CN-120938386-B - Method for rapidly detecting human physiological signals

Abstract

The application relates to a human physiological signal rapid detection method, belonging to the field of human signal detection, in particular to a method for detecting human signals through facial information, which is provided by the scheme of the application, has shorter measurement time, and can acquire IPPG signals by sliding and intercepting IPPG original signals with 1.8s cut-off time after acquiring original IPPG signals, wherein the sliding frame number is 1 frame frequency, and performing polynomial fitting and averaging on each segment of fragment set; the high-precision human physiological signal can be obtained after the noise is corrected and weighted due to low-frequency noise caused by relative motion and illumination change, and the shortest calculation time of the human physiological signal is 1.8 seconds.

Inventors

• ZHOU WEI
• LI YUPENG
• ZHANG JING
• WANG JUN
• NING XUEMEI
• LI QIAN
• XIA XIGANG
• YU TAO
• JIN WEI
• LIU WENJIN
• WANG YANG
• WANG HE
• QI ZHAO
• WANG XIN

Assignees

• 中浦慧联信息科技(上海)有限公司

Dates

Publication Date

20260508

Application Date

20250901

Claims (3)

1. 1. The rapid detection method of the human physiological signals is characterized by comprising the following steps of: S1, acquiring face videos, wherein an image acquisition module adopts a high-bit-depth camera and is used for recording high-bit-depth video data of face scattered light signals; S2, extracting human facial feature nodes of the high-bit-depth video data obtained in the step S1, automatically dividing a high signal-to-noise ratio area based on the feature nodes according to the vascular distribution of the human face, and extracting an original IPPG signal, wherein the original IPPG signal can be expressed as follows; (1) reflecting the time-dependent variation of IPPG signal, where Is the gray value corresponding to the camera RGB channel at the kth pixel, is a matrix of 3 x N, N is the length of t, Is the noise of the sound and, Is the vector of the blood flow pulse BVP, reflects the intensity of the response of the camera RGB channel to the BVP signal, Is the intensity of the incident light, Is a BVP signal, t is time information; s3, extracting BVP signals from IPPG signals obtained in the step S2, wherein the BVP signals mainly comprise the following steps; S31 original signal obtained by equation (1) The method comprises the steps of dividing into fragment sets with the length of 1.8s which are connected end to end; s32 for the original signal processed in S31 The signal after 1.8s sliding window interception is expressed in discrete form: (2) Wherein, the Representing the original signal of the discrete sampled k (k=r, G, B) channel, n being the nth sample point, N is the length of the original signal; in the formula (1) Wherein Is a multiplicative signal that is independent of time, Is a time dependent multiplicative signal; the BVP signal is represented by a signal, The part can be regarded as 'noise-like' with weak magnitude, and the 'noise-like' is fitted and eliminated by the following data processing mode " ; Using a cubic polynomial Fitting A part in which Is a mathematical expression of a cubic polynomial, and in order to minimize the error of the fitting result, it is necessary to make And (3) with The sum of squares of the residuals is minimal: (3) For a pair of Differentiating and making the differentiation equal to 0, the following system of equations can be obtained: (4) Equation 4 is expressed in matrix form: (5) Wherein: , , t represents the transpose of the matrix; The matrix is a Van der Monte matrix, pair Matrix QR decomposition can be obtained: (6) Wherein Q is a unitary matrix of NxN and R is an upper triangular matrix of Nx4, and obtaining coefficients b of a cubic polynomial from the formula (6) to obtain a fitting trend of noise components ; S33, obtaining BVP signals through weighting processing; Further eliminate the acquisition in S32 From hemoglobin absorption spectrum, red light carries a large amount of noise information and a small amount of BVP information, green light carries a large amount of BVP information and a small amount of noise information, blue light carries a noise information and a small amount of BVP information, and after the original signal is subjected to polynomial fitting noise and elimination, a weighted signal processing method is adopted, namely, a red channel signal is subtracted from a green channel signal, namely: (7) in the formula, To weight the processed blood flow pulsation signal, And The green and red channel signals of the BVP signal obtained in the formula (6) are respectively; s4, fitting and weighting the obtained polynomial by the step S3 And calculating the heartbeat frequency and the blood oxygen saturation degree by the signals, and finishing the detection of human physiological signals.
2. 2. The method for rapidly detecting human physiological signals according to claim 1, wherein the high-order depth camera in the step S1 is a camera with a bit depth of 8 bits or 10 bits.
3. 3. A rapid human physiological signal detection method according to claim 1 is characterized in that in the step S2, the high signal-to-noise ratio area is the forehead, namely the upper part of the arch of the eyebrow and the forehead, the area is the branch of the internal carotid artery, the two sides of the nasal wings and the nasal tip are the main branch of the external carotid artery, the region from the angle of the lower jaw to the angle of the oral cavity, the artery in the region directly receives the blood flow of the external carotid artery, and the branch of the maxillary artery is arranged in front of the tragus, so that the pulsation is obvious.

Description

Method for rapidly detecting human physiological signals Technical Field The invention belongs to the field of human body signal detection, and particularly relates to a method for detecting human body signals through facial information. Background Many medical devices of the modern day have been greatly altered by advances in technology, but are still comparable to 50 years ago in the monitoring of key physiological parameters. Heart rate is typically monitored by an Electrocardiogram (ECG) that has been applied clinically for more than a century, the nurse calculating the respiration rate by counting the movements of the patient's chest over a predetermined period of time, and measuring blood pressure by means of inflatable armbands. The last major change was the introduction of the 19 th century 70 oximeter for measuring peripheral blood oxygen saturation. These conventional methods have been well established in clinical applications, but they have disadvantages as well. For example, ECG and oximeter probes may be uncomfortable for the patient to wear for a long period of time, may limit the patient's movement, and may increase the risk of infection. Therefore, the method for measuring the key physiological parameters of the human body, which can be continuously and without physical contact, has great clinical application prospect. The field of non-contact critical physiological parameter monitoring has been rising for the past 10 years, and more studies have demonstrated that imaging photoplethysmography (IPPG, imaging PPG) can estimate critical physiological parameters of the human body by means of a video camera. In 2014, oxford team used an autoregressive model to estimate key signals including respiration and heart rate, finding that the estimated heart rate was comparable to the reference values measured simultaneously by ear and finger oximeters. An overview of non-contact respiration measurement was published by Kranjec in 2014, the authors compared with the new method of non-contact measurement and focused on RGB imaging-based methods, which they consider the most important advantage of this method to be low cost and have the potential to monitor multiple targets simultaneously, but the complexity of the processing method and low time resolution are drawbacks of this method. In the prior art, the signal-to-noise ratio of BVP signals is guaranteed through longer facial video acquisition time, and the typical minimum signal acquisition time is 20-30 seconds, however, in the application scene of the intelligent canteen system, the real-time detection efficiency of human physiological signals is required to be equivalent to the detection efficiency of automatic calculation of meal prices, analysis of meal nutritional ingredients and the like, and the long detection time of the physiological signals can greatly restrict the operation efficiency of the intelligent canteen system, so that the application value of human physiological signal detection is reduced. There is therefore a great need in the art to address the above-mentioned problems. Disclosure of Invention The invention aims to solve the technical problem of providing a method for solving the problems in the background technology. The rapid detection method of the human physiological signals is characterized by comprising the following steps of: S1, acquiring face videos, wherein an image acquisition module adopts a high-bit-depth camera and is used for recording high-bit-depth video data of face scattered light signals; S2, extracting human facial feature nodes of the high-bit-depth video data obtained in the step S1, automatically dividing a high signal-to-noise ratio area based on the feature nodes according to the vascular distribution of the human face, and extracting an original IPPG signal, wherein the original IPPG signal can be expressed as follows; (1) reflecting the time-dependent variation of IPPG signal, where Is the gray value corresponding to the camera RGB channel at the kth pixel, is a matrix of 3 x N, N is the length of t,Is the noise of the sound and,Is the vector of the blood flow pulse BVP, reflects the intensity of the response of the camera RGB channel to the BVP signal,Is the intensity of the incident light,Is a BVP signal, t is time information; s3, extracting BVP signals from IPPG signals obtained in the step S2, wherein the BVP signals mainly comprise the following steps; S31 original signal obtained by equation (1) The method comprises the steps of dividing into fragment sets with the length of 1.8s which are connected end to end; s32 for the original signal processed in S31 The signal after 1.8s sliding window interception is expressed in discrete form: (2) Wherein, the Representing the original signal of the discrete sampled k (k=r, G, B) channel, n being the nth sample point,N is the length of the original signal; in the formula (1) WhereinIs a multiplicative signal that is independent of time,Is a time depend