Heart rate detection method and device, storage medium and electronic equipment

Abstract

The invention discloses a heart rate detection method and device, a storage medium and electronic equipment. Acquiring a pulse wave signal and an acceleration signal of the user, and judging the current state of the user; in the static state, time domain analysis is carried out based on the pulse wave signals to obtain at least one first static heart rate value, and frequency domain analysis is carried out based on the pulse wave signals to obtain a plurality of second static heart rate values; in the exercise state, frequency domain analysis is conducted based on the pulse wave signals, multiple first exercise heart rate values are obtained, deep learning analysis is conducted based on the pulse wave signals and the acceleration signals, and second exercise heart rate values are obtained; and determining a target heart rate value of the user according to the first rest heart rate value and the second rest heart rate value, or determining the target heart rate value of the user according to the first exercise heart rate value and the second exercise heart rate value. The time domain and the frequency domain are combined in a static state, so that the energy consumption is reduced while the accurate heart rate value is obtained; and a more accurate heart rate value is obtained by combining a frequency domain and deep learning in a motion state.

Description

Technical Field

[0001] The present application relates to the technical field of wearable devices, and in particular to a heart rate detection method, device, storage medium and electronic device. Background Art

[0002] With the continuous development of optical sensors and advanced signal processing algorithms, smart wearable devices such as sports watches can monitor the user's heart rate through photoplethysmography (PPG). It is based on optical sensors, which measures the attenuation of light reflected by blood vessels and other tissues on the surface of human skin, records the pulsation state of blood vessels, measures pulse waves, and thus associates the heart beat and measures the heart rate. However, when the wearer is exercising vigorously or the environment is disturbed, the pulse wave signal may be affected by noise, resulting in inaccurate heart rate measurement. Summary of the invention

[0003] The present application provides a heart rate detection method, device, storage medium and electronic device to solve the technical problem of low accuracy of a single pulse wave signal in heart rate detection.

[0004] In a first aspect, an embodiment of the present application provides a heart rate detection method, the method comprising:

[0005] Obtain the user's pulse wave signal and acceleration signal to determine the user's current state, which includes a stationary state or a moving state;

[0006] When the user is in a stationary state, performing time domain analysis based on the pulse wave signal to obtain at least one first stationary heart rate value, and performing frequency domain analysis based on the pulse wave signal to obtain multiple second stationary heart rate values;

[0007] When the user is in motion, frequency domain analysis is performed based on the pulse wave signal to obtain a plurality of first exercise heart rate values, and deep learning analysis is performed based on the pulse wave signal and the acceleration signal to obtain a second exercise heart rate value;

[0008] The user's target heart rate value is determined based on each first resting heart rate value and each second resting heart rate value, or the user's target heart rate value is determined based on each first exercise heart rate value and each second exercise heart rate value.

[0009] In a second aspect, an embodiment of the present application provides a heart rate detection device, the device comprising:

[0010] A state determination unit, used to obtain the user's pulse wave signal and acceleration signal, and determine the user's current state, which includes a static state or a moving state;

[0011] a first calculation unit, configured to, when the user is in a stationary state, perform a time domain analysis based on the pulse wave signal to obtain at least one first stationary heart rate value, and perform a frequency domain analysis based on the pulse wave signal to obtain a plurality of second stationary heart rate values;

[0012] a second calculation unit, configured to, when the user is in motion, perform frequency domain analysis based on the pulse wave signal to obtain a plurality of first exercise heart rate values, and perform deep learning analysis based on the pulse wave signal and the acceleration signal to obtain a second exercise heart rate value;

[0013] The result determination unit is used to determine the user's target heart rate value according to each first resting heart rate value and each second resting heart rate value, or to determine the user's target heart rate value according to each first exercise heart rate value and each second exercise heart rate value.

[0014] In a third aspect, an embodiment of the present application provides a computer storage medium, wherein the computer storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the steps of the method.

[0015] In a fourth aspect, an embodiment of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is suitable for being loaded by the processor and executing the steps of the method.

[0016] The beneficial effects brought about by the technical solutions provided by some embodiments of the present application include at least:

[0017] The present application provides a heart rate detection method, which obtains a user's pulse wave signal and an acceleration signal, and determines the user's current state, which includes a static state or a motion state; when the user is in a static state, a time domain analysis is performed based on the pulse wave signal to obtain at least one first static heart rate value, and a frequency domain analysis is performed based on the pulse wave signal to obtain multiple second static heart rate values; when the user is in a motion state, a frequency domain analysis is performed based on the pulse wave signal to obtain multiple first motion heart rate values, and a deep learning analysis is performed based on the pulse wave signal and the acceleration signal to obtain a second motion heart rate value; a target heart rate value of the user is determined based on each first static heart rate value and each second static heart rate value, or a target heart rate value of the user is determined based on each first motion heart rate value and the second motion heart rate value. By collecting pulse wave signals and acceleration signals, a data basis is provided for subsequent state judgment and heart rate analysis. Acceleration signals can effectively assist in distinguishing pulse wave changes caused by exercise from real heart rate changes, which helps to more comprehensively understand the user's heart rate changes. At the same time, accurately judging the user's current state is the prerequisite for the subsequent selection of an appropriate heart rate detection method, which can ensure that time domain analysis and frequency domain analysis that reduce power consumption are used in a static state, while frequency domain analysis and deep learning analysis are more effectively used to reduce the impact of motion artifacts in a dynamic state. When the user is in a static state, time domain analysis based on the pulse wave signal can directly reflect the changing trend of the pulse wave to obtain basic information about the heart rate, and also The pulse wave signal is converted from the time domain to the frequency domain to observe the distribution of the pulse wave at different frequencies, so as to obtain multiple heart rate values ​​in a static state; when the user is in motion, in addition to frequency domain analysis, a deep learning model is introduced to conduct a comprehensive analysis of the pulse wave signal and the acceleration signal. The deep learning model can learn and identify the pulse wave change pattern caused by motion, thereby effectively reducing the impact of motion artifacts on heart rate detection and more accurately extracting the true heart rate information in motion; finally, the user's target heart rate value is determined according to the heart rate value in a static state or in motion, which can help users understand their heart rate status and serve as an important basis for health management and disease prevention. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 An exemplary system architecture diagram of a heart rate detection method provided in an embodiment of the present application;

[0020] Figure 2A flowchart of a heart rate detection method provided in an embodiment of the present application;

[0021] Figure 3 A schematic diagram of the overall process of implementing a heart rate detection method provided in an embodiment of the present application;

[0022] Figure 4 A flowchart of a heart rate detection method provided in an embodiment of the present application;

[0023] Figure 5 A schematic diagram of a process for implementing time domain analysis in a heart rate detection method provided in an embodiment of the present application;

[0024] Figure 6 A flowchart of a heart rate detection method provided in an embodiment of the present application;

[0025] Figure 7 A flowchart of a heart rate detection method provided in an embodiment of the present application;

[0026] Figure 8 A schematic diagram of a flow chart of implementing frequency domain analysis in a heart rate detection method provided in an embodiment of the present application;

[0027] Fig. 9 A schematic diagram of a process for implementing deep learning analysis in a heart rate detection method provided in an embodiment of the present application;

[0028] Fig.10 A flowchart of a heart rate detection method provided in an embodiment of the present application;

[0029] Fig.11 A structural block diagram of a heart rate detection device provided in an embodiment of the present application;

[0030] Fig.12 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to make the features and advantages of the present application more obvious and easy to understand, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0032] When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Instead, they are only examples of devices and methods consistent with some aspects of the present application as detailed in the attached claims.

[0033] Smart wearable devices such as sports watches can monitor the user's heart rate through pulse wave signals (PPG signals). This is based on optical sensors, which measure the attenuation of light reflected by blood vessels and other tissues on the surface of human skin, record the pulsation state of blood vessels, measure pulse waves, and thus associate the heart beat and measure the heart rate. This realizes non-invasive and continuous heart rate monitoring, greatly improving the convenience of heart rate monitoring.

[0034] Although the pulse wave signal can provide accurate heart rate monitoring in a static state, when the wearer is exercising vigorously, the pulse wave signal may be interfered by noise due to factors such as muscle contraction and skin vibration. In addition, environmental factors such as light changes and temperature changes may also interfere with the pulse wave signal. These factors will lead to inaccurate heart rate measurement, or even the inability to obtain valid heart rate data, thus affecting the accuracy of health assessment.

[0035] Therefore, an embodiment of the present application provides a heart rate detection method to solve the technical problem of low accuracy of a single pulse wave signal in heart rate detection.

[0036] See also Figure 1 , Figure 1 An exemplary system architecture diagram of a heart rate detection method provided in an embodiment of the present application.

[0037] like Figure 1 As shown, the system architecture may include an electronic device 101, a network 102, and a server 103. The network 102 is used to provide a medium for a communication link between the electronic device 101 and the server 103. The network 102 may include various types of wireless communication links, such as a Bluetooth communication link, a Wireless-Fidelity (Wi-Fi) communication link, or a microwave communication link.

[0038] The electronic device 101 can interact with the server 103 through the network 102 to receive messages from the server 103 or send messages to the server 103, or the electronic device 101 can interact with the server 103 through the network 102 to receive messages or data sent by other users to the server 103. For example, the electronic device 101 can upload pulse wave signals and acceleration signals to the server 103 through the network 102 (such as Bluetooth, Wi-Fi), and then the server 103 processes these data using a deep learning algorithm and returns multiple exercise heart rate values ​​to the watch to reflect the user's heart health during exercise.

[0039] The electronic device 101 may be hardware or software. When the electronic device 101 is hardware, it may be various electronic devices, including but not limited to smart watches, smart rings, etc. When the electronic device 101 is software, it may be installed in the electronic devices listed above, and it may be implemented as multiple software or software modules (for example: used to provide distributed services), or it may be implemented as a single software or software module, which is not specifically limited here.

[0040] In an embodiment of the present application, the electronic device 101 first obtains the user's pulse wave signal and acceleration signal to determine the user's current state, which includes a stationary state or a motion state; then, when the user is in a stationary state, the electronic device 101 performs a time domain analysis based on the pulse wave signal to obtain at least one first stationary heart rate value, and performs a frequency domain analysis based on the pulse wave signal to obtain multiple second stationary heart rate values; when the user is in motion, the electronic device 101 performs a frequency domain analysis based on the pulse wave signal to obtain multiple first motion heart rate values, and performs a deep learning analysis based on the pulse wave signal and the acceleration signal to obtain a second motion heart rate value; finally, the electronic device 101 determines the user's target heart rate value based on each first stationary heart rate value and each second stationary heart rate value, or determines the user's target heart rate value based on each first motion heart rate value and each second motion heart rate value.

[0041] The server 103 may be a business server that provides various services. It should be noted that the server 103 may be hardware or software. When the server 103 is hardware, it may be implemented as a distributed server cluster consisting of multiple servers, or it may be implemented as a single server. When the server 103 is software, it may be implemented as multiple software or software modules (for example, for providing distributed services), or it may be implemented as a single software or software module, which is not specifically limited here.

[0042] Alternatively, the system architecture may not include the server 103. In other words, the server 103 may be an optional device in the embodiments of this specification. That is, the method provided in the embodiments of this specification may be applied to a system structure that only includes the electronic device 101, and the embodiments of this application do not limit this.

[0043] It should be understood that Figure 1 The number of electronic devices, networks and servers in the figure is only illustrative, and any number of electronic devices, networks and servers may be used according to implementation requirements.

[0044] See also Figure 2 , Figure 2 A flowchart of a heart rate detection method provided for an embodiment of the present application. The execution subject of the embodiment of the present application can be an electronic device that performs heart rate detection, or a processor in an electronic device that performs the heart rate detection method, or a heart rate detection service in an electronic device that performs the heart rate detection method. For the convenience of description, the specific execution process of the heart rate detection method is introduced below by taking the execution subject as an example of a processor in an electronic device.

[0045] like Figure 2 As shown, the heart rate detection method may at least include:

[0046] S202: Obtain the user's pulse wave signal and acceleration signal to determine the user's current state, where the current state includes a stationary state or a moving state.

[0047] Optionally, the pulse wave signal is the direct data source for heart rate detection, which provides basic information about heartbeats. When the user is in a stationary state, the pulse wave signal is easily affected by environmental noise such as light and temperature, and the influence of motion interference factors on the pulse wave signal is small. At this time, the pulse wave signal is relatively stable. When the user is in motion, due to the increase in physical activity, the pulse wave signal may be affected by motion artifacts, resulting in a decrease in the accuracy of the heart rate detection results. Based on this, it is possible to consider using different heart rate detection methods according to the different states of the user. When the user is in a stationary state, the influence of environmental interference factors on the pulse wave signal is considered, and the influence of motion interference factors does not need to be considered, thereby reducing the utilization of computing resources and reducing power consumption; when the user is in motion, it is necessary to filter out motion artifacts to improve the accuracy of heart rate detection.

[0048] Optionally, when the user is in a stationary state, the activities of various parts of the body decrease, and the corresponding acceleration changes are relatively small and present a relatively stable feature; when the user starts to move, the activities of various parts of the body increase, resulting in obvious acceleration changes, such as a sudden increase in acceleration, frequent changes in direction, etc. Based on this, it is possible to determine whether the user is in a stationary or moving state by processing and analyzing the user's acceleration signal.

[0049] Optionally, when performing heart rate detection through a wearable device, the user's pulse wave signal and acceleration signal can be obtained through sensors built into the wearable device (such as a sports watch, a sports ring, etc.), such as a photoelectric volume pulse wave sensor and an acceleration sensor, respectively, wherein the pulse wave signal is the direct data source for heart rate detection; the acceleration signal is used as auxiliary information to help us determine the user's state and optimize the heart rate detection method. Specifically, the acceleration signal can be processed to extract activity parameters that can quantify the user's activity, such as acceleration mean and variance, step count, energy consumption estimation, etc. Then, the user's current state is comprehensively judged based on these activity parameters. For example, when the acceleration mean is close to zero, the variance is very small, and the step count is zero or very low, it can be judged that the user is in a stationary state; when the acceleration signal shows a high mean, a large variance, and frequent peaks, and the step count is very high or cannot be accurately counted (such as running, jumping, etc.), it can be judged that the user is in motion.

[0050] It should be noted that in addition to the acceleration signal, the user's current state can also be judged by other means, such as judging whether the user is in motion by physiological parameters such as breathing rate and skin conductivity, and a more detailed state judgment can be made by combining information such as the user's activity history. The embodiment of the present application does not specifically limit the method for judging the user's state.

[0051] Optionally, Figure 3 A schematic diagram of the overall process of implementing a heart rate detection method provided in an embodiment of the present application is shown in FIG. Figure 3As shown in S302-306, in order to remove noise interference, such as environmental electromagnetic interference, baseline drift caused by breathing, etc., and retain the frequency components related to the heart rate, the acquired pulse wave signal and acceleration signal can also be subjected to time domain bandpass filtering of the heart rate band respectively to obtain the pulse wave bandpass signal and acceleration bandpass signal after filtering, and the user state judgment and subsequent heart rate detection steps can be performed based on the pulse wave bandpass signal and acceleration bandpass signal. Specifically, appropriate filter parameters can be selected according to the preset heart rate band (usually the adult heart rate range is 35-240 times / minute, and the corresponding frequency band is about 0.58-4Hz). At least one method of finite impulse response filter (FIR), infinite impulse response filter (IIR), and discrete wavelet transform (DWT) can be used here to remove noise (e.g., noise with a heart rate band of 0.1 Hz) and interference components in the signal, and retain the effective signal related to the heart rate.

[0052] S204. When the user is in a stationary state, perform time domain analysis based on the pulse wave signal to obtain at least one first stationary heart rate value, and perform frequency domain analysis based on the pulse wave signal to obtain multiple second stationary heart rate values.

[0053] Optionally, since when the user is in a stationary state, the influence of motion interference factors on the pulse wave signal is small and the pulse wave signal is relatively stable, when selecting a heart rate detection method in a stationary state, it is possible to consider saving power consumption as much as possible while ensuring the accuracy of the result. Based on this, if Figure 3 As shown in S308, in a stationary state, time domain analysis and frequency domain analysis can be combined to fully utilize information of different dimensions of the signal, verify and supplement each other, and improve the accuracy of heart rate detection while maintaining a low power consumption level.

[0054] Optionally, time domain analysis regards the signal as a function of time, which is intuitive and can directly show the changes of the signal over time, such as the waveform, amplitude, phase, etc. of the signal. At the same time, it is sensitive to the instantaneous characteristics of the signal and can capture the slight changes of the signal. Moreover, since time domain analysis mainly focuses on the changes of the signal over time, its calculation and processing process is relatively simple, and does not require complex calculations or a large amount of data conversion, so it can reduce the power consumption of the processor. Frequency domain analysis converts the signal from the time domain to the frequency domain through methods such as Fourier transform, analyzes the spectral characteristics of the signal, and further extracts the heart rate information. This process regards the signal as a function of frequency, and analyzes the characteristics of the signal by decomposing and reconstructing the signal on the frequency axis. Frequency domain analysis can accurately identify the components and distribution of the signal at different frequencies, and provide deeper heart rate information. Therefore, in a static state, the pulse wave signal is relatively stable, and the time domain analysis can intuitively display the waveform and timing relationship of the signal, while the frequency domain analysis can reveal the frequency components of the signal. The combination of the two can more comprehensively understand the characteristics of the heart rate.

[0055] Specifically, in the time domain analysis, the search algorithm can be used to identify the peaks or troughs of the pulse wave signal (or the pulse wave bandpass signal), calculate the pulse cycle between adjacent peaks (or troughs), and then calculate multiple first resting heart rate values ​​based on the pulse cycle. In addition, since the heart rate is relatively stable in the resting state, the heart rate values ​​of multiple consecutive cycles can also be calculated, and the average value can be taken as one of the resting heart rate values ​​of this stage to reduce accidental errors. In the frequency domain analysis, the pulse wave signal (or the pulse wave bandpass signal) can be subjected to a fast Fourier transform (Fast Fourier Transform, FFT) to obtain a spectrum diagram of the signal. Then, in the spectrum diagram, multiple spectral components corresponding to the heart beat frequency are found, which are usually manifested as obvious peaks, and the corresponding frequencies are the heart rate frequencies, and these heart rate frequencies are converted into second resting heart rate values. The first resting heart rate value obtained by the time domain analysis and the second resting heart rate value obtained by the frequency domain analysis are the resting heart rate values ​​in the resting state.

[0056] S206. When the user is in motion, frequency domain analysis is performed based on the pulse wave signal to obtain multiple first exercise heart rate values, and deep learning analysis is performed based on the pulse wave signal and the acceleration signal to obtain a second exercise heart rate value.

[0057] Optionally, in the state of motion, since there are many motion interference signals generated by physical activities, it is necessary to consider filtering out motion artifacts when selecting a heart rate detection method in the state of motion to improve the accuracy of heart rate detection. Based on this, Figure 3As shown in S310, in a stationary state, the pulse wave signal and the acceleration signal can be processed in combination with frequency domain analysis and deep learning analysis to cope with complex motion interference and heart rate changes and provide a more accurate heart rate value.

[0058] Optionally, the heart rate changes greatly during exercise, and frequency domain analysis can identify and process such changes. It has good anti-interference ability to pulse interference in time domain signals and can extract heart rate information more accurately. Deep learning algorithms are good at processing complex nonlinear signals and sequence data. They can automatically learn and extract key features in signals and capture the dynamic changes of pulse wave signals during exercise, thereby improving the accuracy of heart rate detection. In addition, during exercise, the acceleration signal can reflect the user's exercise state and exercise intensity, provide additional information for heart rate detection, and further improve the accuracy of heart rate detection.

[0059] Specifically, in the deep learning analysis under motion state, a large amount of labeled data (such as electrocardiogram, pulse wave signal, acceleration signal and real heart rate value recorded at the same time, etc.) can be used to train the model in the training stage, and by optimizing the loss function, the model can learn the mapping relationship from input signal to heart rate value. In the actual heart rate detection process, the pre-processed pulse wave signal and acceleration signal can be selected to be input into the trained deep learning model, and the second exercise heart rate value is obtained by multiple predictions or averages of the input signal by the model. Then the first exercise heart rate value obtained by frequency domain analysis and the second exercise heart rate value obtained by deep learning analysis are used as the exercise heart rate value under motion state.

[0060] It should be noted that the deep learning model can be a convolutional neural network, a recurrent neural network, or other neural network regression models, and the selection is made according to actual needs. For example, on wearable devices such as sports watches or sports rings that need to consider power consumption, a relatively power-saving neural network regression model can be selected.

[0061] S208: Determine a target heart rate value of the user according to each first resting heart rate value and each second resting heart rate value, or determine a target heart rate value of the user according to each first exercise heart rate value and each second exercise heart rate value.

[0062] Alternatively, if Figure 3 As shown in S312, after a plurality of resting heart rate values ​​or exercise heart rate values ​​are preliminarily determined, it is necessary to select the most accurate and representative heart rate value from the plurality of heart rate values ​​to output as the user's target heart rate value.

[0063] Optionally, first remove the obviously abnormal or unreasonable data points caused by signal interference, measurement error, etc. Next, the target heart rate value can be determined by a weighted average method. Specifically, a weight can be assigned to each heart rate value, and the determination of the weight can be based on a variety of factors, such as signal quality, the effectiveness of the analysis method, consistency with other heart rate values, the accuracy of historical data, etc. For example, when the acceleration signal shows that the user is in motion, the heart rate value obtained by deep learning analysis can be relied on more. At this time, a higher weight is assigned to the exercise heart rate value obtained by deep learning analysis, so that the target heart rate value is closer to the user's true heart rate. Then, the target heart rate value is output to the user or related system in an appropriate form so that the user can understand his or her health status. It should be noted that in addition to weighted average, other methods such as median and mode can also be used to determine the target heart rate value, and the specific acquisition method of the target heart rate value is not limited in the embodiment of the present application.

[0064] In an embodiment of the present application, a heart rate detection method is provided, which obtains a user's pulse wave signal and an acceleration signal to determine the user's current state, which includes a stationary state or a motion state; when the user is in a stationary state, a time domain analysis is performed based on the pulse wave signal to obtain at least one first stationary heart rate value, and a frequency domain analysis is performed based on the pulse wave signal to obtain multiple second stationary heart rate values; when the user is in a motion state, a frequency domain analysis is performed based on the pulse wave signal to obtain multiple first motion heart rate values, and a deep learning analysis is performed based on the pulse wave signal and the acceleration signal to obtain a second motion heart rate value; a target heart rate value of the user is determined based on each first stationary heart rate value and each second stationary heart rate value, or the target heart rate value of the user is determined based on each first motion heart rate value and each second motion heart rate value. By collecting pulse wave signals and acceleration signals, a data basis is provided for subsequent state judgment and heart rate analysis. Acceleration signals can effectively assist in distinguishing pulse wave changes caused by exercise from real heart rate changes, which helps to more comprehensively understand the user's heart rate changes. At the same time, accurately judging the user's current state is the prerequisite for the subsequent selection of an appropriate heart rate detection method, which can ensure that time domain analysis and frequency domain analysis that reduce power consumption are used in a static state, while frequency domain analysis and deep learning analysis are more effectively used to reduce the impact of motion artifacts in a dynamic state. When the user is in a static state, time domain analysis based on the pulse wave signal can directly reflect the changing trend of the pulse wave to obtain basic information about the heart rate, and also The pulse wave signal is converted from the time domain to the frequency domain to observe the distribution of the pulse wave at different frequencies, so as to obtain multiple heart rate values ​​in a static state; when the user is in motion, in addition to frequency domain analysis, a deep learning model is introduced to conduct a comprehensive analysis of the pulse wave signal and the acceleration signal. The deep learning model can learn and identify the pulse wave change pattern caused by motion, thereby effectively reducing the impact of motion artifacts on heart rate detection and more accurately extracting the true heart rate information in motion; finally, the user's target heart rate value is determined according to the heart rate value in a static state or in motion, which can help users understand their heart rate status and serve as an important basis for health management and disease prevention.

[0065] See also Figure 4 , Figure 4 A flowchart of a heart rate detection method provided in an embodiment of the present application.

[0066] like Figure 4 As shown, the heart rate detection method may at least include:

[0067] S402: Obtain the user's pulse wave signal and acceleration signal to determine the user's current state, where the current state includes a stationary state or a moving state.

[0068] Optionally, regarding step S402, please refer to the detailed description in step S202, which will not be repeated here.

[0069] S404, performing time-domain bandpass filtering on the pulse wave signal according to a preset heart rate frequency band to obtain a pulse wave bandpass signal; when the user is in a stationary state, performing peak and trough search on the pulse wave bandpass signal and / or establishing a probability density function to obtain at least one first stationary heart rate value.

[0070] Optionally, since the pulse wave signal may contain multiple frequency components, including components related to the heart rate and other interfering components (such as signals generated by breathing frequency, light effects, equipment noise, etc.), after obtaining the pulse wave signal, the pulse wave signal can be subjected to time domain bandpass filtering to obtain a pulse wave bandpass signal, so as to remove irrelevant noise components and obtain a purer pulse wave bandpass signal, and the subsequent heart rate value acquisition is performed based on this.

[0071] Optionally, in the time domain analysis, the pulse wave signal is represented by time as the horizontal axis and the pulse wave amplitude (heartbeat amplitude) as the vertical axis, which directly reflects the change of the pulse wave generated by each heart beat over time. In a static state, due to the periodic beating of the heart, the pulse wave signal will show a periodic fluctuation, in which the peak usually corresponds to the systole period and the trough corresponds to the diastole period. The time interval between the peaks or troughs reflects the cycle of the heart beat, that is, the heart rate. Based on this, Figure 5 A schematic diagram of a process for implementing time domain analysis in a heart rate detection method provided in an embodiment of the present application, such as Figure 5 As shown in S502-S504, in the time domain analysis method in the static state, the filtered pulse wave bandpass signal can be first obtained, and the pulse wave bandpass signal can be searched for peaks and troughs. This can be done by setting appropriate thresholds and search algorithms to identify the peaks and troughs in the signal, and according to the time intervals between peaks or troughs in different periods, multiple corresponding first static heart rate values ​​can be calculated. In addition, there may still be some small fluctuations or abnormal values ​​in the pulse wave bandpass signal after filtering. Considering the periodic characteristics of the signal, we can obtain a more stable first static heart rate value by calculating the average time interval between multiple peaks or troughs, further reducing the impact of random errors and noise interference. At the same time, other statistical methods (such as median, mode, etc.) can also be used to process multiple heart rate values ​​to obtain more reliable results.

[0072] Optionally, the probability density function is used to describe the probability distribution of continuous random variables. In the analysis of signals, since the peak or trough interval can take any real value (within a certain range), it can be regarded as a continuous random variable. Therefore, we can use the probability density function to describe the probability distribution of these intervals, thereby intuitively showing the distribution characteristics of the peak or trough intervals in the signal. Based on this, Figure 5 As shown in S506, the probability values ​​of different heart rates can be represented by establishing a probability density function for the pulse wave bandpass signal, wherein the peak position usually corresponds to the most likely first heart rate value. In addition, one or more probability thresholds are set according to actual needs, and the heart rate value that meets the probability threshold is used as the possible first resting heart rate value. In some cases, the probability density function may present multiple peaks, and there are also multiple heart rate values ​​that meet the probability threshold. These heart rate values ​​can be preliminarily determined as the first resting heart rate value, and then combined with other analysis methods (such as frequency domain analysis, etc.) to further screen and determine the target heart rate value.

[0073] S406, differentiating the pulse wave bandpass signal to obtain a pulse wave differential signal; performing a peak and trough search and / or establishing a probability density function on the pulse wave differential signal to obtain at least one first resting heart rate value.

[0074] Alternatively, if Figure 5 As shown in S502-508, in order to further improve the accuracy of obtaining the heart rate value and reduce the influence of low-frequency interference such as baseline drift and light changes, the pulse wave bandpass signal can be differentially processed. Specifically, the bandpass signal is differentiated in turn, and the difference between adjacent sampling points is used as the new signal value to obtain the pulse wave differential signal. In this way, the low-frequency interference will be weakened in the differential signal, and the characteristic points such as peaks and troughs that are directly related to the heart rate will be enhanced, making them easier to identify and extract in subsequent analysis. Next, the pulse wave differential signal is also searched for peaks and troughs and a probability density function is established to obtain a possible first resting heart rate value. Specifically, regarding how to search for peaks and troughs and establish a probability density function for the pulse wave differential signal, please refer to the detailed record in step S404, which will not be repeated here.

[0075] S408: Perform frequency domain analysis based on the pulse wave signal to obtain multiple second resting heart rate values.

[0076] Optionally, in addition to the time domain analysis, the pulse wave bandpass signal can also be subjected to frequency domain analysis in a stationary state. Specifically, regarding step S408, please refer to the detailed record in step S204, which will not be repeated here.

[0077] S410: When the user is in motion, frequency domain analysis is performed based on the pulse wave signal to obtain multiple first exercise heart rate values, and deep learning analysis is performed based on the pulse wave signal and the acceleration signal to obtain a second exercise heart rate value.

[0078] Optionally, regarding step S410, please refer to the detailed description in step S206, which will not be repeated here.

[0079] S412: Determine a target heart rate value of the user based on each first resting heart rate value and each second resting heart rate value, or determine a target heart rate value of the user based on each first exercise heart rate value and each second exercise heart rate value.

[0080] Optionally, regarding step S412, please refer to the detailed description in step S208, which will not be repeated here.

[0081] In an embodiment of the present application, a heart rate detection method is provided, which performs a peak and trough search on a pulse wave bandpass signal, and can calculate a first resting heart rate value through the peaks and troughs, which are key feature points that directly reflect the heart beat cycle; at the same time, by establishing a probability density function for the pulse wave bandpass signal, it is possible to describe the probability distribution of different heart rates, thereby intuitively obtaining a possible first resting heart rate value, reducing the utilization of computing resources and reducing power consumption; further, the pulse wave bandpass signal is differentially processed, and the peak and trough search and probability density function are performed on the differential pulse wave difference signal, which can enhance the changing components in the signal, thereby removing the low-frequency interference components in the pulse wave signal, which helps to more accurately identify and extract the feature points directly related to the heart rate, and further enhance the accuracy of heart rate detection.

[0082] See also Figure 6 , Figure 6 A flowchart of a heart rate detection method provided in an embodiment of the present application.

[0083] like Figure 6 As shown, the heart rate detection method may at least include:

[0084] S602. Obtain the pulse wave signal and acceleration signal of the user, perform time-domain bandpass filtering on the pulse wave signal according to a preset heart rate frequency band to obtain a pulse wave bandpass signal; perform time-domain bandpass filtering on the acceleration signal according to a preset heart rate frequency band to obtain an acceleration bandpass signal; determine the user's activity parameter based on the acceleration bandpass signal; when the activity parameter is less than a preset threshold, determine that the user is in a stationary state; when the activity parameter is greater than or equal to the preset threshold, determine that the user is in motion.

[0085] Optionally, firstly, the user's pulse wave signal and acceleration signal are obtained respectively through the built-in sensor of the wearable device, and then the pulse wave signal is subjected to time-domain bandpass filtering to obtain a pulse wave bandpass signal; similarly, multiple acceleration signals can also be subjected to time-domain bandpass filtering of the heart rate band to obtain multiple acceleration bandpass signals (for example, three acceleration bandpass signals of x, y, and z). Specifically, regarding how to perform time-domain bandpass filtering of the heart rate band on the signal, please refer to the detailed record in step S202, which will not be repeated here. Furthermore, the current state of the user is judged based on the filtered acceleration bandpass signal.

[0086] Specifically, the acceleration bandpass signal is processed to extract activity parameters that can quantify the user's activity, such as acceleration mean and variance, step count, energy consumption estimation, etc. The above activity parameters are comprehensively analyzed and combined with preset thresholds or models to determine the user's current state. For example, when the activity parameter is lower than a certain threshold, the user can be considered to be in a static state; when the activity parameter is higher than the threshold and meets the motion characteristics, the user can be considered to be in motion.

[0087] Optionally, a three-axis acceleration sensor is used to measure the acceleration signal of the user, and the acceleration signals in the three directions of x, y, and z are output respectively. A three-axis acceleration sensor is a sensor that can simultaneously measure the acceleration of the user in three orthogonal directions. It can provide the user with a complete acceleration vector in three-dimensional space, thereby more accurately reflecting the user's motion state in three-dimensional space. Among them, the static state is usually manifested as the value of the acceleration signal in each direction is close to zero, while the motion state is manifested as a significant change in the acceleration signal in at least one direction.

[0088] S604: When the user is in a stationary state, perform time domain analysis based on the pulse wave signal to obtain at least one first stationary heart rate value.

[0089] Optionally, regarding step S604, please refer to the detailed description in step S204, which will not be repeated here.

[0090] S606: Obtain a first frequency spectrum corresponding to the pulse wave bandpass signal, and search for heart rate values ​​on the first frequency spectrum to obtain a plurality of second resting heart rate values.

[0091] Optionally, when frequency domain analysis is required based on the pulse wave signal in a stationary state, the pulse wave bandpass signal is first obtained, and then the pulse wave bandpass signal is subjected to a fast Fourier transform to obtain its spectrum representation, i.e., the first spectrum. The first spectrum is the representation of the pulse wave bandpass signal in the frequency domain. In the spectrum diagram, the horizontal axis usually represents the frequency, and the vertical axis represents the energy of the signal, showing the energy distribution of the signal at different frequencies. Therefore, by observing the first spectrum, we can intuitively understand which frequency components in the signal dominate and which frequency components are weak or almost non-existent. Among them, the peak value usually corresponds to the user's heart rate value. At this time, the frequency corresponding to the peak value with higher energy in the first spectrum is converted into the second stationary heart rate value. These first stationary heart rate values ​​obtained by time domain analysis and the second stationary heart rate values ​​obtained by frequency domain analysis are the stationary heart rate values ​​of the user in a stationary state.

[0092] S608: When the user is in motion, a first frequency spectrum corresponding to the pulse wave bandpass signal is obtained, and a heart rate value search is performed on the first frequency spectrum to obtain a plurality of first motion heart rate values.

[0093] Optionally, when the user is in motion, multiple first exercise heart rate values ​​can also be determined according to the frequency domain analysis method. Specifically, regarding step S608, please refer to the detailed description in step S606, which will not be repeated here.

[0094] S610, performing deep learning analysis based on the pulse wave signal and the acceleration signal to obtain a second exercise heart rate value.

[0095] Optionally, regarding step S610, please refer to the detailed description in step S206, which will not be repeated here.

[0096] S612: Determine a target heart rate value of the user based on each first resting heart rate value and each second resting heart rate value, or determine a target heart rate value of the user based on each first exercise heart rate value and each second exercise heart rate value.

[0097] Optionally, regarding step S612, please refer to the detailed description in step S208, which will not be repeated here.

[0098] In an embodiment of the present application, a heart rate detection method is provided. By using a three-axis acceleration sensor to simultaneously measure acceleration in three directions, the error caused by measurement in a single direction can be reduced, thereby providing more accurate acceleration information and improving the accuracy of overall detection. By extracting multiple activity parameters in the acceleration bandpass signal and comprehensively considering their changes, the current state of the user can be more accurately judged, further improving the accuracy of heart rate detection. By performing a fast Fourier transform on the pulse wave bandpass signal to obtain a first spectrum, the energy distribution of the pulse wave bandpass signal at different frequencies can be intuitively displayed, and spectrum analysis can be performed based on this to obtain a second resting heart rate value or a first exercise heart rate value, which helps to reduce interference from noise and other non-heart rate related signals, thereby improving the accuracy of heart rate detection.

[0099] See also Figure 7 , Figure 7 A flowchart of a heart rate detection method provided in an embodiment of the present application.

[0100] like Figure 7 As shown, the heart rate detection method may at least include:

[0101] S702. Obtain the pulse wave signal and acceleration signal of the user, perform time-domain bandpass filtering on the pulse wave signal according to a preset heart rate frequency band to obtain a pulse wave bandpass signal; perform time-domain bandpass filtering on the acceleration signal according to a preset heart rate frequency band to obtain an acceleration bandpass signal; determine the user's activity parameter based on the acceleration bandpass signal; when the activity parameter is less than a preset threshold, determine that the user is in a stationary state; when the activity parameter is greater than or equal to the preset threshold, determine that the user is in motion.

[0102] Optionally, regarding step S702, please refer to the detailed description in step S602, which will not be repeated here.

[0103] S704: When the user is in a stationary state, perform time domain analysis based on the pulse wave signal to obtain at least one first stationary heart rate value.

[0104] Optionally, regarding step S704, please refer to the detailed description in step S204, which will not be repeated here.

[0105] S706, and obtain the pulse wave power spectrum density corresponding to the pulse wave bandpass signal, and obtain the acceleration power spectrum density corresponding to the acceleration bandpass signal; obtain the priori signal-to-noise ratio and the a posteriori signal-to-noise ratio of the pulse wave bandpass signal corresponding to the acceleration bandpass signal based on the pulse wave power spectrum density and the acceleration power spectrum density.

[0106] Optionally, the acceleration signal can reflect the user's exercise intensity and exercise pattern, and this information is crucial for accurately calculating the heart rate value. For example, in different types of exercise such as running, walking or cycling, the pulse wave signal may be interfered with to varying degrees, while the acceleration signal can provide useful information about the type and intensity of exercise, thereby helping to more accurately estimate the heart rate value. Therefore, the acquired acceleration signal can not only be used to determine the user's current state, but also used in the frequency domain analysis method (the frequency domain analysis method suitable for the exercise state), and combined with the pulse wave signal to improve the accuracy of heart rate detection. Based on this, Figure 8 A flowchart of implementing frequency domain analysis in a heart rate detection method provided in an embodiment of the present application is shown in FIG. Figure 8 As shown in S802, in the frequency domain analysis method in a static state, firstly, a pulse wave bandpass signal and multi-channel acceleration bandpass signals (x, y, z three-channel acceleration bandpass signals) are obtained.

[0107] Alternatively, if Figure 8 As shown in S804, in the frequency domain analysis method, in order to quantitatively analyze the energy distribution of the pulse wave bandpass signal and the acceleration bandpass signal in a specific frequency range, and to evaluate the noise level in the signal, the power spectral density (PSD) of the two signals can be established respectively. The power spectral density is a physical quantity that describes the change of signal power with frequency, which can reflect the energy distribution of the signal at different frequencies. Since the acceleration bandpass signal contains motion-related noise, by obtaining the multi-channel acceleration power spectral density of the multi-channel acceleration bandpass signal, we can evaluate in which frequency range these noises are most significant, so as to more accurately identify the frequency components related to the heart beat by comparing the pulse wave power spectral density and the acceleration power spectral density, and avoid misjudging the motion noise as a heart rate signal. At the same time, based on this information, a suitable filter can be designed to remove the noise component while retaining useful heart rate information, thereby improving the robustness and accuracy of heart rate detection.

[0108] Alternatively, if Figure 8 As shown in S806, the prior signal-to-noise ratio (Prior SNR) of the pulse wave bandpass signal corresponding to the multi-channel acceleration bandpass signal is first obtained from the power spectrum density. This signal-to-noise ratio reflects the strength of the pulse wave bandpass signal relative to the acceleration noise. Then, the signal is filtered by the Wiener filter in the frequency domain analysis, and the corresponding posterior signal-to-noise ratio (Posterior SNR) is obtained again, so that the first spectrum of the pulse wave bandpass signal can be filtered based on the prior signal-to-noise ratio and combined with the posterior signal-to-noise ratio.

[0109] It should be noted that when there are three acceleration bandpass signals x, y, and z, each frequency point of each pulse wave bandpass signal corresponds to the three acceleration bandpass signals x, y, and z, and each has its own signal-to-noise ratio, so as to optimize the spectrum through the cascade of multiple filters. For example, assuming that there are 100 frequency points (expressed as a frequency point matrix [1,100], which are 1.01 Hz, 1.02 Hz, 1.03 Hz, etc.) in the 1-2 Hz pulse wave bandpass signal, each frequency point corresponds to the signal-to-noise ratio on the three acceleration bandpass signals x, y, and z (expressed as the x-path signal-to-noise ratio matrix [1,100], the y-path signal-to-noise ratio matrix [1,100], and the z-path signal-to-noise ratio matrix [1,100]), and the pulse wave bandpass signal at each frequency point can be subsequently filtered in real time according to the signal-to-noise ratio at each frequency point. Therefore, the number of pulse wave bandpass signals (such as one, two, etc.) and the frequency resolution (such as 10 frequencies within 1 Hz with a resolution of 0.1 Hz; 100 frequencies with a resolution of 0.01 Hz) can be set according to actual needs. For example, on wearable devices such as sports watches, considering the issue of power saving, only one pulse wave bandpass signal can be obtained and a frequency with a lower resolution can be set.

[0110] S708, obtaining a first spectrum corresponding to the pulse wave bandpass signal; filtering the first spectrum according to the priori signal-to-noise ratio and the a posteriori signal-to-noise ratio to obtain a second spectrum corresponding to the pulse wave bandpass signal, searching the second spectrum for heart rate values ​​to obtain multiple second resting heart rate values.

[0111] Optionally, the first spectrum corresponding to the pulse wave bandpass signal can be filtered according to the power spectrum density and signal-to-noise ratio information. Specifically, the filtering process can perform weighted processing on different frequency components in the spectrum according to the signal-to-noise ratio. In frequency segments with higher signal-to-noise ratios (usually the range corresponding to the heart rate), more signal components are retained; in frequency segments with lower signal-to-noise ratios (usually the range corresponding to the noise), more noise components are suppressed. In this way, the spectrum can be further optimized to improve the accuracy of heart rate detection. After spectrum filtering based on the signal-to-noise ratio, the second spectrum of the pulse wave bandpass signal is obtained. This spectrum is purer than the first spectrum, and the noise components are effectively suppressed. Next, based on the frequency components and intensity distribution in the second spectrum, the heart rate detection algorithm is used to calculate the possible second resting heart rate value.

[0112] S710. When the user is in motion, obtain the pulse wave power spectrum density corresponding to the pulse wave bandpass signal, and obtain the acceleration power spectrum density corresponding to the acceleration bandpass signal; obtain the priori signal-to-noise ratio and the a posteriori signal-to-noise ratio of the pulse wave bandpass signal corresponding to the acceleration bandpass signal based on the pulse wave power spectrum density and the acceleration power spectrum density.

[0113] Optionally, regarding step S710, please refer to the detailed description in step S706, which will not be repeated here.

[0114] S712, obtaining a first spectrum corresponding to the pulse wave bandpass signal; filtering the first spectrum according to the priori signal-to-noise ratio and the a posteriori signal-to-noise ratio to obtain a second spectrum corresponding to the pulse wave bandpass signal, searching the second spectrum for heart rate values ​​to obtain a plurality of first exercise heart rate values.

[0115] Optionally, the first spectrum in the motion state is filtered according to the calculation of the signal-to-noise ratio to obtain a purer second spectrum, and then the first exercise heart rate value is obtained from the second spectrum. Specifically, for step S712, please refer to the detailed record in step S708, which will not be repeated here.

[0116] S714, and obtain the acceleration spectrum corresponding to the acceleration bandpass signal; input the second spectrum, the acceleration spectrum and the activity parameters into a preset deep learning model; obtain the exercise heart rate value output by the deep learning model to obtain a second exercise heart rate value.

[0117] Optionally, Fig. 9 A flowchart of implementing deep learning analysis in a heart rate detection method provided in an embodiment of the present application is shown in FIG. Fig. 9 As shown in S902, in the deep learning analysis method under the motion state, in order to convert the acceleration bandpass signal from the time domain to the frequency domain so as to perform combined analysis with the pulse wave bandpass signal in the frequency domain, the acceleration bandpass signal is firstly subjected to spectrum analysis to obtain the acceleration spectrum. Fig. 9 As shown in S904-S906, the second spectrum, acceleration spectrum and activity parameters are input as parameters into the trained deep learning model, and the second exercise heart rate value is obtained through the model. Specifically, for how to perform deep learning analysis, please refer to the detailed record in step S206, which will not be repeated here.

[0118] Optionally, the first exercise heart rate value obtained through frequency domain analysis and the second exercise heart rate value obtained through deep learning analysis are used as the exercise heart rate values ​​in the exercise state.

[0119] S716: Determine a target heart rate value of the user based on each first resting heart rate value and each second resting heart rate value, or determine a target heart rate value of the user based on each first exercise heart rate value and each second exercise heart rate value.

[0120] Optionally, regarding step S716, please refer to the detailed description in step S208, which will not be repeated here.

[0121] In an embodiment of the present application, a heart rate detection method is provided, which uses pulse wave power spectral density and acceleration power spectral density to calculate a priori signal-to-noise ratio and a posteriori signal-to-noise ratio, and filters a first spectrum of the pulse wave bandpass signal based on the priori signal-to-noise ratio and the posteriori signal-to-noise ratio to obtain a second spectrum. This process further optimizes the original spectrum, filters out frequency components with lower signal-to-noise ratio, and retains frequency components that are more relevant to heart rate information, thereby improving the accuracy of heart rate detection; through a deep learning analysis method, it is possible to fully utilize the feature extraction and pattern recognition capabilities of a deep learning model, extract features closely related to heart rate from complex signals based on pulse wave signals, acceleration signals, and activity parameters, and more accurately identify the heart rate of a user in different states, thereby improving the accuracy of heart rate detection.

[0122] See also Fig.10 , Fig.10 A flowchart of a heart rate detection method provided in an embodiment of the present application.

[0123] like Fig.10 As shown, the heart rate detection method may at least include:

[0124] S1002. Obtain the pulse wave signal and acceleration signal of the user, perform time-domain bandpass filtering on the pulse wave signal according to a preset heart rate frequency band to obtain a pulse wave bandpass signal; perform time-domain bandpass filtering on the acceleration signal according to a preset heart rate frequency band to obtain an acceleration bandpass signal; determine the user's activity parameter based on the acceleration bandpass signal; when the activity parameter is less than a preset threshold, determine that the user is in a stationary state; when the activity parameter is greater than or equal to the preset threshold, determine that the user is in motion.

[0125] Optionally, regarding step S1002, please refer to the detailed description in step S602, which will not be repeated here.

[0126] S1004: When the user is in a stationary state, perform time domain analysis based on the pulse wave signal to obtain at least one first stationary heart rate value.

[0127] Optionally, regarding step S1004, please refer to the detailed description in step S204, which will not be repeated here.

[0128] S1006, and obtain the pulse wave power spectrum density corresponding to the pulse wave bandpass signal, and obtain the acceleration power spectrum density corresponding to the acceleration bandpass signal; obtain the priori signal-to-noise ratio and the a posteriori signal-to-noise ratio of the pulse wave bandpass signal corresponding to the acceleration bandpass signal based on the pulse wave power spectrum density and the acceleration power spectrum density.

[0129] Optionally, regarding step S1006, please refer to the detailed description in step S706, which will not be repeated here.

[0130] S1008. Obtain a first spectrum corresponding to the pulse wave bandpass signal; determine a first frequency domain filter coefficient through a priori signal-to-noise ratio and a posteriori signal-to-noise ratio, and filter the first spectrum through the first frequency domain filter coefficient; or determine the first frequency domain filter coefficient through a priori signal-to-noise ratio and a posteriori signal-to-noise ratio, and determine a second frequency domain filter coefficient through the first frequency domain filter coefficient and an activity parameter, and filter the first spectrum according to the second frequency domain filter coefficient to obtain a second spectrum corresponding to the pulse wave bandpass signal.

[0131] Optionally, Figure 8 A flowchart of implementing frequency domain analysis in a heart rate detection method provided in an embodiment of the present application is shown in FIG. Figure 8 As shown in S808, in a feasible implementation, when optimizing the first spectrum, an algorithm such as Wiener filtering can be used to construct and dynamically adjust multiple first frequency domain filter coefficients according to the changes in the priori signal-to-noise ratio and the a posteriori signal-to-noise ratio, wherein each pulse wave bandpass signal corresponds to the filter coefficients on the three acceleration bandpass signals x, y, and z, for example, the x-channel first frequency domain filter coefficient matrix [1, n] (n is the number of frequency points representing the resolution). Then, the first spectrum is filtered using the calculated multiple first frequency domain filter coefficients (the multiple first frequency domain filter coefficient matrices [1, n] are multiplied point by point with the frequency point matrix [1, n]), and the optimized second spectrum is obtained, thereby effectively suppressing noise while retaining the heart rate-related signals.

[0132] Alternatively, if Figure 8 As shown in S810-814, in another feasible implementation, the user's activity parameters are further introduced on the basis of the multiple first frequency domain filter coefficients to adjust the second spectrum. Specifically, in combination with the user's activity parameters, the multiple first frequency domain filter coefficients are further optimized to obtain corresponding multiple second frequency domain filter coefficients. The optimization process here can be an adjustment based on preset rules, or it can be a model obtained through pre-training, which can automatically adjust the filter coefficients according to different activity parameters to adapt to the heart rate detection requirements under different motion states. Then, the multiple second frequency domain filter coefficients are applied to filter the first spectrum of the pulse wave bandpass signal to obtain an optimized second spectrum.

[0133] S1010, performing a peak search on the second frequency spectrum according to the activity parameter to determine a plurality of heart rate values ​​to be confirmed; and correcting each heart rate value to be confirmed by a resolution refinement method to obtain a plurality of second resting heart rate values.

[0134] Optionally, in the frequency domain analysis, the second spectrum is represented by frequency as the horizontal axis and energy as the vertical axis, showing the energy distribution of the signal at different frequencies, where the frequency points with high energy are most likely to correspond to the true heart rate, while noise is usually manifested as widely distributed low-energy components. Since we are concerned with the spectral components corresponding to the heart beat frequency, Figure 8 As shown in S816, these high-energy frequency points related to heart rate can be identified by peak search. At the same time, since the interference of various noises may not be completely filtered out during the aforementioned filtering process, these noises may cause multiple high-energy peaks to appear on the second spectrum graph, not just the peak corresponding to the heart rate. Therefore, simply finding the maximum peak may not accurately reflect the true heart rate value, and multiple possible peak positions (that is, the heart rate value to be confirmed) can be found to increase the probability of identifying the correct heart rate value. In addition, when the second spectrum is searched for peaks, the search range can be narrowed in combination with activity parameters to improve the efficiency of peak search. For example, when the known motion parameters are large, the heart rate usually increases accordingly. If the heart rate value corresponding to a peak position is significantly low, then the peak is likely to be caused by noise rather than the true heart rate value. At this time, a higher search threshold can be set according to the activity parameters to avoid unnecessary searches in the low-frequency area and further improve the accuracy of heart rate detection.

[0135] Optionally, in actual applications, due to limitations such as power consumption, the spectrum may not be subdivided into many frequency points, which means that within a limited frequency range, the true frequency corresponding to the heart rate value may not be accurately captured. For example, if there are only 10 frequency points in the range of 1-2 Hz and the resolution is 0.1 Hz, then within this range we can only identify specific heart rate values ​​such as 66, 72, 78 (times / minute). Obviously, this resolution is not enough for accurate heart rate measurement, because the true heart rate value may be between these discrete points. Based on this, if Figure 8 As shown in S818-S820, in order to overcome the limitation of the resolution of the second spectrum, we need to refine the resolution of the heart rate value to be confirmed, so as to determine the heart rate value more accurately. Specifically, this can be achieved through a variety of technical means, including at least one of Zoom-FFT transform, Chirp-Z transform and Yip-Zoom transform. After the resolution of the second spectrum is refined, the position of the heart rate value to be confirmed is corrected according to the refinement result to obtain a corrected second resting heart rate value, which is closer to the user's true heart rate. The first resting heart rate value obtained by time domain analysis and the second resting heart rate value obtained by frequency domain analysis are the resting heart rate values ​​of the user in a stationary state.

[0136] S1012. When the user is in motion, obtain the pulse wave power spectrum density corresponding to the pulse wave bandpass signal, and obtain the acceleration power spectrum density corresponding to the acceleration bandpass signal; obtain the priori signal-to-noise ratio and the a posteriori signal-to-noise ratio of the pulse wave bandpass signal corresponding to the acceleration bandpass signal based on the pulse wave power spectrum density and the acceleration power spectrum density.

[0137] Optionally, regarding step S1012, please refer to the detailed description in step S706, which will not be repeated here.

[0138] S1014, obtaining a first spectrum corresponding to the pulse wave bandpass signal; determining a first frequency domain filter coefficient by a priori signal-to-noise ratio and a posteriori signal-to-noise ratio, and filtering the first spectrum by the first frequency domain filter coefficient; or determining a first frequency domain filter coefficient by a priori signal-to-noise ratio and a posteriori signal-to-noise ratio, and determining a second frequency domain filter coefficient by the first frequency domain filter coefficient and activity parameters, filtering the first spectrum according to the second frequency domain filter coefficient, and obtaining a second spectrum corresponding to the pulse wave bandpass signal; performing a peak search on the second spectrum according to the activity parameters, and determining a plurality of heart rate values ​​to be confirmed; correcting each heart rate value to be confirmed by a resolution refinement method, and obtaining a plurality of first exercise heart rate values; and performing deep learning analysis based on the pulse wave signal and the acceleration signal, and obtaining a second exercise heart rate value.

[0139] Optionally, in the motion state, multiple first motion heart rate values ​​can also be acquired according to the frequency domain analysis. Specifically, regarding step S1014, please refer to the detailed records in steps S1008, S1010, and S206, which will not be repeated here.

[0140] S1016: Determine a target heart rate value of the user based on each first resting heart rate value and each second resting heart rate value, or determine a target heart rate value of the user based on each first exercise heart rate value and each second exercise heart rate value.

[0141] Optionally, regarding step S1016, please refer to the detailed description in step S208, which will not be repeated here.

[0142] In an embodiment of the present application, a heart rate detection method is provided. By determining a first frequency domain filter coefficient based on a priori signal-to-noise ratio and a posteriori signal-to-noise ratio, interference signals generated during exercise can be more effectively identified and suppressed, thereby improving the accuracy of heart rate detection; based on the first frequency domain filter coefficient and in combination with activity parameters, a second frequency domain filter coefficient is further adjusted to obtain a first frequency spectrum for more refined filtering processing according to signal characteristics under different motion states, thereby further optimizing the filtering effect; next, a peak search is performed on the second spectrum through activity parameters to determine a heart rate value to be confirmed, which can preliminarily screen out signals related to the heart rate and lay the foundation for subsequent determination of the heart rate value; then, each heart rate value to be confirmed is corrected through a resolution refinement method to obtain a final heart rate value, thereby further improving the accuracy of the heart rate value and reducing errors.

[0143] See also Fig.11 , Fig.11 This is a structural block diagram of a heart rate detection device provided in an embodiment of the present application. Fig.11 As shown, the heart rate detection device 1100 includes:

[0144] The state determination unit 1110 is used to obtain the pulse wave signal and acceleration signal of the user and determine the current state of the user, which includes a static state or a moving state;

[0145] The first calculation unit 1120 is used to perform time domain analysis based on the pulse wave signal to obtain at least one first resting heart rate value, and perform frequency domain analysis based on the pulse wave signal to obtain multiple second resting heart rate values ​​when the user is in a stationary state;

[0146] The second calculation unit 1130 is used to perform frequency domain analysis based on the pulse wave signal to obtain multiple first exercise heart rate values, and perform deep learning analysis based on the pulse wave signal and the acceleration signal to obtain a second exercise heart rate value when the user is in a state of exercise;

[0147] The result determination unit 1140 is used to determine the user's target heart rate value based on each first resting heart rate value and each second resting heart rate value, or to determine the user's target heart rate value based on each first exercise heart rate value and each second exercise heart rate value.

[0148] Optionally, the heart rate detection device 1100 also includes: a bandpass filtering unit, used to perform time-domain bandpass filtering on the pulse wave signal according to a preset heart rate frequency band to obtain a pulse wave bandpass signal; a first calculation unit 1120, also used to perform peak and trough search on the pulse wave bandpass signal and / or establish a probability density function to obtain at least one first resting heart rate value.

[0149] Optionally, the heart rate detection device 1100 also includes: a signal differential unit, which is used to differentiate the pulse wave bandpass signal to obtain a pulse wave differential signal after the bandpass filtering unit obtains the pulse wave bandpass signal; perform peak and trough search on the pulse wave differential signal and / or establish a probability density function to obtain at least one first resting heart rate value.

[0150] Optionally, the first calculation unit 1120 or the second calculation unit 1130 is further configured to obtain a first frequency spectrum corresponding to the pulse wave bandpass signal, and search for a heart rate value on the first frequency spectrum.

[0151] Optionally, the bandpass filtering unit is further used to perform time-domain bandpass filtering on the acceleration signal according to a preset heart rate frequency band to obtain an acceleration bandpass signal; the first calculation unit 1120 or the second calculation unit 1130 is further used to obtain a pulse wave power spectrum density corresponding to the pulse wave bandpass signal, and obtain an acceleration power spectrum density corresponding to the acceleration bandpass signal; obtain a priori signal-to-noise ratio and a posteriori signal-to-noise ratio of the acceleration bandpass signal corresponding to the pulse wave bandpass signal based on the pulse wave power spectrum density and the acceleration power spectrum density; filter the first spectrum based on the priori signal-to-noise ratio and the posteriori signal-to-noise ratio to obtain a second spectrum corresponding to the pulse wave bandpass signal, and search for heart rate values ​​on the second spectrum.

[0152] Optionally, the first calculation unit 1120 or the second calculation unit 1130 is further configured to determine a first frequency domain filter coefficient according to a priori signal-to-noise ratio and a posteriori signal-to-noise ratio, and filter the first spectrum according to the first frequency domain filter coefficient.

[0153] Optionally, the heart rate detection device 1100 also includes: an activity parameter determination unit, which is used to determine the user's activity parameters according to the acceleration bandpass signal after the bandpass filtering unit performs time domain bandpass filtering on the acceleration signal according to a preset heart rate frequency band to obtain an acceleration bandpass signal; the first calculation unit 1120 or the second calculation unit 1130 is also used to determine the first frequency domain filter coefficient through the prior signal-to-noise ratio and the a posteriori signal-to-noise ratio, and determine the second frequency domain filter coefficient through the first frequency domain filter coefficient and the activity parameter, and filter the first spectrum according to the second frequency domain filter coefficient.

[0154] Optionally, the first calculation unit 1120 or the second calculation unit 1130 is further configured to perform a peak search on the second frequency spectrum according to the activity parameter to determine a plurality of heart rate values ​​to be confirmed; and to correct each heart rate value to be confirmed by a resolution refinement method.

[0155] Optionally, the second calculation unit 1130 is further used to obtain an acceleration spectrum corresponding to the acceleration bandpass signal; input the second spectrum, the acceleration spectrum and the activity parameters into a preset deep learning model; and obtain the exercise heart rate value output by the deep learning model.

[0156] Optionally, the state judgment unit 1110 is further configured to judge that the user is in a stationary state when the activity parameter is less than a preset threshold value; and to judge that the user is in a moving state when the activity parameter is greater than or equal to the preset threshold value.

[0157] In an embodiment of the present application, a heart rate detection device is provided, wherein a state judgment unit is used to obtain a user's pulse wave signal and an acceleration signal to judge the user's current state, wherein the current state includes a static state or a motion state; a first calculation unit is used to perform time domain analysis based on the pulse wave signal when the user is in a static state to obtain at least one first static heart rate value, and to perform frequency domain analysis based on the pulse wave signal to obtain multiple second static heart rate values; a second calculation unit is used to perform frequency domain analysis based on the pulse wave signal when the user is in a motion state to obtain multiple first motion heart rate values, and to perform deep learning analysis based on the pulse wave signal and the acceleration signal to obtain a second motion heart rate value; a result determination unit is used to determine the user's target heart rate value based on each first static heart rate value and each second static heart rate value, or to determine the user's target heart rate value based on each first motion heart rate value and the second motion heart rate value. When performing heart rate detection, the pulse wave signal and acceleration signal are first collected through the state judgment unit, which provides a data basis for subsequent state judgment and heart rate analysis. The acceleration signal can effectively assist in distinguishing the pulse wave changes caused by movement from the real heart rate changes, which helps to more comprehensively understand the user's heart rate changes; at the same time, accurately judging the user's current state is a prerequisite for the subsequent selection of an appropriate heart rate detection method, which can ensure that time domain analysis and frequency domain analysis that reduce power consumption are used in a static state, while frequency domain analysis and deep learning analysis are more effectively used to reduce the impact of motion artifacts in a dynamic state; when the user is in a static state, the first calculation unit performs time domain analysis based on the pulse wave signal, which can directly reflect the changing trend of the pulse wave to obtain the basic information of the heart rate. This information is collected and the pulse wave signal is converted from the time domain to the frequency domain to observe the distribution of the pulse wave at different frequencies, thereby obtaining multiple heart rate values ​​in a static state; when the user is in motion, the second calculation unit, in addition to frequency domain analysis, also introduces a deep learning model to conduct a comprehensive analysis of the pulse wave signal and the acceleration signal. The deep learning model can learn and identify the pulse wave change pattern caused by motion, thereby effectively reducing the impact of motion artifacts on heart rate detection and more accurately extracting the true heart rate information in motion; finally, the result determination unit determines the user's target heart rate value based on the heart rate value in a static state or the heart rate value in motion, which can help the user understand his or her heart rate status and serve as an important basis for health management and disease prevention.

[0158] An embodiment of the present application further provides a computer storage medium, which can store multiple instructions, and the instructions are suitable for being loaded by a processor and executing the steps of any method in the above embodiments.

[0159] See also Fig.12 , Fig.12 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Fig.12 As shown, the electronic device 1200 may include: at least one processor 1201 , at least one network interface 1204 , a user interface 1203 , a memory 1205 , and at least one communication bus 1202 .

[0160] The communication bus 1202 is used to realize the connection and communication between these components.

[0161] The user interface 1203 may include a display screen (Display) and a camera (Camera), and the optional user interface 1203 may also include a standard wired interface and a wireless interface.

[0162] The network interface 1204 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0163] Among them, the processor 1201 may include one or more processing cores. The processor 1201 uses various interfaces and lines to connect various parts within the entire electronic device 1200, and executes various functions and processes data of the electronic device 1200 by running or executing instructions, programs, code sets or instruction sets stored in the memory 1205, and calling data stored in the memory 1205. Optionally, the processor 1201 can be implemented in at least one hardware form of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), and programmable logic array (Programmable Logic Array, PLA). The processor 1201 can integrate one or a combination of a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU) and a modem. Among them, the CPU mainly processes the operating system, user interface and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 1201, and it can be implemented separately through a chip.

[0164] Among them, the memory 1205 may include a random access memory (Random Access Memory, RAM) and may also include a read-only memory (Read-Only Memory, ROM). Optionally, the memory 1205 includes a non-transitory computer-readable storage medium. The memory 1205 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 1205 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 1205 may also be optionally at least one storage device located away from the aforementioned processor 1201. As Fig.12 As shown, the memory 1205 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a heart rate detection program.

[0165] exist Fig.12 In the electronic device 1200 shown, the user interface 1203 is mainly used to provide an input interface for the user and obtain the data input by the user; and the processor 1201 can be used to call the heart rate detection program stored in the memory 1205 and specifically perform the following operations:

[0166] Obtain the user's pulse wave signal and acceleration signal to determine the user's current state, which includes a stationary state or a moving state;

[0167] When the user is in a stationary state, performing time domain analysis based on the pulse wave signal to obtain at least one first stationary heart rate value, and performing frequency domain analysis based on the pulse wave signal to obtain multiple second stationary heart rate values;

[0168] When the user is in motion, frequency domain analysis is performed based on the pulse wave signal to obtain a plurality of first exercise heart rate values, and deep learning analysis is performed based on the pulse wave signal and the acceleration signal to obtain a second exercise heart rate value;

[0169] The user's target heart rate value is determined based on each first resting heart rate value and each second resting heart rate value, or the user's target heart rate value is determined based on each first exercise heart rate value and each second exercise heart rate value.

[0170] In some embodiments, after acquiring the user's pulse wave signal and acceleration signal, the processor 1201 further specifically performs the following steps: performing time-domain bandpass filtering on the pulse wave signal according to a preset heart rate frequency band to obtain a pulse wave bandpass signal; when the processor 1201 performs time-domain analysis based on the pulse wave signal to obtain at least one first resting heart rate value, the processor 1201 further specifically performs the following steps: performing peak and trough search on the pulse wave bandpass signal and / or establishing a probability density function to obtain at least one first resting heart rate value.

[0171] In some embodiments, after obtaining the pulse wave bandpass signal, the processor 1201 further specifically performs the following steps: differentiating the pulse wave bandpass signal to obtain a pulse wave differential signal; performing peak and trough search and / or establishing a probability density function on the pulse wave differential signal to obtain at least one first resting heart rate value.

[0172] In some embodiments, after obtaining the user's pulse wave signal and acceleration signal, the processor 1201 further specifically performs the following steps: performing time domain bandpass filtering on the pulse wave signal according to a preset heart rate frequency band to obtain a pulse wave bandpass signal; when performing frequency domain analysis based on the pulse wave signal, the processor 1201 specifically performs the following steps: obtaining a first spectrum corresponding to the pulse wave bandpass signal, and searching for a heart rate value for the first spectrum.

[0173] In some embodiments, after the processor 1201 obtains the pulse wave signal and acceleration signal of the user, it further specifically performs the following steps: performing time-domain bandpass filtering on the acceleration signal according to a preset heart rate frequency band to obtain an acceleration bandpass signal; when the processor 1201 performs a heart rate value search on the first spectrum, it further specifically performs the following steps: obtaining a pulse wave power spectrum density corresponding to the pulse wave bandpass signal, and obtaining an acceleration power spectrum density corresponding to the acceleration bandpass signal;

[0174] Obtaining a priori signal-to-noise ratio and a posteriori signal-to-noise ratio of the pulse wave bandpass signal corresponding to the acceleration bandpass signal according to the pulse wave power spectrum density and the acceleration power spectrum density;

[0175] The first spectrum is filtered according to the priori signal-to-noise ratio and the a posteriori signal-to-noise ratio to obtain a second spectrum corresponding to the pulse wave bandpass signal, and the heart rate value is searched for the second spectrum.

[0176] In some embodiments, when the processor 1201 performs filtering on the first spectrum according to the a priori signal-to-noise ratio and the a posteriori signal-to-noise ratio, it specifically performs the following steps: determining a first frequency domain filter coefficient through the a priori signal-to-noise ratio and the a posteriori signal-to-noise ratio, and filtering the first spectrum through the first frequency domain filter coefficient.

[0177] In some embodiments, after performing time-domain bandpass filtering on the acceleration signal according to a preset heart rate frequency band to obtain an acceleration bandpass signal, the processor 1201 further specifically performs the following steps: determining the user's activity parameters based on the acceleration bandpass signal; when the processor 1201 performs filtering of the first spectrum based on the priori signal-to-noise ratio and the a posteriori signal-to-noise ratio, the processor 1201 further specifically performs the following steps: determining a first frequency domain filter coefficient through the priori signal-to-noise ratio and the a posteriori signal-to-noise ratio, determining a second frequency domain filter coefficient through the first frequency domain filter coefficient and the activity parameters, and filtering the first spectrum based on the second frequency domain filter coefficient.

[0178] In some embodiments, when the processor 1201 searches for the heart rate value on the second spectrum, it specifically performs the following steps: searching for peaks on the second spectrum based on activity parameters to determine multiple heart rate values ​​to be confirmed; and correcting each heart rate value to be confirmed by a resolution refinement method.

[0179] In some embodiments, when the processor 1201 performs deep learning analysis based on the pulse wave signal and the acceleration signal, it specifically performs the following steps: obtaining the acceleration spectrum corresponding to the acceleration bandpass signal; inputting the second spectrum, the acceleration spectrum and the activity parameters into a preset deep learning model; and obtaining the exercise heart rate value output by the deep learning model.

[0180] In some embodiments, when the processor 1201 determines the current state of the user, it specifically performs the following steps: when the activity parameter is less than a preset threshold, it determines that the user is in a stationary state; when the activity parameter is greater than or equal to the preset threshold, it determines that the user is in motion.

[0181] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0182] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0183] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The above computer program product includes one or more computer instructions. When the above computer program instructions are loaded and executed on a computer, the above process or function according to the embodiment of this specification is generated in whole or in part. The above computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The above computer instructions can be stored in a computer-readable storage medium or transmitted by the above computer-readable storage medium. The above computer instructions can be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (Digital Subscriber Line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode. The above computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server, data center, etc. that contains one or more available media integrated. The above-mentioned available media can be magnetic media (for example, floppy disks, hard disks, tapes), optical media (for example, digital versatile discs (DVD)), or semiconductor media (for example, solid state drives (SSD)), etc.

[0184] It should be noted that, for the convenience of description, the aforementioned method embodiments are all described as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described action sequence, because according to the present application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0185] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0186] The above is a description of a heart rate detection method, device, storage medium and electronic device provided in the present application. For technicians in this field, according to the ideas of the embodiments of the present application, there may be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A heart rate detection method, characterized in that: The method comprises: Acquire a pulse wave signal and an acceleration signal of a user to determine a current state of the user, where the current state includes a stationary state or a moving state; When the user is in a stationary state, performing a time domain analysis based on the pulse wave signal to obtain at least one first stationary heart rate value, and performing a frequency domain analysis based on the pulse wave signal to obtain a plurality of second stationary heart rate values; When the user is in motion, frequency domain analysis is performed based on the pulse wave signal to obtain a plurality of first exercise heart rate values, and deep learning analysis is performed based on the pulse wave signal and the acceleration signal to obtain a second exercise heart rate value; The target heart rate value of the user is determined according to each first resting heart rate value and each second resting heart rate value, or the target heart rate value of the user is determined according to each first exercise heart rate value and each second exercise heart rate value.

2. The method according to claim 1, characterized in that: After obtaining the pulse wave signal and acceleration signal of the user, the method further includes: Performing time-domain bandpass filtering on the pulse wave signal according to a preset heart rate frequency band to obtain a pulse wave bandpass signal; The step of performing time domain analysis based on the pulse wave signal to obtain at least one first resting heart rate value includes: The pulse wave bandpass signal is subjected to a peak and trough search and / or a probability density function is established to obtain at least one first resting heart rate value.

3. The method according to claim 2, characterized in that After obtaining the pulse wave bandpass signal, the method further includes: Differentiating the pulse wave bandpass signal to obtain a pulse wave differential signal; The pulse wave differential signal is subjected to a peak and trough search and / or a probability density function is established to obtain at least one first resting heart rate value.

4. The method according to claim 1, characterized in that: After obtaining the pulse wave signal and acceleration signal of the user, the method further includes: Performing time-domain bandpass filtering on the pulse wave signal according to a preset heart rate frequency band to obtain a pulse wave bandpass signal; The frequency domain analysis based on the pulse wave signal comprises: A first frequency spectrum corresponding to the pulse wave bandpass signal is obtained, and a heart rate value is searched for the first frequency spectrum.

5. The method according to claim 4, characterized in that After obtaining the pulse wave signal and acceleration signal of the user, the method further includes: Performing time-domain bandpass filtering on the acceleration signal according to a preset heart rate frequency band to obtain an acceleration bandpass signal; The searching for the heart rate value of the first frequency spectrum includes: Obtaining a pulse wave power spectrum density corresponding to the pulse wave bandpass signal, and obtaining an acceleration power spectrum density corresponding to the acceleration bandpass signal; Acquire a priori signal-to-noise ratio and a posteriori signal-to-noise ratio of the pulse wave bandpass signal corresponding to the acceleration bandpass signal according to the pulse wave power spectrum density and the acceleration power spectrum density; The first spectrum is filtered according to the priori signal-to-noise ratio and the a posteriori signal-to-noise ratio to obtain a second spectrum corresponding to the pulse wave bandpass signal, and a heart rate value is searched for the second spectrum.

6. The method according to claim 5, characterized in that The filtering the first spectrum according to the a priori signal-to-noise ratio and the a posteriori signal-to-noise ratio includes: A first frequency domain filter coefficient is determined by the a priori signal-to-noise ratio and the a posteriori signal-to-noise ratio, and the first frequency spectrum is filtered by the first frequency domain filter coefficient.

7. The method according to claim 5, characterized in that After performing time-domain bandpass filtering on the acceleration signal according to the preset heart rate frequency band to obtain the acceleration bandpass signal, the method further includes: determining an activity parameter of the user according to the acceleration bandpass signal; The filtering the first spectrum according to the a priori signal-to-noise ratio and the a posteriori signal-to-noise ratio includes: A first frequency domain filter coefficient is determined by the priori signal-to-noise ratio and the a posteriori signal-to-noise ratio, and a second frequency domain filter coefficient is determined by the first frequency domain filter coefficient and the activity parameter, and the first frequency spectrum is filtered according to the second frequency domain filter coefficient.

8. The method according to claim 7, characterized in that The searching for the heart rate value of the second frequency spectrum includes: Performing a peak search on the second frequency spectrum according to the activity parameter to determine a plurality of heart rate values ​​to be confirmed; Each heart rate value to be confirmed is corrected by the resolution refinement method.

9. The method according to claim 7, characterized in that: The deep learning analysis based on the pulse wave signal and the acceleration signal includes: Acquire an acceleration spectrum corresponding to the acceleration bandpass signal; Inputting the second spectrum, the acceleration spectrum and the activity parameter into a preset deep learning model; Obtain the exercise heart rate value output by the deep learning model.

10. The method according to claim 7, characterized in that The determining the current state of the user includes: When the activity parameter is less than a preset threshold, it is determined that the user is in a stationary state; When the activity parameter is greater than or equal to the preset threshold, it is determined that the user is in motion.

11. A heart rate detection device, characterized in that: The device comprises: A state determination unit, used to obtain a pulse wave signal and an acceleration signal of a user, and determine a current state of the user, wherein the current state includes a stationary state or a moving state; a first calculation unit, configured to, when the user is in a stationary state, perform a time domain analysis based on the pulse wave signal to obtain at least one first stationary heart rate value, and perform a frequency domain analysis based on the pulse wave signal to obtain a plurality of second stationary heart rate values; a second calculation unit, configured to, when the user is in motion, perform frequency domain analysis based on the pulse wave signal to obtain a plurality of first exercise heart rate values, and perform deep learning analysis based on the pulse wave signal and the acceleration signal to obtain a second exercise heart rate value; The result determination unit is used to determine the target heart rate value of the user according to each first resting heart rate value and each second resting heart rate value, or to determine the target heart rate value of the user according to each first exercise heart rate value and each second exercise heart rate value.

12. A computer storage medium, characterized in that: The computer storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the steps of the method according to any one of claims 1 to 10.

13. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method according to any one of claims 1 to 10 when executing the program.

Images