A non-contact human blood oxygen saturation detection method, device and equipment

Abstract

The non-contact human blood oxygen saturation detection method, device and computer equipment based on machine learning provided in the embodiments of the present invention obtain the image information of the target human body, determine the facial region of interest ROI area from the image information, divide the facial ROI area into multiple sub-ROIs, evaluate the quality of the facial sub-ROI area, remove the sub-ROI with low signal-to-noise ratio, and at the same time, perform weighted averaging on the retained ROI signals to improve the signal-to-noise ratio of the blood oxygen saturation solution signal. Secondly, in addition to the ratio of ratio parameter used in the traditional algorithm, the present invention also adds the gray value average of the three channels of Red, Green and Blue as input parameters for machine learning training. In addition to considering the reflection of light by the skin, this method also considers the influence of light scattering on blood oxygen saturation estimation, thereby improving the accuracy of non-contact measurement.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular to a non-contact human blood oxygen saturation detection method, device and computer equipment based on machine learning. Background Art

[0002] Blood oxygen saturation is an important life indicator of the human body. It is an indicator that reflects the respiratory and circulatory functions of the human body and measures the oxygen-carrying capacity of the human blood. Once the blood oxygen saturation decreases, it means that the person's life is threatened. There are two methods to measure arterial blood oxygen saturation. One is to draw arterial blood, use a blood gas analyzer to measure the arterial oxygen partial pressure (PaO2), and then obtain SaO2 through the hemoglobin-oxygen dissociation curve. The disadvantage is that repeated blood draws are required to observe changes in the condition. Another method is to use a pulse oximeter monitor for measurement. It is small and flexible, does not require blood drawing, has high accuracy, and is widely used in clinical practice. However, the blood oxygen saturation monitor is a contact measurement, which increases the risk of cross-infection. Therefore, non-contact blood oxygen saturation measurement has received widespread attention.

[0003] The existing non-contact blood oxygen measurement technology is based on the color camera to collect data of the three bands of red, green and blue of the measured target. Then, the ratio of ratio algorithm is used to calculate the ratio of the AC component and DC component of the red and green (or blue) bands respectively. The blood oxygen saturation is proportional to the ratio of the red and green (or) two bands. Finally, after a certain number of sample training, the blood oxygen saturation can be accurately estimated.

[0004] The existing technology has two disadvantages:

[0005] First, the existing technology mainly selects an area on the face of the target to be measured, extracts the RGB signal of the entire area for processing and solves the blood oxygen saturation. However, due to the different distribution of facial blood vessels, different areas of the face have different intensities of blood pulsation, and the blood pulsation in some areas is very weak. If these low signal-to-noise ratio areas are added to the calculation of blood oxygen saturation, the accuracy of blood oxygen saturation solution will be reduced.

[0006] Second, the prior art only uses the ratio of ratio parameters calculated from the red and green (or blue) two-band PPG signals to estimate the blood oxygen saturation, but does not take into account the influence of the light scattering effect on the blood oxygen saturation estimation. Summary of the invention

[0007] In view of this, an embodiment of the present invention provides a non-contact human blood oxygen saturation detection method, device and computer equipment based on machine learning to improve the accuracy of blood oxygen saturation estimation.

[0008] In a first aspect, an embodiment of the present invention provides a non-contact human blood oxygen saturation detection method based on machine learning, comprising:

[0009] Acquire image information of the target human body;

[0010] Determine a facial region of interest (ROI) from the image information;

[0011] Dividing the facial ROI region into a plurality of sub-ROIs;

[0012] Performing quality evaluation on the sub-ROIs, removing sub-ROIs with low signal-to-noise ratios, and performing weighted averaging on the retained ROI signals;

[0013] Based on the ratio of ratio results of any two bands of red, green or blue as input parameters, the grayscale values ​​of the red, green and blue channels in the ROI are added as input parameters of blood oxygen saturation, and multi-parameter machine learning training is used to obtain the estimated value of blood oxygen saturation.

[0014] As an optional solution, the quality evaluation of the sub-ROIs is performed, the sub-ROIs with low signal-to-noise ratios are eliminated, and the weighted average of the retained ROI signals is performed, including:

[0015] The PPG signal of each sub-ROI is extracted, filtered, and then fast Fourier transform is performed, and the quality of each sub-ROI is evaluated in the frequency domain. The quality evaluation formula is shown in the following formula:

[0016]

[0017] In the formula, Metric i is the evaluation coefficient of each sub-ROI, [PR-b, PR+b] is the small frequency range near the heartbeat frequency, and [B1, B2] is the bandpass filter range;

[0018] An evaluation coefficient of each sub-ROI is determined, and sub-ROIs with evaluation coefficients lower than a preset threshold are eliminated, and all remaining sub-ROIs are used to form an ROI subset.

[0019] As an optional solution, performing weighted averaging on the retained ROI signals includes:

[0020] The weighted average of the ROI subsets is used to obtain the PPG signal for blood oxygen calculation. Let the ROI subset be [ROI1, ROI2, ..., ROI m ], the corresponding PPG signals and evaluation coefficients are [PPG1,PPG2,...,PPG m],[Metric1,Metric2,...,Metric m ], where m is the number of sub-ROIs, and the weighted average process is as follows:

[0021] PPG calc =Metric1·PPG1+Metric2·PPG2+…+Metric m ·PPG m (2).

[0022] As an optional solution, based on using the ratio ofratio results of any two bands of red, green or blue as input parameters, the grayscale values ​​of the red, green and blue channels in the ROI are added as input parameters of blood oxygen saturation, and multi-parameter machine learning training is used to obtain an estimated value of blood oxygen saturation, including:

[0023] Calculate the ratio of ratio results and extract PPG respectively calc The time series of the red and green bands are used to calculate the AC signal and DC signal of the two bands respectively, and the ratio of ratio is shown in formula (3);

[0024]

[0025] In the formula, AC calc_red and DC calc_red is the red channel ROI calc The AC and DC components of the regional average over time, AC calc_green and DC calc_green It is the green channel ROI calc The AC and DC components of the regional averages varying with time;

[0026] Extract ROI calc Weighted average PPG in the region calc Gray value average of the three channels of Red, Green and Blue of the signal red Gray green and Gray blue , as the input parameter for calculating blood oxygen saturation;

[0027] The results of the 4 parameters ratio of ratio, Gray red Gray green and Gray blue As input parameters, a certain amount of samples are trained through the machine learning algorithm to obtain the final estimated value of blood oxygen saturation.

[0028] In a second aspect, an embodiment of the present invention provides a non-contact human blood oxygen saturation detection device based on machine learning, comprising:

[0029] An acquisition module is used to acquire image information of a target human body;

[0030] A determination module, used to determine a facial region of interest (ROI) from the image information;

[0031] A division module, used for dividing the facial ROI area into a plurality of sub-ROIs;

[0032] An evaluation module, used to evaluate the quality of the sub-ROIs, remove sub-ROIs with low signal-to-noise ratios, and perform weighted averaging on the retained ROI signals;

[0033] The training module is used to add the grayscale values ​​of the red, green and blue channels in the ROI as input parameters of blood oxygen saturation based on the ratio of ratio results of any two bands of red, green or blue as input parameters, and use multi-parameter machine learning training to obtain the estimated value of blood oxygen saturation.

[0034] As an optional solution, the evaluation module is specifically used for:

[0035] The PPG signal of each sub-ROI is extracted, filtered, and then fast Fourier transform is performed, and the quality of each sub-ROI is evaluated in the frequency domain. The quality evaluation formula is shown in the following formula:

[0036]

[0037] In the formula, Metric i is the evaluation coefficient of each sub-ROI, [PR-b, PR+b] is the small frequency range near the heartbeat frequency, and [B1, B2] is the bandpass filter range;

[0038] An evaluation coefficient of each sub-ROI is determined, and sub-ROIs with evaluation coefficients lower than a preset threshold are eliminated, and all remaining sub-ROIs are used to form an ROI subset.

[0039] As an optional solution, the evaluation module is specifically used for:

[0040] The weighted average of the ROI subsets is used to obtain the PPG signal for blood oxygen calculation. Let the ROI subset be [ROI1, ROI2, ..., ROI m ], the corresponding PPG signals and evaluation coefficients are [PPG1,PPG2,...,PPG m ],[Metric1,Metric2,...,Metricm ], where m is the number of sub-ROIs, and the weighted average process is as follows:

[0041] PPG calc =Metric1·PPG1+Metric2·PPG2+…+Metric m ·PPG m (2).

[0042] As an optional solution, the training module is specifically used for:

[0043] Calculate the ratio of ratio results and extract PPG respectively calc The time series of the red and green bands are used to calculate the AC signal and DC signal of the two bands respectively, and the ratio of ratio is shown in formula (3);

[0044]

[0045] In the formula, AC calc_red and DC calc_red is the red channel ROI calc The AC and DC components of the regional average over time, AC calc_green and DC calc_green It is the green channel ROI calc The AC and DC components of the regional averages varying with time;

[0046] Extract ROI calc Weighted average PPG in the region calc Gray value average of the three channels of Red, Green and Blue of the signal red Gray green and Gray blue , as the input parameter for calculating blood oxygen saturation;

[0047] The results of the 4 parameters ratio of ratio, Gray red Gray green and Gray blue As input parameters, a certain amount of samples are trained through the machine learning algorithm to obtain the final estimated value of blood oxygen saturation.

[0048] In a third aspect, an embodiment of the present invention provides a computer device, including:

[0049] at least one processor; and

[0050] a memory communicatively connected to the at least one processor; wherein,

[0051] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned non-contact human blood oxygen saturation detection method based on machine learning.

[0052] In a fourth aspect, an embodiment of the present invention provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable the computer to execute the above-mentioned non-contact human blood oxygen saturation detection method based on machine learning.

[0053] The non-contact human blood oxygen saturation detection method, device and computer equipment based on machine learning provided in the embodiments of the present invention obtain the image information of the target human body, determine the facial region of interest ROI area from the image information, divide the facial ROI area into multiple sub-ROIs, evaluate the quality of the facial sub-ROI area, remove the sub-ROI with low signal-to-noise ratio, and at the same time, perform weighted averaging on the retained ROI signal to improve the signal-to-noise ratio of the blood oxygen saturation solution signal. Secondly, in addition to the ratio of ratio parameter used in the traditional algorithm, the present invention also adds the gray value average of the three channels of Red, Green and Blue as input parameters for machine learning training. In addition to considering the reflection of light by the skin, this method also considers the influence of light scattering on blood oxygen saturation estimation, thereby improving the measurement accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 A flowchart of a non-contact human blood oxygen saturation detection method based on machine learning is provided in an embodiment of the present invention;

[0055] Figure 2 A flowchart diagram of another non-contact human blood oxygen saturation detection method based on machine learning is provided in an embodiment of the present invention;

[0056] Figure 3 A schematic diagram of ROI segmentation effect in a non-contact human blood oxygen saturation detection method based on machine learning is provided in an embodiment of the present invention;

[0057] Figure 4 A flowchart of a ROI quality evaluation process in a non-contact human blood oxygen saturation detection method based on machine learning is provided in an embodiment of the present invention;

[0058] Figure 5 A structural block diagram of a non-contact human blood oxygen saturation detection device based on machine learning is provided in an embodiment of the present invention;

[0059] Figure 6A structural block diagram of a computer device is provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0060] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0061] The terms "first", "second", "third", "fourth", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0062] Combination Figure 1 As shown, an embodiment of the present invention provides a non-contact human blood oxygen saturation detection method based on machine learning, comprising:

[0063] S101, acquiring image information of a target human body, for example, by reading in a video signal;

[0064] S102, determining a facial region of interest (ROI) from the image information, and selecting a suitable facial ROI region according to feature points identified by a face recognition algorithm;

[0065] S103, dividing the facial ROI area into a plurality of sub-ROIs;

[0066] S104, performing quality evaluation on the sub-ROIs, removing sub-ROIs with low signal-to-noise ratios, and performing weighted averaging on the retained ROI signals;

[0067] S105. On the basis of using the ratio of ratio results of any two bands of red, green or blue as input parameters, adding the grayscale values ​​of the red, green and blue channels in the ROI as input parameters of blood oxygen saturation, and using multi-parameter machine learning training to obtain an estimated value of blood oxygen saturation.

[0068] The non-contact human blood oxygen saturation detection method based on machine learning provided in the embodiment of the present invention obtains the image information of the target human body, determines the facial region of interest ROI area from the image information, divides the facial ROI area into multiple sub-ROIs, performs quality evaluation on the facial sub-ROI area, removes the sub-ROI with low signal-to-noise ratio, and at the same time, performs weighted averaging on the retained ROI signal to improve the signal-to-noise ratio of the blood oxygen saturation solution signal. Secondly, in addition to the ratio of ratio parameter used in the traditional algorithm, the present invention also adds the gray value average of the three channels of Red, Green and Blue as input parameters for machine learning training. In addition to considering the reflection of light by the skin, this method also considers the influence of light scattering on blood oxygen saturation estimation, thereby improving the measurement accuracy.

[0069] Combination Figure 2 and Figure 3 As shown, another non-contact human blood oxygen saturation detection method based on machine learning provided in an embodiment of the present invention includes:

[0070] Obtain image information of the target human body, determine a facial region of interest (ROI) from the image information, and divide the facial ROI into multiple sub-ROIs, such as Figure 2 shown.

[0071] The PPG signal of each sub-ROI is extracted, filtered and then subjected to fast Fourier transform, and the quality of each sub-ROI is evaluated in the frequency domain, such as Figure 3 The quality evaluation formula is as follows:

[0072]

[0073] In the formula, Metric i is the evaluation coefficient of each sub-ROI, [PR-b, PR+b] is the small frequency range near the heartbeat frequency, and [B1, B2] is the bandpass filter range;

[0074] An evaluation coefficient of each sub-ROI is determined, and sub-ROIs with evaluation coefficients lower than a preset threshold are eliminated, and all remaining sub-ROIs are used to form an ROI subset.

[0075] The weighted average of the ROI subsets is used to obtain the PPG signal for blood oxygen calculation. Let the ROI subset be [ROI1, ROI2, ..., ROI m ], the corresponding PPG signals and evaluation coefficients are [PPG1,PPG2,...,PPG m ],[Metric1,Metric2,...,Metric m], where m is the number of sub-ROIs, and the weighted average process is as follows:

[0076] PPG calc =Metric1·PPG1+Metric2·PPG2+…+Metric m ·PPG m (2).

[0077] Calculate the ratio of ratio results and extract PPG respectively calc The time series of the red and green bands are used to calculate the AC signal and DC signal of the two bands respectively, and the ratio of ratio is shown in formula (3);

[0078]

[0079] In the formula, AC calc_red and DC calc_red is the red channel ROI calc The AC and DC components of the regional average over time, AC calc_green and DC calc_green It is the green channel ROI calc The AC and DC components of the regional averages varying with time;

[0080] Extract ROI calc Weighted average PPG in the region calc Gray value average of the three channels of Red, Green and Blue of the signal red Gray green and Gray blue , as the input parameter for calculating blood oxygen saturation;

[0081] The results of the 4 parameters ratio of ratio, Gray red Gray green and Gray blue As input parameters, a certain amount of samples are trained through the machine learning algorithm to obtain the final estimated value of blood oxygen saturation.

[0082] Combination Figure 2 and 4 As shown, an embodiment of the present invention provides a non-contact human blood oxygen saturation detection method based on machine learning, and the method includes the following steps:

[0083] Step 1: Read the video signal and select the appropriate facial ROI area based on the feature points identified by the face recognition algorithm;

[0084] Step 2: Divide the facial ROI into n sub-ROIs;

[0085] Step 3: Extract the PPG signal of each sub-ROI, perform FFT transformation after filtering, and evaluate the quality of each sub-ROI in the frequency domain. The quality evaluation formula is shown as follows:

[0086]

[0087] In the formula, Metric i is the evaluation coefficient of each sub-ROI; [PR-b, PR+b] is the small frequency range near the heart rate, and [B1, B2] is the bandpass filter range.

[0088] After calculating the evaluation coefficient of each sub-ROI, the ROI with an evaluation coefficient lower than the threshold is eliminated, and all the remaining ROIs are used to establish ROI subsets;

[0089] Step 4: Perform weighted averaging on the ROI subsets to obtain the PPG signal for subsequent blood oxygen calculation. Assume that the ROI subset is [ROI1, ROI2, ..., ROI m ], the corresponding PPG signals and evaluation coefficients are [PPG1,PPG2,...,PPG m ],[Metric1,Metric2,...,Metric m ], where m is the number of sub-ROIs, and the weighted average process is as follows:

[0090] PPG calc =Metric1·PPG1+Metric2·PPG2+…+Metric m ·PPG m (2)

[0091] Step 5: Calculate the ratio of ratio signal. Extract PPG separately calc The time series of the red and green bands are calculated, and the AC signal (AC) and DC signal (DC) of the two bands are calculated respectively, and the ratio of ratio can be obtained as shown in formula (3);

[0092]

[0093] In the formula, AC calc_red and DC calc_red is the red channel ROI calc The AC and DC components of the regional average over time, AC calc_green and DC calc_green It is the green channel ROI calcThe AC and DC components of the regional averages varying with time.

[0094] Step 6: Extract ROI calc Weighted average PPG in the region calc Gray value average of the three channels of Red, Green and Blue of the signal red Gray green and Gray blue , as the input parameter for subsequent calculations.

[0095] Step 7: Use the four parameters ratio of ratio, Gray obtained in steps 5 and 6 red , Gray green and Gray blue As an input parameter, through a machine learning algorithm, the machine learning algorithm can use multiple linear regression to train a certain amount of samples to obtain an estimate of the final blood oxygen saturation.

[0096] The non-contact human blood oxygen saturation detection method based on machine learning provided in the embodiment of the present invention obtains the image information of the target human body, determines the facial region of interest ROI area from the image information, divides the facial ROI area into multiple sub-ROIs, performs quality evaluation on the facial sub-ROI area, removes the sub-ROI with low signal-to-noise ratio, and at the same time, performs weighted averaging on the retained ROI signal to improve the signal-to-noise ratio of the blood oxygen saturation solution signal. Secondly, in addition to the ratio of ratio parameter used in the traditional algorithm, the present invention also adds the gray value average of the three channels of Red, Green and Blue as input parameters for machine learning training. In addition to considering the reflection of light by the skin, this method also considers the influence of light scattering on blood oxygen saturation estimation, thereby improving the measurement accuracy.

[0097] Combination Figure 5 The embodiment of the present invention provides a non-contact human blood oxygen saturation detection device based on machine learning, comprising:

[0098] An acquisition module 501 is used to acquire image information of a target human body;

[0099] A determination module 502, configured to determine a facial region of interest (ROI) from the image information;

[0100] A division module 503 is used to divide the facial ROI area into a plurality of sub-ROIs;

[0101] An evaluation module 504 is used to evaluate the quality of the sub-ROIs, remove sub-ROIs with low signal-to-noise ratios, and perform weighted averaging on the retained ROI signals;

[0102] The training module 505 is used to add the grayscale values ​​of the red, green and blue channels in the ROI as input parameters of blood oxygen saturation based on the ratio of ratio results of any two bands of red, green or blue as input parameters, and use multi-parameter machine learning training to obtain an estimated value of blood oxygen saturation.

[0103] The evaluation module 404 is specifically used for:

[0104] The PPG signal of each sub-ROI is extracted, filtered, and then fast Fourier transform is performed, and the quality of each sub-ROI is evaluated in the frequency domain. The quality evaluation formula is shown in the following formula:

[0105]

[0106] In the formula, Metric i is the evaluation coefficient of each sub-ROI, [PR-b, PR+b] is the small frequency range near the heartbeat frequency, and [B1, B2] is the bandpass filter range;

[0107] An evaluation coefficient of each sub-ROI is determined, and sub-ROIs with evaluation coefficients lower than a preset threshold are eliminated, and all remaining sub-ROIs are used to form an ROI subset.

[0108] In some possible embodiments, the evaluation module 404 is specifically used to:

[0109] The weighted average of the ROI subsets is used to obtain the PPG signal for blood oxygen calculation. Let the ROI subset be [ROI1, ROI2, ..., ROI m ], the corresponding PPG signals and evaluation coefficients are [PPG1,PPG2,...,PPG m ],[Metric1,Metric2,...,Metric m ], where m is the number of sub-ROIs, and the weighted average process is as follows:

[0110] PPG calc =Metric1·PPG1+Metric2·PPG2+…+Metric m ·PPG m (2).

[0111] In some possible embodiments, the training module 405 is specifically used for:

[0112] Calculate the ratio of ratio results and extract PPG respectively calcThe time series of the red and green bands are used to calculate the AC signal and DC signal of the two bands respectively, and the ratio of ratio is shown in formula (3);

[0113]

[0114] In the formula, AC calc_red and DC calc_red is the red channel ROI calc The AC and DC components of the regional average over time, AC calc_green and DC calc_green It is the green channel ROI calc The AC and DC components of the regional averages varying with time;

[0115] Extract ROI calc Weighted average PPG in the region calc Gray value average of the three channels of Red, Green and Blue of the signal red Gray green and Gray blue , as the input parameter for calculating blood oxygen saturation;

[0116] The results of the 4 parameters ratio of ratio, Gray red Gray green and Gray blue As input parameters, a certain amount of samples are trained through the machine learning algorithm to obtain the final estimated value of blood oxygen saturation.

[0117] The non-contact human blood oxygen saturation detection device based on machine learning provided in the embodiment of the present invention obtains the image information of the target human body, determines the facial region of interest ROI area from the image information, divides the facial ROI area into multiple sub-ROIs, performs quality evaluation on the facial sub-ROI area, removes the sub-ROI with low signal-to-noise ratio, and at the same time, performs weighted averaging on the retained ROI signal to improve the signal-to-noise ratio of the blood oxygen saturation solution signal. Secondly, in addition to the ratio of ratio parameter used in the traditional algorithm, the present invention also adds the gray value average of the three channels of Red, Green and Blue as input parameters for machine learning training. In addition to considering the reflection of light by the skin, this method also considers the influence of light scattering on blood oxygen saturation estimation, thereby improving the measurement accuracy.

[0118] Accordingly, according to an embodiment of the present invention, the present invention also provides a computer device, a readable storage medium and a computer program product.

[0119] Figure 6FIG. 1 is a schematic diagram of the structure of a computer device 12 provided in an embodiment of the present invention. Figure 6 A block diagram of an exemplary computer device 12 suitable for use in implementing embodiments of the present invention is shown. Figure 6 The computer device 12 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0120] like Figure 6 As shown, computer device 12 is in the form of a general-purpose computing device. Computer device 12 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic devices can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0121] Components of computer device 12 may include, but are not limited to, one or more processors or processing units 16 , a system memory 28 , and a bus 18 that connects various system components including system memory 28 and processing unit 16 .

[0122] Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor or a local bus using any of a variety of bus architectures. By way of example, these architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.

[0123] The computer device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the computer device 12, including volatile and non-volatile media, removable and non-removable media.

[0124] The system memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. The computer device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 34 may be used to read and write non-removable, non-volatile magnetic media ( Figure 6 not shown, usually called a "hard drive"). Although Figure 6Not shown in the figure, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, a DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data medium interfaces. The memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the embodiments of the present invention.

[0125] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in the memory 28, such program modules 42 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules 42 generally perform the functions and / or methods of the embodiments described herein.

[0126] The computer device 12 may also communicate with one or more external devices 14 (e.g., keyboards, pointing devices, displays 24, etc.), may communicate with one or more devices that enable a user to interact with the computer device 12, and / or may communicate with any device that enables the computer device 12 to communicate with one or more other computing devices (e.g., network cards, modems, etc.). Such communication may be performed via an input / output (I / O) interface 22. Furthermore, the computer device 12 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 20. As shown, the network adapter 20 communicates with other modules of the computer device 12 via a bus 18. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the computer device 12, including, but not limited to, microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0127] The processing unit 16 executes various functional applications and data processing by running the programs stored in the system memory 28, such as implementing the non-contact human blood oxygen saturation detection method based on machine learning provided in an embodiment of the present invention.

[0128] An embodiment of the present invention also provides a non-transitory computer-readable storage medium storing computer instructions, on which a computer program is stored, wherein when the program is executed by a processor, the non-contact human blood oxygen saturation detection method based on machine learning provided in all the inventive embodiments of the present application is implemented.

[0129] The computer storage medium of the embodiment of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, an apparatus, or a device.

[0130] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0131] The program code included in the computer-readable medium can be transmitted with any appropriate medium, including but not limited to wireless, electric wire, optical cable, RF, etc., or any suitable combination of the above. The computer program code for performing the operation of the present invention can be written in one or more programming languages ​​or their combinations, and the programming language includes object-oriented programming languages ​​such as Java, Smalltalk, C++, and also includes conventional procedural programming languages-such as "C" language or similar programming languages. The program code can be executed completely on the user's computer, partially on the user's computer, executed as an independent software package, partially on the user's computer and partially on the remote computer, or completely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer by any type of network including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, using an Internet service provider to connect to the Internet).

[0132] An embodiment of the present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements the above-mentioned non-contact human blood oxygen saturation detection method based on machine learning.

[0133] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in the present disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution disclosed in the present invention can be achieved, and this document does not limit this.

[0134] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A non-contact human blood oxygen saturation detection method based on machine learning, characterized in that: include: Acquire image information of the target human body; Determine a facial region of interest (ROI) from the image information; Dividing the facial ROI region into a plurality of sub-ROIs; Performing quality evaluation on the sub-ROIs, removing sub-ROIs with low signal-to-noise ratios, and performing weighted averaging on the retained ROI signals; performing quality evaluation on the sub-ROIs, removing sub-ROIs with low signal-to-noise ratios, and performing weighted averaging on the retained ROI signals, including: The PPG signal of each sub-ROI is extracted, filtered and then fast Fourier transform is performed, and the quality of each sub-ROI is evaluated in the frequency domain. The quality evaluation formula is shown in the following formula: In the formula, Metric i is the evaluation coefficient of each sub-ROI, [PR-b, PR+b] is the small frequency range near the heartbeat frequency, and [B1, B2] is the bandpass filter range; Determine the evaluation coefficient of each sub-ROI, remove the sub-ROI with an evaluation coefficient lower than a preset threshold, and establish an ROI subset for all the remaining sub-ROIs; Based on the ratio of ratio results of any two bands of red, green or blue as input parameters, the grayscale values ​​of the red, green and blue channels in the ROI are added as input parameters of blood oxygen saturation, and multi-parameter machine learning training is used to obtain the estimated value of blood oxygen saturation.

2. The non-contact human blood oxygen saturation detection method based on machine learning according to claim 1 is characterized in that: The weighted averaging of the retained ROI signals comprises: The weighted average of the ROI subsets is used to obtain the PPG signal for blood oxygen calculation. Let the ROI subset be [ROI1, ROI2, ..., ROI m ], the corresponding PPG signals and evaluation coefficients are [PPG1,PPG2,...,PPG m ],[Metric1,Metric2,...,Metric m ], where m is the number of sub-ROIs, and the weighted average process is as follows: PPG calc =Metric1·PPG1+Metric2·PPG2+…+Metric m ·PPG m (2)。 3. The non-contact human blood oxygen saturation detection method based on machine learning according to claim 1 is characterized in that: On the basis of using the ratio of ratio results of any two bands of red, green or blue as input parameters, adding the grayscale values ​​of the red, green and blue channels in the ROI as input parameters of blood oxygen saturation, and using multi-parameter machine learning training to obtain an estimated value of blood oxygen saturation, including: Calculate the ratio of ratio results and extract PPG respectively calc The time series of the red and green bands are used to calculate the AC signal and DC signal of the two bands respectively, and the ratio of ratio is shown in formula (3); In the formula, AC calc_red and DC calc_red is the red channel ROI calc The AC and DC components of the regional average over time, AC calc_green and DC calc_green It is the green channel ROI calc The AC and DC components of the regional averages varying with time; Extract ROI calc Weighted average PPG in the region calc Gray value average of the three channels of Red, Green and Blue of the signal red Gray green and Gray blue , as the input parameter for calculating blood oxygen saturation; The results of the 4 parameters ratio of ratio, Gray red Gray green and Gray blue As input parameters, a certain amount of samples are trained through the machine learning algorithm to obtain the final estimated value of blood oxygen saturation.

4. A non-contact human blood oxygen saturation detection device based on machine learning, characterized in that: include: An acquisition module is used to acquire image information of a target human body; A determination module, used to determine a facial region of interest (ROI) from the image information; A division module, used for dividing the facial ROI area into a plurality of sub-ROIs; An evaluation module is used to evaluate the quality of the sub-ROIs, remove sub-ROIs with low signal-to-noise ratios, and perform weighted averaging on the retained ROI signals; the evaluation module is specifically used to: The PPG signal of each sub-ROI is extracted, filtered and then fast Fourier transform is performed, and the quality of each sub-ROI is evaluated in the frequency domain. The quality evaluation formula is shown in the following formula: In the formula, Metric i is the evaluation coefficient of each sub-ROI, [PR-b, PR+b] is the small frequency range near the heartbeat frequency, and [B1, B2] is the bandpass filter range; Determine the evaluation coefficient of each sub-ROI, remove the sub-ROI with an evaluation coefficient lower than a preset threshold, and establish an ROI subset for all the remaining sub-ROIs; The training module is used to add the grayscale values ​​of the red, green and blue channels in the ROI as input parameters of blood oxygen saturation based on the ratio of ratio results of any two bands of red, green or blue as input parameters, and use multi-parameter machine learning training to obtain the estimated value of blood oxygen saturation.

5. The non-contact human blood oxygen saturation detection device based on machine learning according to claim 4 is characterized in that: The evaluation module is specifically used for: The weighted average of the ROI subsets is used to obtain the PPG signal for blood oxygen calculation. Let the ROI subset be [ROI1, ROI2, ..., ROI m ], the corresponding PPG signals and evaluation coefficients are [PPG1,PPG2,...,PPG m ],[Metric1,Metric2,...,Metric m ], where m is the number of sub-ROIs, and the weighted average process is as follows: ·PPG calc =Metric1·PPG1+Metric2·PPG2+…+Metric m ·PPG m (2)。 6. The non-contact human blood oxygen saturation detection device based on machine learning according to claim 4 is characterized in that: The training module is specifically used for: Calculate the ratio of ratio results and extract PPG respectively calc The time series of the red and green bands are used to calculate the AC signal and DC signal of the two bands respectively, and the ratio of ratio is shown in formula (3); In the formula, AC calc_red and DC calc_red is the red channel ROI calc The AC and DC components of the regional average over time, AC calc_green and DC calc_green It is the green channel ROI calc The AC and DC components of the regional averages varying with time; Extract ROI calc Weighted average PPG in the region calc Gray value average of the three channels of Red, Green and Blue of the signal red Gray green and Gray blue , as the input parameter for calculating blood oxygen saturation; The results of the 4 parameters ratio of ratio, Gray red Gray green and Gray blue As input parameters, a certain amount of samples are trained through the machine learning algorithm to obtain the final estimated value of blood oxygen saturation.

7. A computer device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the non-contact human blood oxygen saturation detection method based on machine learning as described in any one of claims 1 to 3.

8. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to enable the computer to execute the non-contact human blood oxygen saturation detection method based on machine learning as described in any one of claims 1 to 3.

Images