measurement can be realized on a common computer.

Drawings

Fig. 1 is a diagram showing a real environment when heart rate is extracted in the present invention.

Fig. 2 is a face test chart in the present invention.

Fig. 3 is a diagram of the result of skin color detection of a human face.

Fig. 4 is a route block diagram of the heart rate extraction technique.

Fig. 5 is a video energy spectrum and heart rate curve generated after wavelet analysis.

Fig. 6 is a graph comparing the results of the video heart rate measurement and the pulse oximeter measurement in the resting state.

FIG. 7 is a graph of heart rate measurements for the method of the present invention.

Detailed Description

The present invention is described in more detail below with reference to specific measurement examples to facilitate understanding by those skilled in the art. The human subject shown in the figure is a simulated character.