A non-contact blood pressure measurement system

Abstract

This invention belongs to the field of non-contact vital sign detection technology, specifically relating to a non-contact blood pressure measurement system, comprising: an acquisition module configured to acquire a face image; a processing module configured to generate a face recognition model based on training the acquired face image using cascaded regression factors, extract the region of interest (ROI) from the acquired face image according to the generated face recognition model, and perform noise reduction processing on the ROI in the face image using the optimization variational mode decomposition method to obtain an imaging photoplethysmography (IPD) signal; and a measurement module configured to process the obtained IPD signal using Euler image magnification technology to obtain a non-contact cardiac impulse signal and pulse transit time, thereby realizing non-contact blood pressure measurement.

Description

Technical Field

[0001] This invention belongs to the field of non-contact vital sign detection technology, specifically relating to a non-contact blood pressure measurement system. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Blood pressure (BP) is the lateral pressure exerted by blood on the walls of blood vessels per unit area as blood flows through them; it is the driving force propelling blood flow. Depending on the location of the blood vessels, blood pressure can be classified as arterial blood pressure, venous blood pressure, and capillary pressure. The blood pressure commonly referred to is the arterial blood pressure of the systemic circulation. Hypertension has become a significant risk factor for human health and can induce various diseases leading to death. Accurate blood pressure monitoring plays a crucial role in the prevention of hypertension and the monitoring of the health status of hypertensive patients.

[0004] Traditional blood pressure measurement is divided into invasive and non-invasive methods. While invasive blood pressure measurement is highly accurate, it requires inserting a detector into a blood vessel, so it is typically used in hospitals for critically ill patients or in intensive care units. In everyday life, non-invasive cuff-based blood pressure measurement is more common. This method involves tightly wrapping a cuff around the subject's upper arm and obtaining estimates of systolic and diastolic pressure by monitoring pressure changes during inflation and deflation. However, this method is time-consuming, generally requiring at least two minutes per test, and only provides one systolic and one diastolic reading, making it unsuitable for rapid, real-time, and continuous blood pressure monitoring. Furthermore, as a contact-based method, the accuracy of blood pressure measurement is highly dependent on the tightness of the cuff and the sensor's sensitivity to pressure changes, resulting in poor reliability of repeated tests and making it unsuitable for patients with arm injuries, bleeding, infection, or amputations.

[0005] Photoplethysmography (PPG) is a blood volume measurement technique. Its principle is that pulsating blood traveling through the human cardiovascular system alters the blood volume in skin tissue. The circulation of oxygenated blood causes fluctuations in the number of hemoglobin molecules and proteins, resulting in fluctuations in the optical absorption across the entire spectrum, forming a pulse wave. This pulse wave can effectively reflect a large amount of physiological information such as respiration, heart rate, and blood oxygen saturation. Pulse wave velocity (PWV) refers to the propagation speed of the pressure wave generated by each heartbeat along the wall of the aorta. It is a simple, effective, and economical non-invasive indicator for assessing arterial stiffness and is strongly correlated with blood pressure. PWV can be obtained by measuring the phase difference of the pulse wave peaks in different regions, i.e., the pulse transit time (PTT). However, the application scope of traditional PPG is limited by its single monitoring area and the need for skin contact.

[0006] Imaging photoplethysmography (IPPG) obtains pulse waves based on imaging. Its technical principle involves using an RGB camera to capture minute color changes reflected from the skin, thereby identifying the stage of blood circulation and extracting the pulse wave signal. It has advantages such as being non-contact with the test site and being simple and easy to operate, solving the problem of measuring physiological parameters using contact instruments for patients with burns or limb defects, as well as infants. However, IPPG is sensitive to light and motion; currently, it only provides relatively accurate pulse wave information under sufficient light and at rest. IPPG acquires pulse waves based on the blood volume of the superficial layer of the skin. Compared to fingertip pulse oximeters that directly record the blood volume of the finger veins, the non-contact pulse waves obtained using IPPG may have millisecond-level deviations in waveform peak value. Furthermore, predicting blood pressure using PTT requires extremely high signal accuracy; therefore, using areas close to the face such as the nose and forehead to predict blood pressure values ​​introduces significant errors. Non-contact pulse wave signals acquired using carotid artery pulsation and IPPG (intra-pulse pulse measurement) are used to obtain PTT (post-traumatic heart rate) signals, mitigating the impact of millisecond-level peak errors. Carotid artery pulsation is a type of ballistocardiogram (BCG), a physiological signal reflecting the mechanical motion of the heart and effectively acquiring heartbeat pulse information. A common method for acquiring BCG is using a pressure sensor to record pressure waveforms to obtain heartbeat pulse information; however, this method requires hardware and is cumbersome to wear. Summary of the Invention

[0007] To address the aforementioned issues, this invention proposes a non-contact blood pressure measurement system. This system achieves precise localization through face detection, selection of high signal-to-noise ratio skin regions, and removal of complex noise, enabling non-contact acquisition of BCG signals. The PTT (post-traumatic stress test) signal is then extracted to obtain the blood pressure value.

[0008] According to some embodiments, the present invention provides a non-contact blood pressure measurement system, which adopts the following technical solution:

[0009] A non-contact blood pressure measurement system, comprising:

[0010] The acquisition module is configured to acquire face images;

[0011] The processing module is configured to generate a face recognition model based on the face image obtained by training with cascaded regression factors, extract the region of interest in the obtained face image according to the generated face recognition model, and perform noise reduction processing on the region of interest in the face image according to the optimization variational mode decomposition method to obtain the imaging photoplethysmography signal.

[0012] The measurement module is configured to process the imaging photoplethysmography signal obtained by Euler image magnification technology to obtain non-contact cardiac impact signal and pulse transit time, thereby realizing non-contact blood pressure measurement.

[0013] As a further technical limitation, in the acquisition module, a face video sequence is acquired, and the acquired face video sequence is processed by frame segmentation to obtain the face image of each frame in the continuous video sequence, that is, to acquire the face image.

[0014] As a further technical limitation, the face recognition model adopts the Landmark model, which obtains several face markers in the face image, performs graphic processing on the obtained face markers to obtain a face mask, and processes the obtained face mask to obtain the region of interest in the face image.

[0015] As a further technical limitation, the region of interest in the face image contains environmental pixels and non-skin pixels. A high signal-to-noise ratio skin region selection method is used to preserve skin region information and remove non-skin interference in the region of interest of the obtained face image.

[0016] Furthermore, the peak summation of the non-contact pulse wave extracted from each skin patch is performed. The skin patch with the most peaks and frequencies is considered to have a high signal-to-noise ratio, and the skin patches with other peak frequencies and frequencies are discarded. The common RGB signal of all the remaining skin patches is obtained, and CHROM and OVMD are performed to obtain the final non-contact pulse wave signal.

[0017] As a further technical limitation, the constrained variational model in the optimization variable mode decomposition method adopted is:

[0018]

[0019]

[0020] Among them, {u k}={u1,u2,…,u k} and {w k}={w1,w2,…,w k} represents the k-th modal component and its corresponding center frequency, for a total of K modes; For partial derivative operations; δ(t) is the unit impulse function; j represents the imaginary unit; * represents convolution operation; f is the target signal; by introducing a penalty factor α and a Lagrange multiplier λ to solve the variational constraint problem, the augmented Lagrange expression is obtained as follows:

[0021]

[0022] An alternating multiplier algorithm is used for iterative updating and solving, updating {u} in the frequency domain. k},{w k},λ.

[0023] Furthermore, the update iterative solution process is as follows: initialization and n; according to and Implement u respectively k w k Iterative updates of λ and λ are performed until the iteration termination condition is met. Output modal components; where τ is the fidelity coefficient; ∧ represents the Fourier transform; n is the number of iterations; ε is the discrimination precision, and ε > 0; the superscript n is the number of iteration steps, and the subscript k represents the current number of modes.

[0024] As a further technical limitation, the formula for obtaining the non-contact cardiac impact signal is as follows: Where x and y are the horizontal and vertical coordinates of the marker point, respectively, and the subscript i of x and y is the serial number of the landmark point.

[0025] As a further technical limitation, the obtained non-contact cardiac impact signal is smoothed, and peak detection is used to obtain the peak index of the non-contact cardiac impact signal and the imaging photoplethysmography signal. Combined with the sampling frequency, the pulse transmission time is obtained.

[0026] As a further technical limitation, the non-contact blood pressure measurement is Wherein, PTT is pulse transit time, SBP is the systolic blood pressure, and DBP is the diastolic blood pressure.

[0027] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0028] This invention utilizes Landmark masking and Region of Interest (ROI) extraction techniques based on high signal-to-noise ratio (SNR) skin patch selection, combined with an automatically optimizing variational mode decomposition algorithm, to acquire high SNR IPPG signals. By using IPPG pulse waves to guide the EVM (Electronic Vessel) within the correct bandpass filtering range for ideal EVM motion amplification, it achieves non-contact acquisition of pulse waves and non-contact cardiac impact signals (PTT). This enables non-contact blood pressure acquisition using only head video, possessing significant application value for non-contact real-time monitoring of vital signs. Attached Figure Description

[0029] The accompanying drawings, which form part of this embodiment, are used to provide a further understanding of this embodiment. The illustrative embodiments and their descriptions are used to explain this embodiment and do not constitute an improper limitation of this embodiment.

[0030] Figure 1 This is a schematic diagram of the non-contact blood pressure measurement system in an embodiment of the present invention;

[0031] Figure 2(a) is a schematic diagram of the 68 face markers in the face landmark of this invention.

[0032] Figure 2(b) is a face mask diagram in an embodiment of the present invention;

[0033] Figure 3 This is a schematic diagram of ROI segmentation in an embodiment of the present invention;

[0034] Figure 4 This is a schematic diagram of the non-contact pulse wave spectrum in an embodiment of the present invention;

[0035] Figure 5 This is a schematic diagram of the non-contact pulse wave Hilbert instantaneous frequency transformation in an embodiment of the present invention;

[0036] Figure 6 This is a schematic diagram of the pulse wave peak value in an embodiment of the present invention;

[0037] Figure 7 This is a schematic diagram of a high signal-to-noise ratio human face skin region in an embodiment of the present invention;

[0038] Figure 8(a) is a schematic diagram of the face and marker position at time t1 in an embodiment of the present invention;

[0039] Figure 8(b) is a schematic diagram of the face and marker positions at time t2 in an embodiment of the present invention;

[0040] Figure 9 This is a schematic diagram of pulse transmission time extraction in an embodiment of the present invention;

[0041] The components include: 1. Industrial camera; 2. Halogen lamp; 3. PC processor; 4. Tester; and 5. Data transmission cable. Detailed Implementation

[0042] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0043] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0044] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0045] This invention provides a non-contact blood pressure measurement system.

[0046] To address the challenge of non-contact blood pressure measurement, this embodiment proposes a region of interest extraction technique based on landmark masking and high signal-to-noise ratio (SNR) skin patch selection. Combined with an automatically optimizing variational mode decomposition algorithm, it achieves high SNR IPPG signal acquisition. Using the IPPG pulse wave to guide the EVM (Electronic Video Magnification) within the correct bandpass filtering range, ideal EVM motion amplification is achieved, enabling non-contact acquisition of pulse waves and non-contact cardiac impulse signals (PTT). This allows for non-contact blood pressure measurement using only head video. This embodiment also utilizes Eulerian Video Magnification (EVM) technology to amplify minute motion and color changes, completing the non-contact acquisition of BCG signals and ultimately achieving non-contact blood pressure measurement.

[0047] Existing technologies utilize PTT (post-contact pulse-thrust) features to calculate blood pressure. However, due to significant individual variability in facial skin tissue blood volume waves, it's impossible to acquire PTT using pulse waves from different facial regions. This embodiment uses non-contact facial pulse waves and non-contact head cardiac impact signals to acquire PTT for the same detection area (head). For complex facial conditions, especially when there is interference from non-skin pixels such as hair, glasses, or beards, how can noise be suppressed as much as possible during ROI selection and physiological signal extraction? This embodiment employs a high signal-to-noise ratio (SNR) skin region selection algorithm to remove non-skin color pixels, obtaining a highly stable and high SNR ROI. Non-contact pulse waves exhibit second and higher harmonic leakage when passing through the carotid artery; this embodiment uses a high SNR skin region selection algorithm to select skin color regions with high SNR and no harmonic leakage as regions of interest.

[0048] When extracting pulse wave signals, noise inevitably exists due to face movement and changes in lighting direction. To further remove this noise, this embodiment employs an automatically optimizing variational mode decomposition algorithm. Compared to the standard variational mode decomposition algorithm, the optimizing variational mode decomposition algorithm can automatically select the mode component most likely to contain blood volume information as the signal output, achieving noise removal. This results in obtaining pulse wave information with a high signal-to-noise ratio, enabling highly accurate non-contact pulse wave detection. EVM (Electronic Vibration Detection) is prone to introducing noise, leading to significant noise in non-contact BCG signals; this embodiment uses non-contact pulse waves to guide the EVM within the correct bandpass frequency domain.

[0049] like Figure 1 The non-contact blood pressure measurement system shown uses an industrial camera 1 to capture facial images and a halogen lamp 2 (whose light wavelength range includes both visible and invisible light) to illuminate the face and obtain better image quality. The captured facial video sequence is input to a PC 3 via a USB 3.0 data cable. Each frame of the facial image is acquired from the continuous video sequence through frame-by-frame processing for image processing.

[0050] In this embodiment, a Landmark model generated by the Ensemble of Regression Trees (ERT) method based on gradient enhancement learning is used for face recognition. This algorithm trains a series of labeled face images using cascaded regression factors to generate a face recognition model. The Landmark model yields 68 face markers as shown in Figure 2(a). These markers are then processed to obtain a face mask as shown in Figure 2(b). Connecting the Landmark markers [0-16, 45, 46, 47, 42, 39-40, 36] forms a closed-loop region image. The skin within this region represents the initial ROI, but the selected region also contains a small number of environmental pixels and non-skin pixels. This embodiment employs a high signal-to-noise ratio skin region selection algorithm to retain skin region information while minimizing non-skin interference. The initial ROI region is first processed as follows... Figure 3 The 10*10 equidistant segmentation is shown; for skin blocks with more than 100 black pixels, they are considered low signal-to-noise ratio skin regions that may introduce noise and are therefore removed.

[0051] Each pixel is composed of red, green, and blue. The RGB values ​​of each processed skin block are obtained. By performing the same processing on each frame, the RGB three-channel signal that changes over time is obtained. Since there is a lot of useless information in the signal, noise reduction processing is required.

[0052] This embodiment uses the CHROM algorithm to remove noise from the linear combination of RGB signals. The specific process is as follows:

[0053] The R, G, and B channels of the continuous video frame pixels are separated and extracted, and then the R, G, and B channel signals are normalized.

[0054] The normalized signal values ​​are projected onto two orthogonal chromaticity vectors X. chrom and Y chrom , that is, X chrom (t)=3x r (t)-2x g (t), Y chrom (t)=1.5x r (t)+x g (t)-1.5x b (t); Output S(t) = X f -αY f ; where x r x g x b These are the R channel, G channel, and B channel signals, respectively; X f Y f It is X chromY chrom The signal after being filtered by a fifth-order Butterworth bandpass filter (passband: 0.7Hz~4Hz); α is X f Y f The ratio of standard deviations; S(t) is the IPPG signal processed by the CHROM algorithm.

[0055] After the CHROM algorithm, the following is obtained: Figure 4 The diagram shows a non-contact pulse wave spectrum, where A represents the noisy modal component and B represents the pulse wave information modal component. Most of the noise in the original IPPG signal is eliminated; however, due to the influence of complex environments, some noise may remain in the IPPG signal that cannot be eliminated. This embodiment proposes an optimized variational mode decomposition (OVMD) to obtain the main frequency domain of the pulse wave from S(t) non-contactly, thus achieving the correct heart rate frequency. The constrained variational model involved in the VMD algorithm is as follows:

[0056]

[0057]

[0058] Among them, {u k}={u1,u2,…,u k} and {w k}={w1,w2,…,w k} represents the k-th modal component and its corresponding center frequency, for a total of K modes; δ(t) represents partial derivative operation; δ(t) represents the unit impulse function; j represents the imaginary unit; * represents convolution operation; f represents the target signal.

[0059] By introducing a penalty factor α and Lagrange multipliers λ to solve the variational constraint problem, the resulting augmented Lagrange expression is as follows:

[0060]

[0061] The alternating multiplier algorithm is used for iterative updating and solving, updating {u} in the frequency domain. k},{w k The process of updating and iteratively solving},λ is as follows: Initialization and n; according to and Implement u respectively k w k Iterative updates of λ and λ are performed until the iteration termination condition is met. Output modal components; where τ is the fidelity coefficient; ∧ represents the Fourier transform; n is the number of iterations; ε is the discrimination precision, and ε > 0; the superscript n is the number of iteration steps, and the subscript k represents the current number of modes.

[0062] Before performing mode decomposition, parameters need to be set. An inappropriate mode number K may lead to mode aliasing and loss of signal frequency components. Since the main frequency range of non-contact pulse waves is relatively small (approximately between 0-4Hz), this embodiment sets the mode number K to 2. After variational mode decomposition, two modal components are obtained. The mode containing pulse wave information is selected from these two modal components; that is:

[0063]

[0064]

[0065]

[0066] Where, h(t) i Let represent the modal components after variational mode decomposition, where i is the i-th modal component after variational mode decomposition. It is h(t) i The signal after Hilbert transform, where arctan() is the arctangent function. Representing the instantaneous phase, ω(t) i Indicates instantaneous frequency.

[0067] In a short period of time, the human heart rate will not change drastically or fluctuate significantly, such as Figure 5 The more stable the Hilbert instantaneous frequency signal of the pulse wave shown, the greater the likelihood that it contains pulse wave information, i.e.:

[0068] R i =std(ω(t)) i ) / mean(ω(t) i )

[0069] IPPG = argmin(R 1 ,R 2 )

[0070] Where std(·) is the standard deviation, mean(·) is the mean, which is used in this embodiment to represent the stationarity of the signal, argmin represents the index of the minimum value, and IPPG is the modal component containing pulse wave information returned last, thus completing the non-contact pulse wave extraction for each skin block.

[0071] like Figure 6As shown, the peak summation of the non-contact pulse wave extracted from each skin patch is performed. The skin patch with the most peaks and frequencies is identified as having a high signal-to-noise ratio, while other skin patches with different peak frequencies are discarded. A common RGB signal is obtained from all the remaining skin patches, and then CHROM and OVMD are performed to obtain the final non-contact pulse wave signal.

[0072] The EVM action method uses a Laplace pyramid for spatial filtering and an ideal bandpass filter for time filtering. The filtered signal δ(t) is then obtained, and the amplification formula is:

[0073] I(x,t)=f(x+δ(t)) t>0

[0074] I(x,t)=f(x) t=0

[0075] Where δ(t) is the transformed signal, and I(x,t) represents the intensity of the image at position x and time t.

[0076] Performing a first-order Taylor expansion of I(x,t)=f(x+δ(t))t>0, we get The result of applying an ideal bandpass filter to I(x,t) can be obtained. The bandpass filter is set to be 0.3 Hz around the peak IPPG frequency at 10 seconds. For example, if the peak IPPG frequency is 1.2 Hz (heart rate 72 bpm), the bandpass filter is set to 0.9–1.5 Hz.

[0077] The bandpass filtered portion is amplified by a factor of α and then added back to the original signal I(x,t), as shown in the following formula:

[0078]

[0079]

[0080]

[0081] in, Let x be the image intensity at time t after local magnification by (1+α).

[0082] After obtaining the image sequence after EVM, the Landmark face marker algorithm is used to perform face acquisition on the images, resulting in the following: Figure 7 The high signal-to-noise ratio facial skin region is shown.

[0083] This embodiment selects the first 35 marker points because this can remove noise caused by mouth and eye movements. Figure 8(a) shows the face and marker point positions at time t1, and Figure 8(b) shows the face and marker point positions at time t2. The non-contact BCG acquisition formula is:

[0084]

[0085] Where x and y are the horizontal and vertical coordinates of the marker point, respectively, and the subscript i of x and y is the serial number of the landmark point.

[0086] After obtaining the BCG waveform, smooth it and use peak detection to obtain the waveform as shown below. Figure 9 The peak indices of BCG and IPPG shown are divided by the sampling frequency to obtain PTT;

[0087] There is a linear relationship between transmission time and blood pressure value. Through experimental fitting, the blood pressure formula is:

[0088]

[0089]

[0090] Wherein, PTT is pulse transit time, SBP is the systolic blood pressure, and DBP is the diastolic blood pressure.

[0091] This embodiment accurately extracts high signal-to-noise ratio (SNR) ROIs by combining Landmark face marker masking with high SNR skin color region detection. The CHROM and OVMD algorithms are used to denoise the original IPPG signal, resulting in a precise IPPG signal. The IPPG signal is then used to guide EVM reconstruction of the face video. Non-contact BCG is obtained by acquiring the coordinate transformation of the face markers. PTT (post-traumatic stress test) is performed using the acquired dual signals, and finally, the blood pressure value is calculated; thus, non-contact facial blood pressure detection is achieved.

[0092] The above description is merely a preferred embodiment of this practice and is not intended to limit the scope of this practice. Various modifications and variations can be made to this practice by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of this practice should be included within the protection scope of this practice.

Claims

1. A non-contact blood pressure measurement system, characterized by, Comprise: An acquisition module configured to acquire a face image; A processing module configured to train the acquired face image based on a level regression factor, generate a face recognition model, extract a region of interest in the acquired face image according to the generated face recognition model, and perform noise reduction processing on the region of interest in the face image according to an optimization variable mode decomposition method to obtain an imaging photoelectric plethysmography signal; The face recognition model uses Landmark The model, through the aforementioned Landmark The model acquires several facial markers in a face image, performs graphic processing on the acquired facial markers to obtain a face mask; processes the obtained face mask to obtain the region of interest in the face image; the region of interest in the face image contains environmental pixels and non-skin pixels, and a high signal-to-noise ratio skin region selection method is used to preserve skin region information and remove non-skin interference in the obtained region of interest in the face image; Sum the peak values of the non-contact pulse waves extracted from each skin block, wherein the skin block with the most peak value and the most number of times is the skin block with high signal-to-noise ratio, and the skin blocks with other peak value sums are excluded; Obtain the common RGB signal of all the remaining skin blocks, perform CHROM and OVMD, and obtain the final non-contact pulse wave signal; The measurement module is configured to process the obtained imaging photoplethysmography signal by using an Eulerian image magnification technique to obtain a non-contact cardiac impact signal and a pulse transit time, and to realize non-contact blood pressure measurement; the acquisition formula of the non-contact cardiac impact signal is: ; wherein x and y are the horizontal and vertical coordinates of the marker point, x and y the subscript i is landmark the marker point serial number; the obtained non-contact cardiac impact signal is subjected to waveform smoothing processing, the peak value indexes of the non-contact cardiac impact signal and the imaging photoplethysmography signal are obtained by using peak value detection, and the pulse transit time is obtained in combination with a sampling frequency; The non-contact blood pressure measurement is ; wherein PTT is pulse transit time, SBP is a high value of blood pressure, and DBP is a low value of blood pressure.

2. A non-contact blood pressure measurement system as defined in claim 1, wherein, In the acquisition module, a face video sequence is collected, the collected face video sequence is frame-processed, and the face image of each frame is obtained in the continuous video sequence, i.e. the face image is acquired.

3. A non-contact blood pressure measurement system as defined in claim 1, wherein, The constraint variation model in the optimization variable mode decomposition method is where, and are the kth modal component and its corresponding center frequency, respectively, with K total modes; is the partial derivative operation; is the unit impulse function; denotes the imaginary unit; * denotes the convolution operation; is the target signal; a penalty factor and a Lagrange multiplier are introduced to solve the variational constraint problem, resulting in the augmented Lagrange expression as: The alternating direction method of multipliers is used to update the iteration solution, and the iteration update is performed in the frequency domain .

4. A non-contact blood pressure measurement system as claimed in claim 3, wherein, The process of the update iteration solution is: initialization , , and ; according to , and , the iterative updates of , and are realized respectively, until the iteration termination condition is satisfied, and the modal component is output; wherein is a fidelity coefficient; represents a Fourier transform; n is the number of iterations; is a discrimination accuracy, and ; the superscript is the number of iteration steps, and the subscript represents the current modal number.

Images