A Pulse Wave Calibration Method Based on Millimeter-Wave Radar

Abstract

The present invention discloses a calibration method for pulse waves based on millimeter-wave radar. By synchronously collecting radar pulse waves and standard pulse waves, they are respectively stored in a radar pulse wave matrix and a standard pulse wave matrix. Subsequently, normalization processing is performed on the two groups of pulse wave signals to eliminate amplitude differences; through differential and ternary quantization processing, the characteristics of pulse wave changes are extracted, and the alignment of the radar pulse wave and the standard pulse wave is achieved through cross-correlation operation. Based on the aligned signals, a calibration matrix is generated by least squares fitting. In actual use, the signal distortion detected by the radar pulse wave can be corrected based on this matrix to achieve real-time calibration, eliminate signal deviations caused by individual differences, equipment interference, and long-distance detection, thereby ensuring the accuracy and stability of long-term health monitoring. The method of the present invention is efficient and has a small computational burden, and is applicable to health management and remote monitoring applications.

Description

Technical Field

[0001] The present invention relates to the technical field of pulse wave calibration, and particularly relates to a pulse wave calibration method based on millimeter wave radar. Background Art

[0002] With the increasingly prominent trend of population aging, the health problems of the elderly have received widespread social attention. Continuously monitoring the health indicators of the elderly is of great significance for the health management and disease prevention of the elderly. As an important basis for judging the human health status, monitoring the pulse wave can provide health information in multiple aspects such as heart function, blood vessel health, blood circulation status, and human physiological and pathological states. However, currently, the mainstream pulse wave monitoring solutions are mainly contact-based, which are not suitable for long-term monitoring. Therefore, the relevant fields have turned their attention to non-contact pulse wave monitoring solutions, among which the pulse wave monitoring solution based on millimeter wave radar is more favored.

[0003] In practical applications, the pulse wave mainly assists in health diagnosis in two ways. One is to directly monitor the pulse wave waveform, and medical staff interpret the waveform to assist in health analysis or disease diagnosis; the other is to automatically extract waveform features based on the monitored pulse wave and analyze health indicators (such as blood pressure, heart rate, heart rate variability, etc.) according to an expert model. No matter which implementation method, it highly depends on the accuracy of pulse wave monitoring. However, due to reasons such as remote detection, individual differences, and internal interference in non-contact monitoring solutions such as millimeter wave radar, the monitored pulse wave has an uncertain degree of distortion, which affects subsequent waveform interpretation or feature extraction, and may thus lead to diagnostic deviations. Summary of the Invention

[0004] Technical Objective: Aiming at the deficiencies of the existing millimeter wave radar for detecting pulse waves, the present invention discloses a pulse wave calibration method based on millimeter wave radar, which can reduce the distortion of the pulse wave and improve the accuracy of pulse wave monitoring.

[0005] Technical Solution: To achieve the above technical objective, the present invention adopts the following technical solution:

[0006] A pulse wave calibration method based on millimeter wave radar specifically includes the following steps:

[0007] Synchronously collect the radar pulse wave monitored by a group of radar units and the standard pulse wave monitored by a standard unit, and store them in a radar pulse wave matrix and a standard pulse wave matrix respectively. The standard unit is a contact-type pulse wave sensing unit;

[0008] Perform normalization processing on the radar pulse wave matrix and the standard pulse wave matrix to eliminate the amplitude difference;

[0009] Align the waveform of the normalized radar pulse wave matrix and the normalized standard pulse wave matrix to obtain a reference pulse wave matrix aligned with the normalized radar pulse wave matrix;

[0010] Downsample the normalized radar pulse wave matrix and the reference pulse wave matrix to obtain a downsampled radar pulse wave matrix and a downsampled reference pulse wave matrix;

[0011] Fit the downsampled radar pulse wave matrix and the downsampled reference pulse wave matrix by the least squares method to generate a calibration matrix and store it for subsequent calibration of the radar pulse wave;

[0012] In actual use, save the collected pulse wave to the pulse wave matrix, and downsample the radar pulse wave in the pulse wave matrix to obtain a downsampled pulse wave matrix;

[0013] Use the stored calibration matrix to correct the downsampled pulse wave matrix to obtain a downsampled corrected pulse wave matrix;

[0014] Calculate a correction factor matrix based on the downsampled pulse wave matrix and the downsampled corrected pulse wave matrix, and perform linear interpolation on the correction factor matrix to obtain a fully corrected factor matrix;

[0015] Correct the pulse wave matrix based on the fully corrected factor matrix to obtain a corrected pulse wave matrix, and complete the calibration of the pulse wave.

[0016] Preferably, aligning the waveform of the normalized radar pulse wave matrix and the normalized standard pulse wave matrix to obtain a reference pulse wave matrix aligned with the normalized radar pulse wave matrix specifically includes the following steps:

[0017] Perform differential processing on the normalized radar pulse wave matrix and the normalized standard pulse wave matrix to obtain a differential radar pulse wave matrix and a differential standard pulse wave matrix respectively;

[0018] Perform ternary quantization processing on the differential radar pulse wave matrix and the differential standard pulse wave matrix to quantize the radar pulse wave and the standard pulse wave between -1, 0, and 1;

[0019] Perform cross-correlation operation row by row on the ternary-quantized differential radar pulse wave matrix and the ternary-quantized differential standard pulse wave matrix to obtain their cross-correlation function, and calculate the offset number for each cross-correlation function to obtain an offset number vector;

[0020] Crop the normalized standard pulse wave matrix according to the offset number vector to obtain a reference pulse wave matrix, and the reference pulse wave matrix is aligned with the normalized radar pulse wave matrix.

[0021] Preferably, the differential processing is performed on the normalized radar pulse wave matrix and the normalized standard pulse wave matrix to obtain the differential radar pulse wave matrix and the differential standard pulse wave matrix, and the specific calculation formulas are as follows:

[0022]

[0023] Among them, RPWD(i, j) represents the j-th sampling point of the i-th row of radar pulse wave in the differential radar pulse wave matrix, and RPWN(i, j) represents the j-th sampling point of the i-th row of radar pulse wave in the normalized radar pulse wave matrix; SPWD(i, j) represents the j-th sampling point of the i-th row of standard pulse wave in the differential standard pulse wave matrix, and RPWN(i, j) represents the j-th sampling point of the i-th row of radar pulse wave in the normalized standard pulse wave matrix. i and j respectively represent the row index and the column index.

[0024] Preferably, the ternary quantization processing is performed on the differential radar pulse wave matrix and the differential standard pulse wave matrix, and the specific calculation process formula for quantizing the radar pulse wave and the standard pulse wave between -1, 0, and 1 is as follows:

[0025]

[0026] Among them, RPWT(i, j) represents the j-th sampling point of the i-th row of radar pulse wave in the differential radar pulse wave matrix after ternary quantization processing, and RPWD(i, j) represents the j-th sampling point of the i-th row of radar pulse wave in the differential radar pulse wave matrix. G TR represents the quantization threshold for the differential radar pulse wave matrix; SPWT(i, j) represents the j-th sampling point of the i-th row of radar pulse wave in the differential standard pulse wave matrix after ternary quantization processing, and SPWD(i, j) represents the j-th sampling point of the i-th row of radar pulse wave in the standard radar pulse wave matrix. G TS represents the quantization threshold for the differential standard pulse wave matrix. i and j respectively represent the row index and the column index.

[0027] Preferably, the cross-correlation operation is performed on the differential radar pulse wave matrix after ternary quantization processing and the differential standard pulse wave matrix after ternary quantization processing to obtain the cross-correlation function of the two, and the specific process for calculating the offset number vector for each cross-correlation function is as follows:

[0028] The cross-correlation operation is performed on the differential radar pulse wave matrix after ternary quantization processing and the differential standard pulse wave matrix after ternary quantization processing to obtain the cross-correlation function of the two, and the calculation formula is as follows:

[0029]

[0030] Wherein, R i represents the cross-correlation function of the i-th row of the pulse wave in the differential radar pulse wave matrix after three-value quantization processing and the differential standard pulse wave matrix after three-value quantization processing, k represents the offset, and N R represents the number of sampling points of the radar pulse wave, and N S represents the number of sampling points of the standard pulse wave;

[0031] Peak search is performed on each cross-correlation function in turn, and the k value corresponding to the peak position is used as the offset number and saved to the offset number vector N OFF Among them, the offset number of the i-th row of the pulse wave is:

[0032]

[0033] Wherein, represents the k value corresponding to when R i (k) takes the maximum peak value.

[0034] Preferably, the normalized standard pulse wave matrix is cropped according to the offset number vector to obtain a reference pulse wave matrix. The specific calculation process for aligning the reference pulse wave matrix with the normalized radar pulse wave matrix is as follows:

[0035]

[0036] Wherein, bidx represents the starting index of the reference pulse wave in the standard pulse, eidx represents the ending index of the reference pulse wave in the standard pulse wave, BPWN(i) represents the i-th reference pulse wave, and SPWN(i, bidx:eidx) represents the sampling values of the i-th row and the bidx to eidx index range of the SPWN matrix.

[0037] Preferably, the calculation formula of the calibration matrix is as follows:

[0038] C = (RPWC T × RPWC) -1 × RPWC T × BPWC

[0039] Wherein, C represents the calibration matrix, RPWC represents the downsampled radar pulse wave matrix, and BPWC represents the downsampled reference pulse wave matrix.

[0040] Preferably, the calculation formula of the correction factor matrix is as follows:

[0041] SFC(i, j) = PWA(i, j) / PWC(i, j)

[0042] Among them, SFC(i,j) is the element in the i-th row and j-th column of the correction factor matrix SFC, representing the degree of scaling of the j-th sampling point in the i-th row of the downsampled radar pulse wave after correction. PWA(i,j) represents the j-th sampling point in the i-th row of the radar pulse wave in the downsampled corrected pulse wave matrix PWA, and PWC(i,j) represents the j-th sampling point in the i-th row of the radar pulse wave in the downsampled pulse wave matrix PWC;

[0043] Perform linear interpolation on the correction factor matrix SFC to obtain the complete correction factor matrix SF. The calculation formula for the i-th row data of the complete correction factor matrix is as follows:

[0044]

[0045] Among them, D represents the extraction factor, and N C represents the number of sampling points of the radar pulse wave in the downsampled pulse wave matrix.

[0046] Preferably, the calculation formula for the corrected pulse wave matrix is as follows:

[0047] PWB = PW·SF

[0048] Among them, PWB represents the corrected pulse wave matrix, PW represents the pulse wave matrix, SF represents the complete correction factor matrix, and · represents the matrix Hadamard product.

[0049] Beneficial effects: A pulse wave calibration method based on millimeter-wave radar provided by the present invention has the following beneficial effects:

[0050] 1. The present invention first synchronously collects the radar pulse wave and the standard pulse wave, and obtains the calibration matrix through processing such as normalization, differential processing, ternary quantization, and cross-correlation operation, and stores the calibration matrix in the device. Then, during formal use, the radar pulse wave is calibrated based on this calibration matrix. On the premise that the device does not malfunction or the measured individual does not change, only one calculation is required to obtain the calibration matrix for calibrating the radar pulse wave. This method can effectively correct the pulse wave signal collected by the millimeter-wave radar, eliminate the influence of device interference and individual differences such as body shape and skin thickness on the signal, reduce pulse distortion, improve the accuracy of pulse wave monitoring, and thus improve the waveform interpretation or feature extraction based on the pulse wave in the later stage.

[0051] 2. The present invention uses downsampling and interpolation techniques in calibration, which can maintain high efficiency and low power consumption during the calibration process, ensure the signal stability and accuracy after long-term use of the device, and this correction ability effectively improves the reliability of long-term monitoring. Description of the Drawings

[0052] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following briefly introduces the accompanying drawings required for the description of the embodiments or the prior art.

[0053] Figure 1 This is a detailed flowchart of the pulse wave calibration method based on millimeter-wave radar of the present invention;

[0054] Figure 2 This is the overall process of parameter calibration and application calibration of the pulse wave calibration method based on millimeter-wave radar of the present invention;

[0055] Figure 3 This is the acquisition schematic diagram of synchronously collecting radar pulse waves and standard pulse waves of the present invention;

[0056] Figure 4 This is the flowchart of obtaining a reference pulse wave matrix aligned with the radar pulse wave matrix after normalization processing of the present invention. Detailed implementation manners

[0057] The following more clearly and completely illustrates the present invention by way of a preferred embodiment in conjunction with the accompanying drawings, but the present invention is not limited to the scope of the described embodiments.

[0058] As Figure 1 and Figure 2 shown, the present invention provides a pulse wave calibration method based on millimeter-wave radar. The method can be divided into two major parts: parameter calibration and application calibration. The role of parameter calibration is to generate a calibration matrix by fitting and matching the radar pulse wave and the standard pulse wave. After successful calibration, the calibration matrix will be persistently stored. Usually, the parameter calibration process only needs to be executed once, unless the detection object is changed or the performance changes after long-term use of the device, in which case it is necessary to re-calibrate. Application calibration is to perform real-time correction on the radar pulse wave according to the calibration matrix during actual use to obtain a more accurate pulse wave waveform, and specifically includes the following steps:

[0059] S101. Synchronously collect the radar pulse wave monitored by a group of radar units and the standard pulse wave monitored by a standard unit, and store them in the radar pulse wave matrix and the standard pulse wave matrix respectively. The standard unit is a contact-type pulse wave sensing unit.

[0060] In one embodiment, the standard unit usually selects a contact-type pulse wave sensing unit (such as a photoelectric sensor). The integration method of the standard unit can be fixed integration, that is, each radar is fixedly integrated with a standard unit, or it can be a pluggable integration, that is, the standard module is made into a pluggable module and connected to the radar when the radar needs to perform parameter calibration. In this way, multiple radars can share a standard unit. Since the radar device only needs to perform parameter calibration occasionally, the pluggable integration scheme is a more economical choice.

[0061] As Figure 3 shown, the method for synchronously collecting data is that a smart card integrated with a millimeter-wave radar and a pluggable optoelectronic sensor are used to measure the pulse wave at the same position. When collecting data, the human body posture can be lying flat or sitting still, and the collection time is not less than 5 minutes. A set of radar pulse waves and standard pulse waves are obtained, and the results are stored in the radar pulse wave matrix RPW and the standard pulse wave matrix SPW respectively. The size of the matrix RPW is N PW ×N R . Each row in the matrix represents a radar pulse wave, and the matrix contains a total of N PW radar pulse waves. The number of sampling points for each radar pulse is N R ; the size of the matrix SPW is N PW ×N S . Each row in the matrix represents a standard pulse wave, and the matrix contains a total of N PW standard pulse waves. The number of sampling points for each standard pulse wave is N S .

[0062] The number of radar pulse waves is the same as that of the standard pulse waves, and the sampling frequencies of the pulse waves are also the same. However, the number of sampling points of the standard pulse waves in SPW is greater than that of the radar pulse waves in RPW. The reason is that when selecting the pulse waves, it is impossible to ensure that the data of the radar pulse waves and the standard pulse waves are completely aligned. Therefore, in this scheme, a certain amount of redundant sampling points are reserved before and after the standard pulse waves to facilitate the data alignment operation. The standard pulse waves retain 25% redundant sampling points before and after relative to the radar pulse waves, that is, N S is 1.5 times of N R .

[0063] S102. Normalize the radar pulse wave matrix and the standard pulse wave matrix to eliminate the amplitude difference.

[0064] There may be a large difference in the amplitudes of the monitored radar pulse waves and standard pulse waves. The reasons are as follows: First, the measurement principles of the radar unit and the standard unit are different. Second, the amplitude of the pulse wave is affected by various external factors (such as the measurement point position, human body posture, etc.). Since the absolute amplitude of the pulse wave has great uncertainty, and what we mainly focus on is the shape of the pulse wave, this scheme isolates the influence of the absolute amplitude by performing normalization processing on the radar pulse wave matrix RPW and the standard pulse wave matrix SPW. The normalization processing method is as follows:

[0065]

[0066] where RPWN represents the radar pulse wave matrix after normalization processing, and SPWN represents the standard pulse wave matrix after normalization processing.

[0067] S103. Align the normalized radar pulse wave matrix and the normalized standard pulse wave matrix to obtain a reference pulse wave matrix aligned with the normalized radar pulse wave matrix.

[0068] As Figure 4 shown, there may be data misalignment between the radar pulse wave and the standard pulse wave obtained in the above steps. Before performing fitting and matching, the two must be aligned. The specific steps are as follows:

[0069] S301. Perform differential processing on the normalized radar pulse wave matrix and the normalized standard pulse wave matrix to obtain a differential radar pulse wave matrix and a differential standard pulse wave matrix respectively.

[0070] The specific calculation formula is as follows:

[0071]

[0072] Among them, RPWD(i, j) represents the j-th sampling point of the i-th row of the radar pulse wave in the differential radar pulse wave matrix, RPWN(i, j) represents the j-th sampling point of the i-th row of the radar pulse wave in the normalized radar pulse wave matrix; SPWD(i, j) represents the j-th sampling point of the i-th row of the standard pulse wave in the differential standard pulse wave matrix, RPWN(i, j) represents the j-th sampling point of the i-th row of the radar pulse wave in the normalized standard pulse wave matrix, and i and j represent the row index and column index respectively.

[0073] S302. Perform ternary quantization processing on the differential radar pulse wave matrix and the differential standard pulse wave matrix to quantize the radar pulse wave and the standard pulse wave between -1, 0, and 1.

[0074] To avoid the excessive weight of sampling points with large values in subsequent cross-correlation calculations, here, ternary quantization processing is respectively performed on the differential pulse wave data in the differential radar pulse wave matrix RPWD and the differential standard pulse wave matrix SPWD to obtain the ternary quantization processed radar pulse wave matrix RPWT and the ternary quantization processed standard pulse wave matrix SPWT. The so-called ternary quantization means quantizing the differential pulse wave data to three values of -1, 0, and 1 according to the numerical size. The specific calculation process formula is as follows:

[0075]

[0076] Among them, RPWT(i, j) represents the j-th sampling point of the i-th row of the differential radar pulse wave matrix after ternary quantization processing, RPWD(i, j) represents the j-th sampling point of the i-th row of the radar pulse wave in the differential radar pulse wave matrix, G TRRepresents the quantization threshold for the differential radar pulse wave matrix; SPWT(i,j) represents the j-th sampling point of the i-th row radar pulse wave in the differential standard pulse wave matrix after ternary quantization processing, SPWD(i,j) represents the j-th sampling point of the i-th row radar pulse wave in the standard radar pulse wave matrix, G TS Represents the quantization threshold for the differential standard pulse wave matrix, where i and j represent the row index and column index respectively.

[0077] S303. Perform cross-correlation operations row by row on the differential radar pulse wave matrix after ternary quantization processing and the differential standard pulse wave matrix after ternary quantization processing to obtain N PW cross-correlation functions of two, and calculate the offset number for each cross-correlation function to obtain an offset number vector.

[0078] In one embodiment, the specific process is as follows:

[0079] S3031. Perform cross-correlation operations row by row on the differential radar pulse wave matrix after ternary quantization processing and the differential standard pulse wave matrix after ternary quantization processing to obtain N PW cross-correlation functions of two, and the calculation formula is as follows:

[0080]

[0081] Among them, R i represents the cross-correlation function of the i-th row pulse wave in the differential radar pulse wave matrix after ternary quantization processing and the differential standard pulse wave matrix after ternary quantization processing, k represents the offset, N R represents the number of sampling points of the radar pulse wave, N S represents the number of sampling points of the standard pulse wave;

[0082] S3032. Perform peak search on each cross-correlation function in turn, and use the k value corresponding to the peak position as the offset number, and save it to the offset number vector N OFF The offset number of the i-th row pulse wave is:

[0083]

[0084] Among them, represents the k value corresponding to when R i (k) takes the maximum peak.

[0085] Since sufficient redundant sampling points are reserved on the left side of the standard pulse in step S101, only by shifting the standard pulse to the left by a certain number of points can it be aligned with the radar pulse wave, that is, the values in the N OFF vector must all be positive values.

[0086] S304. Crop the normalized standard pulse wave matrix according to the offset number vector to obtain a reference pulse wave matrix, and align the reference pulse wave matrix with the normalized radar pulse wave matrix.

[0087] The so-called cropping is to remove the front and back sampling points of each standard pulse wave in the normalized standard pulse wave matrix SPWN to align it with the corresponding radar pulse wave. The specific calculation process is as follows:

[0088]

[0089] Among them, bidx represents the starting index of the reference pulse wave in the standard pulse, eidx represents the ending index of the reference pulse wave in the standard pulse wave, BPWN(i) represents the i-th reference pulse wave, and SPWN(i, bidx:eidx) represents the sampling values in the range of the i-th row and the bidx to eidx indexes of the SPWN matrix.

[0090] S104. Downsample the normalized radar pulse wave matrix and the reference pulse wave matrix to obtain a downsampled radar pulse wave matrix and a downsampled reference pulse wave matrix.

[0091] To save computing power and storage space, the present invention first uses the decimation method to perform downsampling on all the pulse waves in the two matrices of the normalized radar pulse wave matrix RPWN and the reference pulse wave matrix BPWN to obtain a downsampled radar pulse wave matrix RPWC and a downsampled reference pulse wave matrix BPWC, and then performs fitting and matching. The number of sampling points of the radar pulse wave before and after downsampling is N R and N C , and the decimation factor is D. Therefore, the following relationship holds:

[0092] N C = N R / D

[0093] The downsampled radar pulse wave matrix RPWC and the downsampled reference pulse wave matrix BPWC obtained by downsampling are respectively:

[0094]

[0095] Among them, RPWC(i, j) represents the j-th sampling point of the i-th row radar pulse wave in the downsampled radar pulse wave matrix, and BPWC(i, j) represents the j-th sampling point of the i-th row radar pulse wave in the reference pulse wave matrix.

[0096] In one embodiment, the decimation factor D is set to 8, that is, the number of sampling points N C of the radar pulse wave after downsampling is one-eighth of the number of sampling points N R of the radar pulse wave before downsampling.

[0097] S105. Fit the downsampled radar pulse wave matrix and the downsampled reference pulse wave matrix by the least squares method to generate a calibration matrix and store it for subsequent calibration of the radar pulse wave.

[0098] Based on the downsampled radar pulse wave matrix RPWC and the downsampled reference pulse wave matrix BPWC, this step uses the least squares criterion to fit the calibration matrix C, that is:

[0099]

[0100] In the formula, the diag(·) operator represents taking the diagonal elements of the matrix. According to the above formula, the analytical result of the calibration matrix C can be calculated as follows, and its size is N C ×N C , and the specific calculation formula is as follows:

[0101] C = (RPWC T × RPWC) -1 × RPWC T × BPWC

[0102] After obtaining the calibration matrix C, store it in the device memory. In the case that the device does not have a fault and an error occurs or the measured individual does not change, this calibration matrix can be used all the time to calibrate and correct the subsequent radar pulse waves that need to be calibrated, and there is no need to recalculate the calibration matrix every time calibration is performed.

[0103] Applying calibration is to correct the radar pulse wave based on the pulse wave calibration matrix C during the formal use process. Since the calibration matrix C is the result of dimensionality reduction (that is, corresponding to the downsampled pulse wave matrix, rather than the original pulse wave matrix), it is also necessary to first perform downsampling processing on the radar pulse wave before applying calibration, and then perform pulse wave correction. After the correction is completed, sampling rate restoration processing is also required. The specific process is as follows:

[0104] S106. In actual use, save the collected pulse wave to the pulse wave matrix and perform downsampling processing on the radar pulse wave in the pulse wave matrix to obtain the downsampled pulse wave matrix.

[0105] First, collect a group of radar pulse waves and save them to the pulse wave matrix PW. The size of PW is N ROW ×N COL , indicating that there are N ROW radar pulse waves in this matrix, and the length of each radar pulse wave is N COL , and N COL is equal to the number of radar pulse wave sampling points N R during parameter calibration.

[0106] The size of the calibration matrix C is N C ×N C while the length of the pulse wave to be calibrated is N COL and it is necessary to first downsample the length of the pulse wave to be calibrated to N C points before using the calibration matrix C for calibration. Since the length N COL of the pulse wave to be calibrated is equal to the radar pulse length N R during the calibration process, the same decimation factor D as during calibration can be used to achieve downsampling.

[0107] Perform sampling processing on all the pulse waves in the pulse wave matrix PW in sequence. The resulting downsampled pulse wave matrix is denoted as PWC. The j-th sampling point of the i-th pulse wave in this matrix is:

[0108] PWC(i,j) = PW(i,D×j)

[0109] S107. Use the stored calibration matrix to correct the downsampled pulse wave matrix to obtain a downsampled corrected pulse wave matrix.

[0110] After obtaining the downsampled pulse wave matrix PWC, the calibration matrix C can be directly used to correct it to obtain a downsampled corrected pulse wave matrix PWA. The calculation method is as follows:

[0111] PWA = PWC × C

[0112] S108. Calculate the correction factor matrix based on the downsampled pulse wave matrix and the downsampled corrected pulse wave matrix, and perform linear interpolation on the correction factor matrix to obtain a fully corrected factor matrix.

[0113] Based on the downsampled pulse wave matrix and the downsampled corrected pulse wave matrix, the correction factor matrix SFC can be obtained. The element in the i-th row and j-th column of SFC represents the scaling degree of the j-th sampling point in the i-th downsampled pulse wave after correction. The calculation method is:

[0114] SFC(i,j) = PWA(i,j) / PWC(i,j)

[0115] where SFC(i,j) is the element in the i-th row and j-th column of the correction factor matrix SFC, representing the scaling degree of the j-th sampling point in the i-th row of the downsampled radar pulse wave after correction. PWA(i,j) represents the j-th sampling point in the i-th row of the radar pulse wave in the downsampled corrected pulse wave matrix PWA, and PWC(i,j) represents the j-th sampling point in the i-th row of the radar pulse wave in the downsampled pulse wave matrix PWC;

[0116] Perform linear interpolation on each row of the correction factor matrix SFC to obtain a fully corrected factor matrix SF. The calculation formula for the i-th row data of the fully corrected factor matrix is as follows:

[0117]

[0118] Among them, D represents the extraction factor, and N C represents the number of radar pulse wave sampling points in the downsampled pulse wave matrix.

[0119] S109. Correct the pulse wave matrix based on the complete correction factor matrix to obtain a corrected pulse wave matrix, and complete the calibration of the pulse wave.

[0120] The calculation formula for the corrected pulse wave matrix is as follows:

[0121] PWB = PW · SF

[0122] Among them, PWB represents the corrected pulse wave matrix, PW represents the pulse wave matrix, SF represents the complete correction factor matrix, and · represents the matrix Hadamard product.

[0123] Each row of the result PWB of the above formula represents a corrected radar pulse wave. Compared with the original uncalibrated pulse wave matrix PW, the radar pulse waves in the corrected pulse wave matrix PWB are closer to the standard pulse wave.

[0124] The above are only the preferred embodiments of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A pulse wave calibration method based on millimeter wave radar, characterized in that: The specific steps include: Synchronously collect a set of radar pulse waves monitored by a radar unit and a standard pulse wave monitored by a standard unit, and store them in a radar pulse wave matrix and a standard pulse wave matrix respectively, wherein the standard unit is a contact pulse wave sensing unit; Normalize the radar pulse wave matrix and the standard pulse wave matrix to eliminate the amplitude difference; Performing waveform alignment on the normalized radar pulse wave matrix and the normalized standard pulse wave matrix to obtain a reference pulse wave matrix aligned with the normalized radar pulse wave matrix; Downsampling the normalized radar pulse wave matrix and the reference pulse wave matrix to obtain a downsampled radar pulse wave matrix and a downsampled reference pulse wave matrix; Fitting the downsampled radar pulse wave matrix and the downsampled reference pulse wave matrix by the least square method, generating a calibration matrix and storing it for subsequent calibration of the radar pulse wave; In actual use, the collected pulse wave is saved in a pulse wave matrix, and the radar pulse wave in the pulse wave matrix is ​​downsampled to obtain a downsampled pulse wave matrix; Correcting the downsampled pulse wave matrix using the stored calibration matrix to obtain a downsampled corrected pulse wave matrix; The correction factor matrix is ​​calculated according to the downsampled pulse wave matrix and the downsampled corrected pulse wave matrix, and the correction factor matrix is ​​linearly interpolated to obtain a complete correction factor matrix; wherein, the calculation formula of the correction factor matrix is ​​as follows: SFC(i,j)=PWA(i,j) / PWC(i,j) Wherein, SFC(i, j) is the i-th row and j-th column element of the correction factor matrix SFC, indicating the scaling degree of the j-th sampling point in the i-th row downsampled radar pulse wave after correction, PWA(i, j) indicates the j-th sampling point in the i-th row radar pulse wave in the downsampled corrected pulse wave matrix PWA, and PWC(i, j) indicates the j-th sampling point in the i-th row radar pulse wave in the downsampled pulse wave matrix PWC; Linear interpolation is performed on the correction factor matrix SFC to obtain the complete correction factor matrix SF. The data calculation formula for the i-th row of the complete correction factor matrix is ​​as follows: Where D represents the extraction factor, N C Indicates the number of radar pulse wave sampling points in the downsampled pulse wave matrix; The pulse wave matrix is ​​corrected based on the complete correction factor matrix to obtain the corrected pulse wave matrix, thereby completing the calibration of the pulse wave. The calculation formula of the corrected pulse wave matrix is ​​as follows: PWB=PW·SF Wherein, PWB represents the modified pulse wave matrix, PW represents the pulse wave matrix, SF represents the complete correction factor matrix, and · represents the matrix Hadamard product.

2. The pulse wave calibration method based on millimeter wave radar according to claim 1, characterized in that: Performing waveform alignment on the normalized radar pulse wave matrix and the normalized standard pulse wave matrix to obtain a reference pulse wave matrix aligned with the normalized radar pulse wave matrix specifically includes the following steps: Performing differential processing on the normalized radar pulse wave matrix and the normalized standard pulse wave matrix to obtain a differential radar pulse wave matrix and a differential standard pulse wave matrix respectively; Performing three-value quantization processing on the differential radar pulse wave matrix and the differential standard pulse wave matrix, and quantizing the radar pulse wave and the standard pulse wave between -1, 0 and 1; Performing cross-correlation operation row by row on the differential radar pulse wave matrix after ternary quantization processing and the differential standard pulse wave matrix after ternary quantization processing to obtain the cross-correlation function of the two, and performing offset calculation on each cross-correlation function to obtain an offset vector; The normalized standard pulse wave matrix is ​​clipped according to the offset number vector to obtain a reference pulse wave matrix, and the reference pulse wave matrix is ​​aligned with the normalized radar pulse wave matrix.

3. The pulse wave calibration method based on millimeter wave radar according to claim 2, characterized in that: The normalized radar pulse wave matrix and the normalized standard pulse wave matrix are subjected to differential processing, and the specific calculation formulas of the differential radar pulse wave matrix and the differential standard pulse wave matrix are obtained as follows: Among them, RPWD(i, j) represents the j-th sampling point of the radar pulse wave in the i-th row in the differential radar pulse wave matrix, and RPWN(i, j) represents the j-th sampling point of the radar pulse wave in the i-th row in the radar pulse wave matrix after normalization; SPWD(i, j) represents the j-th sampling point of the standard pulse wave in the i-th row in the differential standard pulse wave matrix, and RPWN(i, j) represents the j-th sampling point of the radar pulse wave in the i-th row in the standard pulse wave matrix after normalization, and i and j represent row index and column index respectively.

4. The pulse wave calibration method based on millimeter wave radar according to claim 2, characterized in that: The differential radar pulse wave matrix and the differential standard pulse wave matrix are subjected to three-value quantization processing, and the specific calculation process formula for quantizing the radar pulse wave and the standard pulse wave between -1, 0 and 1 is as follows: Wherein, RPWT(i, j) represents the jth sampling point of the radar pulse wave in the i-th row of the differential radar pulse wave matrix after three-value quantization processing, RPWD(i, j) represents the jth sampling point of the radar pulse wave in the i-th row of the differential radar pulse wave matrix, G TR represents the quantization threshold for the differential radar pulse wave matrix; SPWT(i, j) represents the jth sampling point of the radar pulse wave in the ith row of the differential standard pulse wave matrix after three-value quantization processing, SPWD(i, j) represents the jth sampling point of the standard pulse wave in the ith row of the differential standard pulse wave matrix, G TS represents the quantization threshold for the differential standard pulse wave matrix, and i and j represent the row index and column index respectively.

5. The pulse wave calibration method based on millimeter wave radar according to claim 2, characterized in that: The differential radar pulse wave matrix after ternary quantization and the differential standard pulse wave matrix after ternary quantization are cross-correlated row by row to obtain the cross-correlation function of the two. The specific process of calculating the offset number of each cross-correlation function to obtain the offset number vector is as follows: The differential radar pulse wave matrix after ternary quantization processing and the differential standard pulse wave matrix after ternary quantization processing are cross-correlated row by row to obtain the cross-correlation function of the two. The calculation formula is as follows: Among them, R i represents the cross-correlation function of the pulse wave in the ith row in the differential radar pulse wave matrix after ternary quantization and the differential standard pulse wave matrix after ternary quantization, k represents the offset, N R Indicates the number of sampling points of radar pulse wave, N S Represents the number of sampling points of the standard pulse wave; RPWT(i, j) represents the jth sampling point of the radar pulse wave in the i-th row of the differential radar pulse wave matrix after three-value quantization processing, and SPWT(i, j+k) represents the j+kth sampling point of the radar pulse wave in the i-th row of the differential standard pulse wave matrix after three-value quantization processing; Perform peak search for each cross-correlation function in turn, and use the k value corresponding to the peak position as the offset number, and save it to the offset number vector N OFF In , the offset number of the i-th row of pulse wave is: Among them, arg max k (R i (k)) indicates R i (k) Take the k value corresponding to the maximum peak.

6. The pulse wave calibration method based on millimeter wave radar according to claim 2, characterized in that: The normalized standard pulse wave matrix is ​​clipped according to the offset vector to obtain the reference pulse wave matrix. The specific calculation process of aligning the reference pulse wave matrix with the normalized radar pulse wave matrix is ​​as follows: Among them, bidx represents the starting index of the reference pulse wave in the standard pulse, eidx represents the ending index of the reference pulse in the standard pulse wave, BPWN(i) represents the i-th reference pulse wave, and SPWN(i, bidx:eidx) represents the sampling value of the index range from bidx to eidx in the i-th row of the SPWN matrix.

7. The pulse wave calibration method based on millimeter wave radar according to claim 1, characterized in that: The calibration matrix is ​​calculated as follows: C=(RPWC T ×RPWC) -1 ×RPWC T ×BPWC Wherein, C represents the calibration matrix, RPWC represents the downsampled radar pulse wave matrix, and BPWC represents the downsampled reference pulse wave matrix.

Images