WO2013060249A1 - 基于视频图像的非接触式生命体征的检测方法及检测系统 - Google Patents

基于视频图像的非接触式生命体征的检测方法及检测系统 Download PDF

Info

Publication number
WO2013060249A1
WO2013060249A1 PCT/CN2012/083181 CN2012083181W WO2013060249A1 WO 2013060249 A1 WO2013060249 A1 WO 2013060249A1 CN 2012083181 W CN2012083181 W CN 2012083181W WO 2013060249 A1 WO2013060249 A1 WO 2013060249A1
Authority
WO
WIPO (PCT)
Prior art keywords
signal
frequency
vital sign
detection
module
Prior art date
Application number
PCT/CN2012/083181
Other languages
English (en)
French (fr)
Inventor
赵芳
钱卓
俞雅萍
Original Assignee
西双版纳大渡云海生物科技发展有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 西双版纳大渡云海生物科技发展有限公司 filed Critical 西双版纳大渡云海生物科技发展有限公司
Publication of WO2013060249A1 publication Critical patent/WO2013060249A1/zh

Links

Classifications

    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00Measuring for diagnostic purposes; Identification of persons
    • A61B5/02Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
    • A61B5/0205Simultaneously evaluating both cardiovascular conditions and different types of body conditions, e.g. heart and respiratory condition
    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00Measuring for diagnostic purposes; Identification of persons
    • A61B5/02Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
    • A61B5/024Detecting, measuring or recording pulse rate or heart rate
    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00Measuring for diagnostic purposes; Identification of persons
    • A61B5/08Detecting, measuring or recording devices for evaluating the respiratory organs
    • A61B5/0816Measuring devices for examining respiratory frequency
    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B2576/00Medical imaging apparatus involving image processing or analysis
    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16HHEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H30/00ICT specially adapted for the handling or processing of medical images
    • G16H30/40ICT specially adapted for the handling or processing of medical images for processing medical images, e.g. editing

Definitions

  • Non-contact vital sign detection method based on video image and leak detection system
  • the invention relates to a home health care system, a medical monitoring system and an animal health detecting system, in particular to a method and a system for detecting non-contact vital signs based on a video circle image, for detecting heart rate, breathing and human body and/or animal Heart background technology
  • Non-contact Vital signs (including heart rate, call and so on) Detection methods and testing facilities due to its convenience, safety and 3 ⁇ 4 activity, have been widely concerned by the industry! Before the non-contact heart rate and respiratory testing technology ⁇ It should include the following Doppler radar detection and serial image acquisition. Doppler radar non-contact measurement 3 ⁇ 4 method detects the human body representation caused by heartbeat and respiration, and then obtains the human heart rate and respiratory information.
  • the Doppler radar detection system is composed of a controller, an ultra-wideband signal transmitter, and a receiver that detects and receives a return-to-zero signal, and the electromagnetic wave emitted by the transmitter is irradiated to the human body, and is subjected to Doppler modulation of the surface of the human body;
  • the machine performs coherent processing on the backscattered echo, detects the Doppler information in the echo, and finally obtains the heartbeat and respiratory information of the human body through signal processing techniques such as amplification and filtering.
  • the Doppler radar detection method is based on the perception of the body surface, the movement, i is extremely sensitive to body movements and changes in the surrounding environment.
  • the Doppler radar type wine detection system is complex in construction and expensive. , ⁇ It is used for special occasions such as earthquake disasters, landslide accidents, emergency rescues, and police hostage rescue. It is not suitable for normal life.
  • the blood in the skin absorbs the beam and reduces it, and the amount of attenuation depends on the amount of blood volume s in the heart: the volume changes *, the amount of attenuation of the beam 4!
  • the current fluctuations Through the image
  • a method of heart rate loss detection based on infrared sequence images has been proposed in China.
  • Weng first collects the subject image through the infrared dynamic imaging collection system, manually intercepts the subject's temple image, and extracts the time series signal by the two spirits method, passing the first-order and low-pass filtering.
  • a series of steps such as power spectrum analysis of AR model to achieve non-destructive testing of heart rate «
  • the invention overcomes the defects in the structure of the device for detecting the eclipse vital sign detection system in the prior art, the cost is expensive, is not suitable for daily life, and the detection method!
  • the real-time measurement, the denoising and the anti-noise ability cover, the detection time cannot be performed. Long. It does not reflect defects such as real-time changes in vital signs.
  • the detection method and system of the present invention have the advantages of fast catch, real-time, multi-target simultaneous spattering, continuous spattering, robustness, low cost, and wide application range.
  • the invention provides a method for detecting a contact vital sign which is suitable for video concealing, comprising: - a first step, continuously collecting a video image of a target to be measured according to a national frame rate, and automatically detecting an ROL region in the image, In the second step, the multi-channel signal divided by the R0!
  • 3 ⁇ 4 domain separates the 3 ⁇ 4 vital sign signal
  • Step extracting the frequency of the signal of the vital sign and converting the frequency into the vital sign, obtaining a detection>, the frequency of the signal of the vital sign including the frequency of the heartbeat signal and the respiratory signal
  • the frequency- f is characterized by the heart rate, the respiratory rate ⁇ : a, the vital sign is the heart rate, the Ron domain is the target area to be tested: the target skin area; the life sign is the respiratory frequency, the R (3 ⁇ 4) The area is the chest or abdomen position of the target to be tested.
  • the detection method of the invention simultaneously detects a plurality of targets to be detected.
  • the detection method of the present invention extracts the frequency of the vital sign signal with a minimum of test time of 23 signals.
  • Separating the ⁇ vital sign signal from the second step includes: stepping the image of the region by a plurality of times, calculating an average value between the ffi of each channel to form an original time series signal, and the original time series signal After the baseline drift is filtered and normalized, the independent component that is the same as the original time series signal is separated by the blind source separation method; wherein the independent component includes the vital sign signal and the click t .
  • the direct source separation method includes the maximum information amount method, ⁇ 1 gradient method adaptive method, fast independent element analysis method, and matrix eigenvalue decomposition method.
  • the strong disturbance signal is determined and eliminated according to the magnitude of the change of the brightness of the hidden image of the ROI region.
  • the frequency of extracting the vital sign signal from the beginning of the new detection step includes the following steps; The independent component ⁇ signal smoothing a, taking the independent component corresponding to the vital sign signal as the source signal component, and then extracting the frequency of the source signal component by the periodic signal frequency detecting method to obtain the vital sign information frequency.
  • the source signal component is the main energy of the correlation function in the independent component as the 'integral energy ratio 1; the smallest independent component S, or the independent component of the power spectral density peak in the independent component S
  • the cycle signal frequency detection method includes a bispectrum analysis method, a wavelet transform method, and a multi-a autocorrelation method.
  • the multi-spiring autocorrelation method in the present invention extracts a frequency of the source signal ⁇ component,
  • the source signal 3 ⁇ 4 component performs multiple autocorrelation operations for spectrum analysis, and the peak power point in the spectrum is the frequency of the vital sign signal.
  • the conversion of the frequency of the vital sign signal into the vital sign is obtained by using the formula! i ⁇ 60> / and Rs ⁇ Ox/i minutes m to obtain the heartbeat number HR and the number of breaths Rs per minute,
  • the invention also provides a detection system based on video circle image contact type vital signs, which comprises a video surrounding image «set module (i), automatic detection module (2), signal separation module (3), intelligent decision module (4, Signal frequency extraction module (5) and display module C6);
  • the video national image acquisition module (1) continuously collects images at a fixed frame rate and transmits the images to the motion detection module (2.);
  • the automatic detection module (2) automatically detects the number of objects to be measured in real time, and captures and tracks the R01 area; the signal separation module (3) realizes separation of vital signs and noises;
  • the intelligent decision module (4) judges according to the variation amplitude of the brightness of the RCM area and rejects the strong disturbance signal; the signal frequency extraction module (5) extracts the frequency from the vital sign signal, and It is converted into the vital signs;
  • the display module (6) updates the detection result of the anger frequency extraction module (5) in real time, and the invention provides a method for detecting ⁇ contact heart rate and/or breathing.
  • System fast, real-time, continuous monitoring, robust, low cost and wide range of applications, enabling heart rate and breathing! 3 ⁇ 4, real-time, continuous non-contact monitoring in heart beats
  • the blood volume of the internal arteries of the skin changes, so that the reflected light intensity of the skin changes volatility, and thus the brightness value of the image changes.
  • Breathing can cause the chest cavity of the human body to undulate, and at the same time, it will move the body parts such as the human body, the head and the like, so that the brightness on the ghost image changes. Therefore, the brightness change of the image includes both the heartbeat and the breathing information.
  • the beneficial effects of the invention include: Simultaneous measurement of multi-standard and multi-signal signs can be achieved in 2-3 signal periods Investigate and realize the rapid measurement of vital signs such as heart rate and respiration.
  • the detection method of the invention adopts the window source separation method, which can increase the resistance to the interference (the illumination brightness change and the waiting Measure the small movement, etc.), it can realize the trajectory in the unstable environment, the test distance! 3 ⁇ 4 range is large, not limited by the short distance.
  • the detection system of the invention can be used in various illumination conditions, indoor illumination source Realization of all-weather measurement under low-light conditions, etc.
  • the invention has a wide range of applications, both for dry humans and for animals.
  • Non-corrosion type The total structure of the vital sign detection system ⁇ Island
  • Hidden 2 is the structure of the contact-type vital sign detection system based on video concealment of the present invention. 4:3 ⁇ 4 The detection system of the non-corrosive vital sign based on video reading of the present invention shows a circle
  • Picture 5 is a schematic diagram of the original time series signal of R, G, and B channels for heart rate detection.
  • Circle 6 is the actual test of the climber, and the R, G, and B pass of the breath check) start the time series signal to invite the map.
  • Hid 7 is a single-person test example, the normalized time series signal of heart rate detection is more obvious
  • Circle 8 is the actual test of the climber.
  • the normalized valve sequence signal of the breath test is shown in the S chart.
  • painting! 1 is a single-person test example in which the heart rate detection is flattened and the ICA independent component is shown.
  • Hidden i3 is a single-person test example, the correlation function of each independent component in heart rate detection
  • Circle 14 is a schematic diagram of the ⁇ correlation function of each individual component in the breath detection in the climbing test example.
  • the country is a single test case ⁇ , the selected independent component of the heart rate test:
  • the spectrum of the autocorrelation function is shown in Fig. 6 is the spectrum of the heavy autocorrelation function that is the most independent of the breath detection in the single test case.
  • the heart rate detection R, G, and B channel original time series signals are circled.
  • National 18 is a normalized time series signal diagram for heart rate detection in a two-person test case I! 1 is a schematic diagram of the ICA independent component of heart rate detection in a two-person test case.
  • Circle 20 is a two-person test example, and the K:A independent component of the heart rate detection is flattened.
  • Hid 21 is a schematic diagram of the i-related function of each independent component in heart rate detection in a two-person test case.
  • Circle 22 is a spectrogram of the ' ⁇ phase-off function of the independent component of the heart rate detection in a two-person test example.
  • Valve 23 is a schematic diagram of the sequence signal between the original B-inch of the R, G, and B channels of the heart rate detection in the pig test example.
  • Circle 24 is a schematic representation of the normalized time series signal for heart rate detection in a pig test example.
  • Picture 25 is a pig test example, the 1CA independent component of heart rate detection indicates i
  • Hidden 26 is a pig test example, heart rate test «the ICA independent component after the flattening process ⁇ meaning off,
  • Yan 27 is a schematic diagram of the correlation function of each independent component in heart rate detection in the pig test case.
  • Park 28 is the pig test example, the independent component selected in the heart rate test: ⁇ .
  • the detection system of the non-contact type of vital signs based on video imaging includes: video garden image acquisition module i, motion detection module 2, signal separation module 3, intelligent decision module 4, multiple number
  • the frequency extraction module 5 and the display module 6 have one or more targets to be tested, as shown in FIG. 2.
  • the signal separation module 3 includes a heart rate signal separation module 34 and a respiratory signal separation module 3-2. Separation modules for other vital signs are also applicable depending on the actual detection purpose.
  • the letter frequency extraction module 5 includes a heartbeat frequency extraction module 5-1 and a respiratory rhythm extraction module 5-2. According to different actual detection of the west, the signal frequency extraction module of other vital signs can also be applied.
  • the signal acquired by the video image acquisition module 1- is automatically detected by the module 2, the signal separation module 3, the intelligent decision module 4, and the signal frequency extraction module 5 transmits and measures the result, and the detection result is displayed by the display module 6
  • the single person is used as the detection mark for checking
  • the process is as follows;
  • the Long video display module i is by country: Frame frequency continuous acquisition is to be tested: ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ C250)
  • No 24-bit RGB true color image set frame rate 15 3 ⁇ 4, image resolution 640x480
  • the larger the acquisition frame rate and the national image resolution the better, but these two are also subject to the execution speed, so In practical applications, it should be determined according to the specific situation.
  • the 7fps frequency can test the heartbeat of up to 200bpm; for the breath test, the lfps hidden frequency is the fastest 30 times/minute of the breath.
  • ⁇ motion detection module 2 automatically detects the number of ii to be tested in real time.
  • the edge contour detection performs the contour recognition of the upper body of the human body, and the number of the to-be-measured objects is determined by the number of the upper body contours as a single motion detection module 2 to dynamically capture and track the region of interest (RO!), which causes i-chest and abdominal movement due to breathing.
  • the amplitude is the largest, because in this example, the position of the thoracic cavity to be measured is used as the ROI detection module 2 of the respiratory detection to determine the thoracic amine position by edge contour detection.
  • the R01 area of the heart rate detection is the exposed skin area of the target to be measured, and the dynamic recognition and dynamic tracking of the heart rate ROi area is realized by the skin color detection or face detection for each target to be measured, and the motion detection module 2 Converting the pixels in the rim area of the upper body from the RGB space to the YCbCr space for skin color recognition, and detecting the exposed skin area of each target to be detected as the ROi area of the heart rate detection.
  • the vital signs such as heart rate and breathing are ⁇ 3 ⁇ 4, in addition to the ROI region selected above, the subsequent processing steps ...
  • the signal separation module 3 realizes the signal and noise separation detection process, the image signal is susceptible to the intensity noise and distribution noise of the illumination source, the detector noise, and the moving noise of the target to be measured, in order to be weak
  • the heartbeat or respiratory letter '3 ⁇ 4 recovers from the image signal of the mixed window complex noise, now the technology directly uses the filter noise reduction side Method, but the method has limited beam.
  • the invention divides the video image into four-way test, and uses the temporary source K method to realize automatic separation of signal and noise, so as to effectively improve the system's noise immunity and robustness. Sex
  • the signal separation module 3 includes a heart rate signal separation module 3-i and a respiratory signal separation module 3-2.
  • the specific steps are as follows :
  • R, G, ⁇ . ⁇ the road color channels, respectively, each - seeking spatial averaging all pixels within the ROI area of the frame rings of t image region ROi Min heart rate signal sequence to obtain an average heart rate of the original path as required air smell.
  • the red, green, and blue channel signals are RK ⁇ Gl (t), B1 (i); the three-way breath original time series signal obtained by averaging 3 ⁇ 4 of the respiratory ROi region, as shown by the rush 6
  • the red, green, and blue channel signals are R2(t), G2(i), ⁇ 2( ⁇ )
  • the baseline drift is filtered out by the smooth prior method.
  • the cutoff frequency is set to 0.6H.Z.
  • the cutoff frequency is set to 0.16Hz.
  • the baseline drift of the start time series signal caused by the disturbance is filtered out by the flat prior method, so that the subsequent processing effect is better:
  • the cutoff frequency is set to 0.61fe; for the breath detection, the cut will be cut.
  • the frequency is set to 0. i6Hz. ;
  • the normalized time series of the zero mean and the out-of-range variance are obtained by the processing, and the red pass is tested by the heart rate test.
  • the normalization is as follows: Cap (1)
  • R1 (0 and iti '(/) represent the original time series and the time series of the red heart rate of the heart rate test, respectively, and the uniformity and standard of the original time series of the red channel of the heart rate test by ⁇ ⁇ and ⁇ ⁇ , respectively.
  • the heart rate test's red, green, and blue channel maternalization letter #3 ⁇ 4 is R (0, (]i'(0, Bl'(t) as shown in Figure 8, the red color of the breath test,
  • the green and blue channel gestation signals are G2 (i), E2 (t), respectively.
  • the present invention traceback method determines whether there is a strong :: P disturbance according to the variation range of the brightness of the area image, and automatically sets the strong disturbance signal steel brought by the strong interference by setting the brightness change threshold value, thereby ensuring the accuracy of the measurement result.
  • the detection method greatly improves the noise immunity and robustness of the detection system i
  • the blind source separation method uses the fast independent meta-analysis (Fast iCA) to separate the signal and the sound, and the B-time series is used as the observation signal, and the Fast 1CA is used to obtain the independent number of observation signals.
  • Fast iCA fast independent meta-analysis
  • the iCA components of heart rate and respiration detection are shown in Fig. 9 and Fig. 1 respectively.
  • Blind source separation method can also use tofomax method, natural gradient method (Natural GradieM), adaptive method (: EASD, matrix eigenvalue). Decomposition method
  • the signal frequency extraction module 5 first finds the independent component of the corresponding vital sign signal from the Fasi ICA separated output component, hereinafter referred to as the source signal component, and then from the source signal component of the mixed noise.
  • the frequency is extracted as the frequency of the vital sign signal and converted into the body sign, ie heart rate, respiratory meditation.
  • the signal frequency extraction module 5 includes a heartbeat frequency extraction module 5-!-, and a respiratory rhythm extraction module 52. The specific steps are as follows:
  • the point ⁇ averaging filter is used to smooth the independent component of the Fast ICA output, and the independent component of the turbulence disturbance is filtered.
  • the independent component after the respiration detection is shown in circle 12 .
  • the independent component of the Fast!CA output is arbitrarily related, and the ⁇ energy of the correlation function accounts for the proportion of the total energy, and the letter # component with the smallest specific gravity is selected as the source letter # component.
  • the heart rate is tested in this example.
  • the detected source signal component is the second independent component; from the hidden M, the source signal component of the respiratory detection is the first: the independent component of the loop
  • the invention adopts a multi-correlation method, and performs the source signal component: ⁇ photo ⁇ correlation operation, filtering out the click sound on the source signal component, and lifting the signal to noise ratio to the source signal: ⁇ heavy
  • the frequency of the source signal component of the present invention is extracted not only by the multiple correlation method, but also by other methods of frequency detection of the periodic signal (such as bispectral method, wavelet transform method, high-order spectrum analysis, etc.): In this embodiment, the frequency f of the obtained heartbeat letter is converted into a heart rate which is lower than the minute, that is,
  • the heart rate of the test subject is 56 hops per minute.
  • the invention can be used for 3 ⁇ 4 quantity in 2-3 signal periods, in order to achieve fast real-time measurement, according to the type of 3 ⁇ 4 life body letter of the target, respectively, t) heart rate and breathing 3 ⁇ 4 short initial test duration, for example, healthy adult
  • the heartbeat is at 6 (M00 times per minute, breathing about i6 ⁇ 20 times/min, so the heart rate and the initial fibrillation time of the breath (that is, the time required to obtain 2 to 3 signal cycles) can be set to 3 seconds respectively. And 12 seconds.
  • the present invention employs a non-contact vital signs detection system based on the common video images
  • a network camera apparatus as set Bian imaging, computer-hardware platform, a software algorithm, and a display interface with VC programming statement t
  • Bian imaging, computer-hardware platform, a software algorithm, and a display interface with VC programming statement t Set the initial test time of heart rate and breathing to 3 seconds and 12 seconds respectively. After completing the initial test, the image sequence updated every two seconds is real-time I! The new result, 3 ⁇ 4 hair ⁇ strong disturbance, 3 ⁇ 4 move to eliminate the test data of the group and restart a new round of measurement.
  • the motion check «module is detected by the edge wheel gallery.
  • the contour recognition of the upper body of the human body is determined.
  • the number of objects to be tested is determined by the number of contours of the upper body.
  • the chest position is determined as the breath detection R01 area; the heart rate detection Oi field is determined by skin color detection. Then, each test mark is processed separately, and the processing method is the same as the single check.
  • the detection is performed by a single person as the detection fi, and the detection of the detection is performed by the pulse detection and the detection system of the present invention.
  • the detection system of the present invention uses the method of the first embodiment to detect and detect the pulse for 20 seconds. Detecting the number of heartbeats of the target and making a record; at the same time, the detection system of the present invention detects the detection mark, and the detection time is 20 3 ⁇ 4, and the result of the pulse detection is compared with the detection result shown by the detection system of the present invention,
  • the results of the randomly selected test targets are as follows:
  • the detection system of the present invention performs heart rate detection, and the result is basically the same as that of the traditional Chinese medicine.
  • the error detection range of within ⁇ lbpm indicates the detection system of the present invention and
  • the detection method has accurate detection results and good practical application value.
  • the invention also has the advantages of rapid, real-time, multi-west standard simultaneous wine inspection, continuous wine control, strong food bar, low cost and wide application range.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Cardiology (AREA)
  • Physiology (AREA)
  • Medical Informatics (AREA)
  • Animal Behavior & Ethology (AREA)
  • Pathology (AREA)
  • Engineering & Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Physics & Mathematics (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • Biophysics (AREA)
  • General Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Pulmonology (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
  • Measuring Pulse, Heart Rate, Blood Pressure Or Blood Flow (AREA)
  • Measuring And Recording Apparatus For Diagnosis (AREA)

Abstract

公开了一种基于视频图像的非接触式生命体征的检测方法,其包括:按固定帧频连续采集待测目标的视频图像,自动检测图像中的ROI区域;从所述ROI区域划分出的多通道信号中分离出生命体征信号;以及提取出所述生命体征信号的频率并将所述频率转换为所述生命体征,获得检测结果。还公开了一种基于视频图像的非接触式生命体征的检测系统。所述方法和系统具有快速、实时、可实现连续监测、鲁棒性强、低成本以及适用范围广等优点。

Description

基于视频图像的非接触式生命体征的检測方法及检漏系统
技术领域
本发明涉及到家庭保健系统、 医用监护系统 及动物健康检测系统, 尤其渉及一种基于 视频圈像的非接触式 命体征的检測方法及系统, 以检測人体和 /或动物的心率、 呼吸以及心 背景技术
非接触式: 命体征 (包括心率、 呼说吸等) 检测方法及检测设施由于其方便性、 安全性以 及¾活性, 受到业界广泛关注 fi!前已 "的非接触式心率和呼吸检測技术 ±要包括如下的多 普勒雷达探測式和序列顏像采集式两种。 多普勒雷达式非接触测 ¾方法对心跳和呼吸造成的人体体表徵动进行探测, 进而获取人 体心率与呼吸信息。 多普勒雷达式探測系统由控制器、 超宽频信 发送器以及侦测接收返 0 信号的接收器组成 发射机发射的电磁波照射到人体后, 受到人体体表微动的多普勒调制; 接收机对后向散射回波做相干处理, 检测出回波中的多普勒信息, 经放大、 滤波等信号处理 技术最终获得人体的心跳和呼吸信—息
由于多普勒雷达式检测方法的测蕭原理是感知人体体表的徵 理.运动, i 而对身体运动 和周围环境变化极其敏感 此外, 多普勒雷达式探酒系统的构造复杂、 造价昂贵, ^要用于 地震灾害、 塌方事故等紧急人员抢救以及警方人质救援等特殊场合, 不适于在 常生活中普
2. 序列圈像采集式
当光束 a射到皮) 皮肤内的血液对光束产生吸收褒减作 , 并且衰减量取决于 血容积的多寡 s 在心 :容积发 *变化, 照射光束衰减量 4! 应 .¾现波动性变化, 、通过裰像头探
得出生物体心跳信息 基于上述原理, 前国际上 kimo和 Ohta报遒了- '种刺兩延时序列徵像测量心跳和呼吸的 方法 该方法用 CCD相机连续采集 30秒的人脸图像, 在图像屮手动截取人体脸颊部位特定区 域, 求取每帧图像中脸.顯特^区域的平均亮度愤, 将得到的一组亮度时间序列信号依次进行 一齡¾分、 低通滤波和 A.R模型功率谱分析, 所得功率.谱中两个显著的峰愤分别对应心跳和呼 吸频率
国内也^人提出一种基于红外序列图像的心率 损检测方法。 该方法翁先通过红外动态 顯像釆集系统釆集受试者顯像, 手动截取受试者太阳穴处顯像, 釆兩靈心法提取出时间序列 信号, 通过一阶 分、 低通滤波和 AR模型功率谱分析等一系列歩骤实现心率的无损检 «
..!:述方法的局限性表现在以下几方丽: !)不能 动捕获跟踪人体脸颊区域, 只能在完成 顏像 ¾集后, 通过人为手动截取 M区域, 属离线处理方式, 无法实时给出測量结 。 2) m 像信号易受到照明光源的强度噪声和分布噪声、 探測器噪声以及待测 标移动噪声的影响, 因而需要从混裔噪声的国像信号中检測出微弱的心跳和呼吸信号。 釆用单组观测信号迸行滤 波处理的方式, 其去噪和抗噪能力 限, 噪声较强时, 将导致检测结果误差升高, 甚至无 法完成检獰。 3) 需要连续累积 30秒钟的測试时间, 園而仅能给出心率和呼吸在 30秒内的平均 值, 无法反映出心率和呼吸 实时变化
本发明克服了现有技术中 *接蝕 命体征检測系统的装置结构复杂、 造价昂贵、 不适 用于日常生活等缺陷, 以及检测方法 ! 不能进行实时测量、 去噪及抗噪能力盖、 检测时间长 . 不能反映生命体征的实时变化等缺陷, 本发明检測方法及系统具有快逮、 实时、 多目标同 时检濺、 可实现连续监濺、 鲁棒性强、 低成本以及适用范围广等优点
发明内容
本发明提供一种維于视频隱像的 接触式生命体征的检灘 i方法, 包括- 歩骤一, 按國定帧频连续釆集待测 标的视频國像, 自动检测图像中的 ROL区域, 歩骤二, 从所述 R0! ¾域划分出的多通道信号 分离出 ¾命体征信号, 歩骤.: Ξ, 提取出所述^命体征信号的频率并将所述频率转换为所述生命体征, 获得检測 > , 所述^命体征信号的频率包括心跳信号的频率 /和呼吸信号的頻 -f 所述 ¾命体 征包括心率、 呼吸 <: a所述生命体征为心率时, 所述 Ron 域为待测: 标的皮肤区域; 所 述 命体征为呼吸频举时, 所述 R(¾区域为待测目标的胸腔或腹部位置。
本发明检测方法同时检獰多个待獰目标。
本发明检测方法以 2 3个信号.周期为最小测试时长提取所述 命体征信号的频率。
所述 RO! 域划分出的多通道信号的归一化时间序列信号幅度同时超过 4时, 判定此时 存在强扰动信号。 通过将亮度变化阈值设置为 4, 实现将所述强扰动信号剔除 ¾
歩骤二中分离 ±命体征信号包括以 ΐ歩骤; 将所述 区域的图像进行多逾道划分, 计 算各个通道的 ffi间平均值以形成原始时间序列信号, 将所述原始时间序列信号的基线漂移滤 除并归一化处理后, 通过盲源分离法分离出与原始时间序列信¾维数相同的独立分量; 其中, 所述独立分量包括生命体征信号和嗞声 t. 其中, 所述直源分离法包括最大信息量法, ^1然梯 度法 适应法、 快速独立元分析法、 矩阵特征值分解法。
其中, 在所述嫫始时间序列信号的維线濯移滤除并归一化处理后> 根据所述 ROI区域隱 像亮度的变化幅度判定并剔除强扰动信号 ,
在本; 明检测方法中, ¾出现所述的强扰动信号时, '麵謹顯 I一幵始新的检測 歩骤 中提取出生命体征信号的频率包括以下歩骤; 对盲源分离法得到的所述独立分量 迸行信号平滑处 a, 将. 生命特征信号对应的独立分量作为源信号分量, 然后通过周期信号 频率检測法提取所述源信 分量的频率, 获得所述生命体征信¾的频率。
其中, 所述源信号分量是所述独立分量中 ^相关函数的主辮能量占'整体能量比 1;最小的 独立分 S, 或所述独立分 S中功率谱密度峰值凝大的独立分量 , 所述搠期信号频率检濺法包括双谱分析法、 小波变换法和多 a自相关法 本发明 中多靈自相关法提取所述源信 · ·分量的频率的歩骤为, 对所述源信 ¾分量进行多重自相关运 算, 进行频谱分析, 频谱中 峰值功率点为所述生命体征信号的频率
所述将生命体征信号的频率转换为生命体征是利用公式! i ^60> /及 Rs^Ox/i分 m得到每分钟的心跳数 HR和呼吸数 Rs ,
本发明还提供…种基于视频圈像 接触式 命体征的检测系统, 其包括视频围像«集 模块 (i )、 自动检测模块 (2)、 信号分离模块 (3 )、 智能判决模块 (4 、 信号频率提取模块 ( 5) 和显示模块 C6);
其中, 所述视频國像采集模块(1 )按固定帧频连续釆集图像, 并传送给 述 动检獰模 块 (2.);
所述自动检测模块 (2) 自动实时检测待测 标数量, 并捕捉 跟踪所述 R01区域; 所述信号分离模块 ( 3 ) 实现生命体征信 和噪声的分离;
所述智能判决模块 (4) 根据所述 RCM区域顯像亮度的变化幅度判 ¾并剔除强扰动信号; 所述信号频率提取模块(5 )从所述生命体征信号中提取出其频率, 并将其转换为所述生 命体征;
所述显示模块 (6) 实时更新通示所述愤号频率提取模块 (5) 转涣的检测结果 本发明的 的是提供一种 ^动 Φ接触式心率和 /或呼吸的检 «方法及检测系统, 具宵快 速、 实时、 可实现连续监測、 鲁棒性强、 低成本以及适用范围广等特点, 实现了对心率及呼 吸的! ¾动, 实时、 连续的非接触式监测 在心脏搏动作用下, 皮肤内动脉血管的血.容枳发生 变化, 使得皮肤反射光强虽现波动性变化, 从而 起图像亮度值的改变。 呼吸会引起人体胸 腔起伏, 同时还将带动人体腐部、 头都等身体部位移动, 从而致使隱像上的亮度发生改变, 因此, 闘像亮度变化既包含心跳也包含呼吸信息。
本发明中, "2 3个信¾周期"是指心脏搏动或呼吸完成 2 3次
本发明有益效果包括: 实现了多 标、 多 ¾命体征的同时測量 能够在 2-3个信号周期内 究成测量, 实现了心率和呼吸等生命体征的快速测量 ■¾―般的滤波法相比较, 本发明检測 方法采用窗源分离的方法, 能够增加对千扰的抵抗度 (照明光亮度变化和待测者小幅度移动 等), 可以实现在不稳 环境中进行溯量, 测试距离!¾范围较大, 不受短距离的限制 利用本 发明检测系统可在多种照明条件 然光、 室内照明光源、 低光照条件等) 下实现全天候测 量 本发明应用范關广泛, 既适用干人体也适用于动物。
附图说明
画〗 为本; 明基于视频画像! 非接蝕式: 命体征检测系统的总结构^意國
隱 2 为本发明基于视频隱像的 *接触式^命体征检测系统的结构示愈顯 園 4 :¾本发明基于视频閱像的非接蝕式生命体征的检测系统示愈圈
画 5为单人測试实倒中, 心率检测的 R、 G、 B通道原始时间序列信号示意图。
圈 6为攀人測试实 中, 呼吸检獰的 R、 G、 B通遛) 始时间序列信号示邀图。
隱 7为单人测试实例中, 心率检测的归一化时间序列信号^愈顯
圈 8为攀人測试实 中, 呼吸检獰的归一化时阀序列信号示 S图
隱 9为攀人測试实倒中 , 心率检灘的 iCA独立分量示意閱
顯 10为離人测试实例中, 呼吸检测的: iCA独立分 *示意國。
画!1为单人測试实例中, 心率检测经平化处理后的 ICA独立分量示愈图。
12为攀人测试实例中, 呼吸检测经平化处理后的 iCA独立分最示愈圈。
隱 i3为单人测试 例中, 心率检测中各独立分量的 相关函数示意图
圈 14为攀人测试实例中, 呼吸检测中各独立分量的 ^相关函数示意图。
國 is为单人测试实例 μ, 心率检测中所选独立分量的: Ξ赏自相关函数的频谱图 图 Ϊ6为单人测试实例中, 呼吸检测中所逸独立分最的 重自相关函数的频谱圈 國 Π为两人測试实例中, 心率检测的 R、 G, B通道原始时间序列信号示意圈。
國 18为两人測试实例中, 心率检测的归一化时间序列信 示意图 I! 1 为两人测试实例中, 心率检测的 ICA独立分量示意图。
圈 20为两人测试实例中, 心率检测经平化处理 的 K:A独立分量示意 I
隱 21为两人测试实例中, 心率检测中各独立分量的 i相关函数示意图
圈 22为两人测试实例中, 心率检测中所逸独立分量的 Ξ靈 相'关函数的频谱图。
閥 23为猪测试实例中, 心率检測的 R、 G , B通道原始 B寸间序列信号示意图
圈 24为猪测试实例中, 心率检测的归一化时间序列信号示意图。
画 25为猪測试实例中, 心率检测的 1CA独立分量示意 i
隱 26为猪测试实例中, 心率检 «经平化处理后的 ICA独立分量^意關 ,
顏 27为猪测试实例中, 心率检測中各独立分量的 相关函数示意图.
園 28为猪测试实例中, 心率检测中所选独立分量的: Ξ. 相关函数的频f图
具体实施方式
结合以下具体实施例和附圈, 对本发明作迸一歩的详细说明, 本发明的保护内容不局限 于以下实施例 在不背离发明构思的精祌和范闘下, 本领域技术人员能够想到的变化和优点 都被包括在本发明中, 并 i以所附的权禾要求书为保护范围
如顯 ]-所示, 本发明基于视频顯像的 '非接触式^命体征的检测系统包括: 视频園像采集 模块 i、 动检測模块 2、 信号分离模块 3、 智能判决模块 4、 倍号频率提取模块 5和显示模 块 6 待测目标为一个或多个 如图 2所示, 信号分离模块 3包括心率信号分离模块 34、 呼 吸信号分离模块 3-2。依据不同实际检測目的, 其他生命体征信 的分离模块也可适用。信 频率提取模块 5包括心跳频率提取模块 5-1、呼吸节律提取模块 5-2。依据不同实际检测西的, 其他生命体征信 的信号频率提取模块也可适用。 经视頻图像采集模块 1-获取的信号经自动 检測模块 2、 信号分离模块 3 , 智能判决模块 4、 信号频率提取模块 5的传输和处:理得到捡測 结果, 经显示模块 6显示检测结果
实施飼 I
利用本发明基于视频國像的非接触式^命体征的检測方法, 以单人作为检測 标进行检 測, 其过程如下;
获取待测目标的视频園像
如顯 3所示, 朗视频顯像釆集模块 i按國: 帧频连续采集待測: ίίίί标的视频顯像 本实施例屮, 如图 4所示, 可釆用罗技网络裰像头( 号 C250)采無 24位 RGB真彩图 像, 集帧频 15 ¾, 图像分辨举 640x480 本发明中, 采集帧频和国像分辨率是越大越好, 但是这两项同时也受执行速度制约, 所以在实际应用时应根据具体情况分析确定 理论上来 说, 对于心率测试, 7fps的祯频可以测试到最高 200bpm的心跳; 对于呼吸测试, lfps隱 频 W以溯试最快 30次 /分钟的呼吸 圏像分辨率不受限制
自动检測待测目标的数量和 801区域
如圈 3、 衝 4所承, 観频圈像釆集模块 将所釆集的園像传送给 动检测模块 2后, ί':ί 动检测模块 2 自动实时检测待测 标数 ii, 遛过边緣轮廓检测进行 ί动人体上身轮廓识别, 由上身轮廓个数确定待测 标数量为单个 ί 动检測模块 2动态捕获、 跟踪测试感兴趣区域 (RO!), 由于呼吸引起 i 胸腔和腹部运动幅度最大, 因 在本实例将待测 标胸腔位置作为呼 吸检测的 ROI区域 动检測模块 2通过边缘轮廓检測 动确定胸胺位 1
检測心率时, 心率检测的 R01区域为待测 标裸露在外的皮肤区域, 通过肤色检溯或人 脸检测实现对心率 ROi区域的 动识别和动态跟踪 对于每个待测 标, ^动检渕模块 2分 别将其上身轮靡 域内的像素从 RGB空间转换到 YCbCr空间进行肤色识别, 动检测出每 个待測 标裸露在外的皮肤区域作为其心率检测的 ROi区域
本发明对心率和呼吸等生命体征 ίί¾溯量, 除上述选择的 ROI区域不同外, 后续处理歩骤 …..'致
分离生命体征信号和噪声, 同时自动识别并扇除强扰动信号
如闘 1-3所示, 信号分离模块 3实现信号和噪声的分离 检测过程中, 图像信号易受到 照明光源的强度噪 和分布噪声、 探测器噪声以及待测目标移动噪声的影响, 为了将微弱的 心跳或呼吸信' ¾从混窗复杂噪声的图像信号中恢复出来, 现 '技术中直接采用滤波降噪的方 法, 但该方法效梁有限 本发明将视频图像 ¾分为多路测试通遒, 采蹈暫源分 K法实现信号 和噪声的自动分离, ^以有效提高系统的抗噪性和:鲁棒性
如顯 2、圏 3所^,信号分离模块 3中包括心率信号分离模块 3-i、呼吸信号分离模块 3-2。 具体歩骤如下 :
1. 多 试通道划分
将 ROi区域分解为 R、 G、 Β :Ξ个色彩通道, 对心率和呼吸分别获得三路测试通道 ε
2. 生成原始时间序列信号
在 R、 G、 Β .Ξ路色彩通道上, 分别将每一 -帧圏像的 ROI区域内所有像素求空间平均 t. 对心率 ROi区域求空闻平均得到的 路心率原始时闽序列信号 如图 5所示, 红色、 绿色和 蓝色通道信号分别为 RK^ Gl (t), B1 (i); 对呼吸 ROi区域求 ¾间平均得到的三路呼吸原始时 间序列信号, 如衝6所示, 红色、 绿色和蓝色通道信号分别为 R2(t), G2(i), Β2(ί)
3. 滤除基线漂移并 sa ····'化处理
用平滑先验法滤除基线漂移, 对于心率检测, 将截止频率设为 0.6H.Z; 对于呼吸检測, 将截至频率设为 0.16Hz
用平潸先验法滤除扰动带来的嫫始时间序列信号的基线漂移, 使后续处理效.架更好:: 对 于心率检测, 将截止频率设为 0.61fe; 对于呼吸检測, 将截 ¾频率设为 0. i6Hz.; 得迸行归… 化处理, 得到零均值、 離位方差的归一化时间序列 以心率测试的红色通遒为钶, 归一化处 理如下: 帽 (1 )
" · σί
其中 R1(0和 iti '(/)分别代表心率測试红色通遛的原始时间序列和 一化时间序列, μ ΐ和 σϊ分别代衮心率測试红色通道的原始时间序列的均偾和标准 》 如图 7所示, 心率测试的红 色、 绿色和蓝色通道妇 化信 #分 ¾为 R (0, (〕i'(0, Bl'(t) 如图 8所示, 呼吸测试的红色、 绿色和蓝色通道妇一化信号分别为 G2 (i), E2 (t) ,
4. 智能剔除强扰动信号 监测 1、 G , Β .:Ξ路时间序列信号的变化, Ξ路信号同时突变 1寸, 判 此时存在强扰动 信 : 新启动测量., 在本实例中, 将突变阔值设为 4 , ¾ II, G , Β .:Ξ路信 ·同时超过突变 國值时, 判^此时存在强扰动信号
如顏 .-3所示, 能判决模块 4 ί动识别并剔除强扰动信号 5 检测过程中, 偶发的大千 扰或强千扰(例如待测 标身体大幅移动)会对测试结果产生影响,本发明检溯方法根据 区域图像亮度的变化幅度来判定是否存在强 ::P扰, 通过设定亮度变化阈值, 自动将强干扰带 来的强扰动信号鋼除, 从而保证测量结果的准确性 本发明检測方法在很大程度上提高检測 系统 i 抗噪性和鲁棒性
本实施例中, 三路归 ·····化时间序列信号幅度同时超过 4时, 判定此时存在强扰动信号 通过将焭度变化阐愆设覽为 4, 实现将所述强扰动信号剔除
5, 盲源分离
在本实例中盲源分离法采用快速独立元分析法 (Fast iCA)迸行信号和喷声的分离, 将的 B ―化时间序列作为观察信号, 剁用 Fast 1CA得到与观察信号数量相等的独立分量。 心率和呼 吸检测的 iCA分量分别见图 9和闘 1 其中, 盲源分离法还可采用凝大信息量法(tofomax)、 自然梯度法 (Natural GradieM)、 自适应法 (: EASD、 矩阵特征值分解法等方法
提取生命体征信号的頻率 并转换为生命体征信号
如閱 1 3所示, 用信号频率提取模块 5首先从 Fasi ICA分离输出分量中找出对应生命特 征信¾的独立分量, 以下称之为源信号分量, 再从混齊噪声的源信号分量中提取出其频率, 为生命体征信号的频率, 并将其转换为所述 ^命体征, 即心率、 呼吸节禅。 信号频率提取模 块 5包括心跳频率提取模块 5- !-、 呼吸节律提取模块 5 2。 具体歩骤如下:
1 . 信号平滑处理
采用 点潸动平均滤波对 Fast ICA输出的独立分量迸行信'号平滑处理, 滤除波动千扰 其中心率检測平滑后的独立分量见圈 I I; 呼吸检测平灌 '后的独立分量见圈 12.
2. 确定源信号分量 将 Fast !CA输出的独立分量傲 相关运獰, 计獰 相关函数的 辮能量占整体能量的比 重, 比重最小的信 #分量被选为源信 #分量 如图 13·所示, 本实例中测试心率检测的源信^ 分量为第二路独立分量; 由隱 M可知, 呼吸检测的源信号分量为第.: Ξ路独立分量
3. 源惜号分量频率櫞取
本发明采 多 相关法, 将源信号分量进行: Ξ攝^相关运算, 滤除源信号分量上的嗞 声, 提 信噪比 对源信号: Ξ重 |相:关函数进行快速傅立叶变换, 获得的频谱中的峰值功率 点即为生命体钲信号的频率。 本发明的源信号分量频率的提取不仅 R于多重 ί相关法, 其它 的搠期信号频率检濺的方法(诸如双谱分祈法、 小波变换法 高阶频谱分析法等) 也都适用:: 本实施例中, 将所得心跳信 的频率 f转换为以分钟为卑位的心率, 即
Figure imgf000012_0001
如图! 5所示, 本测试实例中 ./Κί.93Ηζ, 由此可得待测者心率为每分钟 56跳。
将所得呼吸信号的频率. /Ϊ转换为以分钟为单位的呼吸次数
Rs===60x/i (3) 如图 所示, 本测试实例中 M).256Hz, 由此可得待测者呼吸节律为每分钟 B次 实时更新显示检测结果
通过显示模块 6实时 ¾新愿示信号频率提取模块 5转换的检測结果, 如图 4所示。
本发明在 2-3个信号周期内卿可究成 ¾量, 为实现快速实时测量, 根据待灣 标的¾命 体征信 类型, 分别确) t心率和呼吸 ¾短初始测试时长 例如, 健康成人的心跳在 6(M00次 每分钟, 呼吸约 i6〜20次 /分钟, 因此可将心率和呼吸的初始纖 I试时间(即获取 2- 3个信号周 期所需要的时间)分别设为 3秒和 12秒。 完成初始测试后, 将每秒钟更新的闘像序列添加 S S始时间信号序列中, 计算新的测量结果《本实例将最长澱试窗口设置为 30s, 当累积 ¾试时 间达到 30秒以后, ¾用窗 滑动方式, 潸动增置设为〖秒, .从而实现实时监测 发生强扰 动时, 利用智能判决模块 4依据 ROI ^域划分出的多通道信号的归一化时间序列信号幅度同 时超过 4, 判定并自动剔除该组强扰动信号涵试数据, 时 B获取待测 H标的视频图像开始 靈新. 动新一轮测量, 由此确保濺量 连续性
如闕 4所示, 本发明基于视频图像的非接触式生命体征的检测系统中采用普通阿络摄像 头作为顯像釆集设备, 以电脑为硬件平台, 用 VC编程 言实现软件算法和显示界面 t. 将心 率和呼吸的初始测试时岡分别设为 3秒.和 12秒, 完成初始測试后 利兩每秒钟更新的图像序 列实时 I!新 量结果, ¾发 ^强扰动时, ¾动剔除该组测试数据同时重新启动新一轮濒量。 实施例 2
以两人为检测目标同时检測, 与单人区别仅在于:
动检«模块通过边缘轮廊检濺进行 ^动人体上身轮廓识别, 由上身轮廓个数前先确定 待測 标数量为两个。
对于每个测试 标分别在其上身轮靡范围内, 确定胸腔位置作为呼吸检测 R01区域; 通 过肤色检测确定心率检测 Oi 域。 而后对每个测试 标分别迸行处理, 处理方法阔单人检 獰相同。 心率检測的 R、 G、 B通道原始时闽序列信号见圏 Π, 心率检测的归一化时间序列 信号见顯 18, 心率检测的 ICA独立分量见顯 19, 心率检溯经平化&理后的 ICA独立分量见 顏 20, 心率检獰中各独立分量的 ^相关函数见 iH 21 ,
我们以双人心率检测为实例对图 22中实验结果迸行了计算, 两个灘 i试 标的心率分别为 5 bpm和 72 bpm。
实施例 3
以动物猪作为检測 标进行检濺 其他实验过程类叙子实施例: 1 , 不重复赞述 ¾ 其中, 区别在于, 猪的心率和呼吸 区域可同时取为猪的腹部 猪腹部的 动检测通过边缘轮廓 检测实现。 具体如下: 通过边缘轮廓检测对猪迸行轮廓识别, 并对猪形体迸行椭闘拟合, 拟 合椭圆的'中心位置即为腹部中心, 将拟合櫞圆长 > 短轴分别取 40%, 得到猪腹部区域作为 ROi区域 后续处理方法与单人检測相同 心率检测的 R、 G, B通遒原始时间序列信号见國 23 , 心率检測的归一化时间序列信号见顯 24, 心率检測的 iCA独立分量见圈 25, 心率检测 经平化处理后的 ICA独立分量见图 26, 心率检«中各独立分量的 相关函数见图 2Ί 我们以猪心率检溯为实飼对隱 28中实验结果进行了计算, 被检測猪的心率为 72 hpm, 实施例 4
以单人作为检测 fi标迸行检测, 对检溯 标分别通过把脉计数及本发明检测系统迸行 率检测, 利用本发明检測系統依照实施例 1的方法迸行检測 检测时把脉 20秒, 计数检測 标的心跳次数并作记录; 同时利服本发明检测系统对检测 标迸行检测, 检測时长为 20 ¾ 将把脉检测的结果与本发明检测系统所显示的检溯结梁迸行比较, 对十名随机选取的检測 标的检 结果如下:
丧 1 通过把脉及本发明检灘人心率的对比结果
Figure imgf000014_0001
从上述表 1 所示对比检测结果可见, 利^本发明检测系统进行心率检测, 其结果与通过 传统中医的把脉检濺结果基本一 ·致,误: 范围在 ± lbpm以内 表明本发明检测系统及检测方 法的检测结果准确, 具有良好的实际应用价值, 同时本发明还具有快速、 实时、 多西标同时 检酒、 可实现连续监酒、 餐棒性强、 低成本以及适用范围广 优点

Claims

权 利 要 求 书 、 一 '种基于视频图像的非接触式生命体征的检测方法, 其特征在予, 包括
歩骤一 , 按阖定帧频连续采集待測 标的视频图像, 自动检测图像中的 ROi 域, 歩骤二, 从所述 R0 区域划分出的多通遒信号中分离出生命体征信号,
歩骤王., 提取出所述生命体征信号的 '频率并将所述频率转换为所述生命体征, 获得检测
、 如权利要求!.所述的检测方法, 其特征在于, 所述生命体征信号的频率包括心跳信号的频 率./和呼吸信号的频率 i ; 所述生命体征包括心率、 呼吸
、 如权利要求 2所述的检测方法, 其特征在于, 当所述生命体征为心率时, 所述 RCH区域 为待測 标的皮肤区域; 所述生命体征为呼吸时, 所述 ROi 区域为待测 标的胸腔或 腹部位置.
、 如权利要求〗所述的检测方法, 其特征¾于, 所述检测方法同时检测多个待测 标 、 如权利要求!所述的检测方法, 其特征在于》 所述检測方法以 2 3个信号周期为最小測试 时长提取所述生命体 £信号的频率
、 如权利要求!.所述的检测方法, 其特征在子, 歩骤二中分离生命体征信号包括以下歩骤- 将所述 RCM 区域的隱像迸行多通遨划分, 计算各个通道的空间平均值以形成原始时间序 列信号, 将所述原始时间序列信号的基线漂移滤除并归 ·······化处理后, 通过窗源分离法分离 出. 原始时间序列信号維数相同的独立分量; 其中,所述独立分量包括生命体征信号和噪
: 、 、 如权利要求 6所述的检测方法, 其特征在于, 所述窗源分离法包括最大信息量法、 然梯 度法、 自适应法、 快速独立元分析法、 矩阵特征值分解法
、 如权利要求 6所述的检测方法, 其特征在 , 在所述原始时间序列信号的基线濯移滤除并 归一化处斑后, 根据所述 RO:i区域簡像亮度的变化幅度判定并:綱除强扰动信号。
¾ 如权利要求 1所述的检測方法, 其特征在子, 所述检测方法中, 当出现如权利要求 8所述 的强扰动信 ¾时, 靈新 ^歩骤一开始新的检雜。 10、 如权利要求 1所述的检«方法,其特征在于, 歩骤.::::中提取 命体征信号的频率包括 以下歩骤: 对窗源分离法得到的所述独立分量进行信 · ·平滑处理, 将与生命特征信¾对应 的独立分量作为源信号分量, 然 通过周期信号频率检測法提取所述:源信号分量 fft频率, 获得所述生命体征信号的频率
1 1、 如权利要求 10所述的检测方法, 其特征在于, 所述源信号分量是所述独立分量中自 相关函数的主辮能量占整体能量比童 ¾小的独立分量,或所述独立分量中功率谱密度峰值 最大的独立分量。
12、 如权利要求 10所述的检测方法, 其特征在于, 所述阀期信号频率检测法包括双谱分 析法、 小波变换法和多靈 相关法。
13、 如权利要求!0所述的检测方法, 其特征在于, 所述周期信号频率捡測法是多 ffi 相 关法, 对所述源信号分 i 进行多重自相关运算, 迸行频谱分析, 频谱中的峰值功率点为所 述生命体征信号的频率
14、 如权利要求 1所述的检测方法,其特征在于, 所述将 ^命体征信号的频率转换为生命 体征是利用公式 FiR^O X /及 /1分别得到每分钟的心跳数 HR和呼吸数 Rs。
15、 一种基于视频隱像的非接触式生命体征的检测系统, 其特征在于, 所述系统包括视频 图像釆集模块(!)、 自动检测模块 (2)、 信号分离模块(3 )、 智能判决模块 (4)、 信号频 率提取模块 (5) 和.1:示模块 (6):
其中, 所述视频圈像釆集模块 按固) 帧频连续采集國像, 并传送给所述 ^动檢測機 块 a);
所述 ^动检测模块 (2) ^动实时检测待测目标数量, 并捕捉、 跟踪所述 )¾01 1 域: 所述信号分离模块 (3 ) 实现生命体征信号和噪声的分离;
所述智能判决模块 (4) 根据所述 RO 区域顯像亮度的变化幅度判定并鋼除强扰动信号; 所述信 频率提取模块 (5 ; 从所述生命体征信号中提取出其频率, 并将其转換为所述生 命体征; 所述显承模块 (6) 实时更新显 所述信号频率提取模块 (5) 转换的检测结.果<
3
PCT/CN2012/083181 2011-10-24 2012-10-19 基于视频图像的非接触式生命体征的检测方法及检测系统 WO2013060249A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201110325533.3 2011-10-24
CN2011103255333A CN102499664B (zh) 2011-10-24 2011-10-24 基于视频图像的非接触式生命体征的检测方法及检测系统

Publications (1)

Publication Number Publication Date
WO2013060249A1 true WO2013060249A1 (zh) 2013-05-02

Family

ID=46211846

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2012/083181 WO2013060249A1 (zh) 2011-10-24 2012-10-19 基于视频图像的非接触式生命体征的检测方法及检测系统

Country Status (2)

Country Link
CN (1) CN102499664B (zh)
WO (1) WO2013060249A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106793976A (zh) * 2014-10-13 2017-05-31 皇家飞利浦有限公司 用于检测对象的生命体征信息的设备和方法

Families Citing this family (61)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102499664B (zh) * 2011-10-24 2013-01-02 西双版纳大渡云海生物科技发展有限公司 基于视频图像的非接触式生命体征的检测方法及检测系统
CN103054569B (zh) * 2012-12-20 2015-04-22 Tcl集团股份有限公司 基于可见光图像测量人体心率的方法、装置及手持设备
CN103006195B (zh) * 2013-01-10 2015-06-17 浙江大学 一种基于图像处理的非接触式生命体征数据监测系统和监测方法
CN103006187B (zh) * 2013-01-10 2014-12-31 浙江大学 一种非接触式生命体征数据监测系统和监测方法
WO2014140978A1 (en) * 2013-03-14 2014-09-18 Koninklijke Philips N.V. Device and method for obtaining vital sign information of a subject
CN103126655B (zh) * 2013-03-14 2014-10-08 浙江大学 一种非约束目标的非接触式脉搏波采集系统及采集方法
US9339210B2 (en) * 2013-05-08 2016-05-17 Koninklijke Philips N.V. Device for obtaining a vital sign of a subject
CN104138254A (zh) * 2013-05-10 2014-11-12 天津点康科技有限公司 非接触式自动心率测量系统及测量方法
CN104173051A (zh) * 2013-05-28 2014-12-03 天津点康科技有限公司 非接触式自动呼吸测量系统及测量方法
CN104173032A (zh) * 2013-05-28 2014-12-03 天津点康科技有限公司 电子健康检测平板系统
CN103505196B (zh) * 2013-09-02 2015-06-17 展讯通信(上海)有限公司 人体脉搏测量方法、装置和移动终端
CN103462597B (zh) * 2013-09-18 2015-04-29 温州医科大学 一种基于智能手机控制的预防阻塞性睡眠猝死的监测装置
CN103654758A (zh) * 2013-12-23 2014-03-26 韩山师范学院 一种抗干扰的心率检测方法
CN103702014B (zh) * 2013-12-31 2017-02-15 中国科学院深圳先进技术研究院 非接触式生理参数检测方法、系统及装置
DE112015000614T5 (de) * 2014-02-03 2016-10-27 The Board Of Trustees Of The Leland Stanford Junior University Kontaktlose physiologische Überwachung während gleichzeitiger Magnetresonanztomographie
JP6191517B2 (ja) * 2014-03-14 2017-09-06 富士通株式会社 検出装置、検出プログラム、及び検出方法
US9892336B2 (en) * 2014-05-16 2018-02-13 Mediatek Inc. Detection devices and methods for detecting regions of interest
US10004411B2 (en) 2014-05-16 2018-06-26 Mediatek Inc. Living body determination devices and methods
CN104000598B (zh) * 2014-06-13 2015-11-18 安徽工业大学 一种基于最大内切圆的猪呼吸急促运动建模方法及系统
US20160055635A1 (en) * 2014-08-21 2016-02-25 Sony Corporation Method and system for video data processing
CN104270658A (zh) * 2014-09-15 2015-01-07 中山市乐居智能技术开发有限公司 基于心率变异性非接触式健康监测系统的电视机及方法
CN104270504A (zh) * 2014-09-15 2015-01-07 中山市乐居智能技术开发有限公司 一种基于互联网非接触式健康监测的手机及方法
CN104257365A (zh) * 2014-09-15 2015-01-07 中山市乐居智能技术开发有限公司 用于笔记本电脑操作者非接触式健康监测的装置及方法
US11771375B2 (en) 2014-09-26 2023-10-03 Pixart Imaging Inc. Respiration rate detection device and breath detection device adopting motion denoising
US9693710B2 (en) * 2014-10-21 2017-07-04 Xerox Corporation System and method for determining respiration rate from a video
US20160140390A1 (en) * 2014-11-13 2016-05-19 Intel Corporation Liveness detection using progressive eyelid tracking
JP6462594B2 (ja) * 2015-04-30 2019-01-30 オリンパス株式会社 撮像装置、画像処理装置および画像処理方法
WO2016184703A1 (en) * 2015-05-21 2016-11-24 Koninklijke Philips N.V. Identifying living skin tissue in a video sequence
CN104943615A (zh) * 2015-06-08 2015-09-30 庄可军 一种具备酒驾探测及空气净化功能的汽车中央后视镜
CN106264532A (zh) * 2015-06-11 2017-01-04 原相科技股份有限公司 光学式呼吸率检测装置及其检测方法
CN105046209B (zh) * 2015-06-30 2019-01-25 华侨大学 一种基于典型相关分析的非接触式心率测量方法
CN105147274B (zh) * 2015-08-04 2018-06-15 河北工业大学 一种从可见光谱段人脸视频信号中提取心率的方法
CN106618483B (zh) * 2015-10-30 2019-11-05 西双版纳生物医学研究院 基于生理状态应激变异的检测方法及系统和应用
CN105266787B (zh) * 2015-11-03 2018-07-06 西安中科创星科技孵化器有限公司 一种非接触式心率检测方法及系统
CN108289620B (zh) * 2015-11-13 2021-06-25 皇家飞利浦有限公司 用于传感器位置引导的设备、系统和方法
CN105996993A (zh) * 2016-04-29 2016-10-12 南京理工大学 一种智能视频生命体征监测系统及方法
CN105956556B (zh) * 2016-04-29 2019-05-07 南京理工大学 一种测量运动目标生命体征的方法
CN105997035B (zh) * 2016-05-04 2020-03-27 深圳欧德蒙科技有限公司 一种非接触式心率测量方法及系统
CN105962915B (zh) * 2016-06-02 2019-04-05 安徽大学 非接触式人体呼吸率与心率同步测量方法及系统
CN106096544B (zh) * 2016-06-02 2019-04-23 安徽大学 基于二阶盲辨识的非接触式眨眼与心率联合检测系统及方法
CN106377240A (zh) * 2016-08-30 2017-02-08 苏州品诺维新医疗科技有限公司 一种检测心跳速率的装置及方法
CN106361340B (zh) * 2016-08-30 2019-05-31 苏州涵轩信息科技有限公司 一种检测呼吸频率的方法及装置
CN106491114B (zh) * 2016-10-25 2020-11-06 Tcl科技集团股份有限公司 一种心率检测方法及装置
CN106845395A (zh) * 2017-01-19 2017-06-13 北京飞搜科技有限公司 一种基于人脸识别进行活体检测的方法
CN107334469A (zh) * 2017-07-24 2017-11-10 北京理工大学 基于支持向量机的非接触多人心率测量方法及装置
CN107616795A (zh) * 2017-08-17 2018-01-23 华东师范大学 一种基于摄像头的非接触式实时呼吸率检测方法
CN107731306B (zh) * 2017-10-10 2021-04-13 江西科技师范大学 一种基于热成像技术的非接触式心率提取方法
CN107886734A (zh) * 2017-12-07 2018-04-06 苏州诚满信息技术有限公司 一种用于道路交通的智能跨线检测方法及其系统
CN110313908B (zh) * 2018-03-30 2022-03-08 深圳市理邦精密仪器股份有限公司 一种获取实时心率的方法和装置
CN109008964A (zh) * 2018-06-27 2018-12-18 浏阳市安生智能科技有限公司 一种生理信号提取的方法及装置
CN109171649B (zh) * 2018-08-30 2021-08-17 合肥工业大学 智能影像式生命体征探测仪
CN109480808A (zh) * 2018-09-27 2019-03-19 深圳市君利信达科技有限公司 一种基于ppg的心率检测方法、系统、设备和存储介质
CN109589505B (zh) * 2018-11-26 2020-12-29 北京科技大学 一种用于放射治疗过程的多功能监测方法
CN111127511B (zh) * 2018-12-18 2022-03-29 玄云子智能科技(深圳)有限责任公司 一种非接触式心率监测方法
CN109754602A (zh) * 2019-01-15 2019-05-14 珠海格力电器股份有限公司 行人闯红灯防误判的方法和装置
CN111345802B (zh) * 2020-03-16 2023-07-07 南京润楠医疗电子研究院有限公司 一种基于心冲击信号获取双人心跳频率的装置及方法
CN113288114A (zh) * 2021-05-28 2021-08-24 南京润楠医疗电子研究院有限公司 一种基于无线信号的多人混合呼吸信号提取方法
CN114469076B (zh) * 2022-01-24 2024-04-19 南京邮电大学 一种融合身份特征的独居老人跌倒识别方法及系统
CN114469036A (zh) * 2022-01-27 2022-05-13 无锡博奥玛雅医学科技有限公司 一种基于视频图像的远程心率监测方法及系统
CN114469037B (zh) * 2022-01-29 2024-01-12 武汉大学 一种基于毫米波雷达的心率测量方法
CN118334020A (zh) * 2024-06-12 2024-07-12 长春大学 一种基于面部视频流的高精度实时心率监测方法和系统

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH09253071A (ja) * 1996-03-22 1997-09-30 Agency Of Ind Science & Technol 呼吸の測定方法
JPH1186002A (ja) * 1997-09-08 1999-03-30 Toshiba Corp 画像処理装置及び被介護者観測装置
US20080045847A1 (en) * 2006-06-30 2008-02-21 University Of Louisville Research Foundation, Inc. Non-contact and passive measurement of arterial pulse through thermal IR imaging, and analysis of thermal IR imagery
CN101299967A (zh) * 2005-11-04 2008-11-05 株式会社东芝 呼吸监视装置、呼吸监视系统、医疗处理系统、呼吸监视方法、呼吸监视程序
CN102499664A (zh) * 2011-10-24 2012-06-20 西双版纳大渡云海生物科技发展有限公司 基于视频图像的非接触式生命体征的检测方法及检测系统

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH09253071A (ja) * 1996-03-22 1997-09-30 Agency Of Ind Science & Technol 呼吸の測定方法
JPH1186002A (ja) * 1997-09-08 1999-03-30 Toshiba Corp 画像処理装置及び被介護者観測装置
CN101299967A (zh) * 2005-11-04 2008-11-05 株式会社东芝 呼吸监视装置、呼吸监视系统、医疗处理系统、呼吸监视方法、呼吸监视程序
US20080045847A1 (en) * 2006-06-30 2008-02-21 University Of Louisville Research Foundation, Inc. Non-contact and passive measurement of arterial pulse through thermal IR imaging, and analysis of thermal IR imagery
CN102499664A (zh) * 2011-10-24 2012-06-20 西双版纳大渡云海生物科技发展有限公司 基于视频图像的非接触式生命体征的检测方法及检测系统

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106793976A (zh) * 2014-10-13 2017-05-31 皇家飞利浦有限公司 用于检测对象的生命体征信息的设备和方法
CN106793976B (zh) * 2014-10-13 2020-05-05 皇家飞利浦有限公司 用于检测对象的生命体征信息的设备和方法

Also Published As

Publication number Publication date
CN102499664B (zh) 2013-01-02
CN102499664A (zh) 2012-06-20

Similar Documents

Publication Publication Date Title
WO2013060249A1 (zh) 基于视频图像的非接触式生命体征的检测方法及检测系统
Tasli et al. Remote PPG based vital sign measurement using adaptive facial regions
EP3405105B1 (en) Method and apparatus for estimating heart rate
CN113925475B (zh) 一种非接触式的人体健康监测系统及监测方法
EP3229676B1 (en) Method and apparatus for physiological monitoring
CN106901741A (zh) 一种适用于昼夜环境的呼吸率检测方法
EP2960862B1 (en) A method for stabilizing vital sign measurements using parametric facial appearance models via remote sensors
Blöcher et al. An online PPGI approach for camera based heart rate monitoring using beat-to-beat detection
CN106413533A (zh) 用于检测对象的呼吸暂停的设备、系统和方法
Seifert et al. Detection of gait asymmetry using indoor Doppler radar
CN108937905B (zh) 一种基于信号拟合的非接触式心率检测方法
LV14514B (lv) Iekārta un metode sirdsdarbības parametru optiskai bezkontakta kontrolei
CN102309318A (zh) 基于红外序列图像的人体生理参数检测的方法
US11083382B2 (en) Method, information processing apparatus and server for determining a physiological parameter of an individual
CN112233813A (zh) 一种基于ppg的非接触式无创心率呼吸测量方法及系统
JP6620999B2 (ja) 生体情報計測装置、生体情報計測プログラム、及び生体情報計測方法
CN110236515A (zh) 一种基于近红外视频的非接触式心率检测方法
CN110801212B (zh) 基于神经网络的bcg信号心率提取方法
CN105869144A (zh) 一种基于深度图像数据的非接触式呼吸监测方法
CN110200607A (zh) 基于光流法与lms算法消除生命体征检测中体动影响的方法
Chen et al. Modulation model of the photoplethysmography signal for vital sign extraction
CN112294282A (zh) 基于rppg的情绪检测装置的自标定方法
KR102150635B1 (ko) 비전 기반 심박수 측정 방법
Cui et al. Non-contact time varying heart rate monitoring in exercise by video camera
Lian et al. Robust and remote photoplethysmography based on smartphone imaging of the human palm

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 12844292

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 12844292

Country of ref document: EP

Kind code of ref document: A1