A sleep posture recognition method and device, electronic equipment and storage medium

Abstract

The application discloses a sleep posture recognition method and device, electronic equipment and storage medium. A first electromagnetic echo signal received by a radar device within a first preset time length is acquired; wherein the radiation range of the radar device includes a sleep area of a target object; based on the first electromagnetic echo signal, a prone reference characteristic value and a supine reference characteristic value corresponding to the target object are determined; when it is detected that the target object changes posture after the first preset time length, a second electromagnetic echo signal received by the radar device within a second preset time length is acquired, and based on the second electromagnetic echo signal, current posture characteristic data corresponding to the target object is determined; based on the current posture characteristic data, the prone reference characteristic value and the supine reference characteristic value, a current sleep posture category corresponding to the target object is determined, so as to realize contactless recognition of a sleep posture, improve the convenience of sleep posture recognition, and improve the sleep posture recognition accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a sleep posture recognition method and device, an electronic device, and a storage medium. BACKGROUND

[0002] Sleep is a process of rest and recovery of physical strength, which is essential to human beings. The quality of sleep and sleep disorders are related to specific body postures during sleep. Recognizing sleep postures can help monitor the sleep quality of a person, thereby further assessing the physical health condition. In addition, the recognition of sleep postures plays an important role in preventing sudden infant death syndrome, nursing pressure ulcer patients, and helping sleep apnea patients, and thus it has important practical significance to recognize sleep postures.

[0003] At present, posture state data of a user during sleep is usually collected by using various sensors, and a neural network model is used to recognize sleep postures based on the collected posture state data. The training data of the neural network model comes from different subjects.

[0004] However, this recognition method requires the user to wear various sensors, which will cause a greater foreign body sensation to the user, thereby affecting the sleep quality. In addition, different users have large individual differences, and the use of a traditional neural network model for recognition results in a low sleep posture recognition accuracy. SUMMARY

[0005] Embodiments of the present application provide a sleep posture recognition method, device, electronic device, and storage medium to realize contactless recognition of sleep postures, improve the convenience of sleep posture recognition, and improve the sleep posture recognition accuracy.

[0006] In a first aspect, the present application provides a sleep posture recognition method, which comprises:

[0007] obtaining a first electromagnetic echo signal received by a radar device within a first preset time length; wherein the radiation range of the radar device includes a sleep area of a target object;

[0008] determining a prone reference feature value and a supine reference feature value corresponding to the target object based on the first electromagnetic echo signal;

[0009] when it is detected that the target object changes posture after the first preset time length, obtaining a second electromagnetic echo signal received by the radar device within a second preset time length, and determining current posture feature data corresponding to the target object based on the second electromagnetic echo signal;

[0010] Determine a current sleep posture category corresponding to the target object based on the current posture feature data, the prone reference feature value and the supine reference feature value.

[0011] In a second aspect, the present application provides a sleep posture recognition device, which comprises:

[0012] A reference signal acquisition module is configured to acquire a first electromagnetic echo signal received by a radar device within a first preset time length, wherein a radiation range of the radar device comprises a sleep area of a target object.

[0013] A reference feature determination module is configured to determine a prone reference feature value and a supine reference feature value corresponding to the target object based on the first electromagnetic echo signal.

[0014] A feature data determination module is configured to acquire a second electromagnetic echo signal received by the radar device within a second preset time length when detecting a posture change of the target object after the first preset time length, and determine current posture feature data corresponding to the target object based on the second electromagnetic echo signal.

[0015] A current sleep posture determination module is configured to determine a current sleep posture category corresponding to the target object based on the current posture feature data, the prone reference feature value and the supine reference feature value.

[0016] In a third aspect, the present application provides a device, which comprises:

[0017] At least one processor; and

[0018] A memory in communication connection with the at least one processor; wherein

[0019] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the sleep posture recognition method of any one of the embodiments of the present application.

[0020] In a fourth aspect, the present application provides a computer readable storage medium, which stores computer instructions for enabling a processor to execute the sleep posture recognition method of any one of the embodiments of the present application.

[0021] The technical scheme provided by the embodiment of the present application comprises the following steps: obtaining a first electromagnetic echo signal received by a radar device within a first preset time length, wherein a radiation range of the radar device comprises a sleep area of a target object; determining a prone reference characteristic value and a supine reference characteristic value corresponding to the target object based on the first electromagnetic echo signal, so as to obtain a personalized reference characteristic value of the target user; when it is detected that the target object has a posture change after the first preset time length, obtaining a second electromagnetic echo signal received by the radar device within a second preset time length, and determining current posture characteristic data corresponding to the target object based on the second electromagnetic echo signal; and then determining a current sleep posture category corresponding to the target object based on the current posture characteristic data, the prone reference characteristic value and the supine reference characteristic value. The present scheme solves the technical problem of low sleep posture recognition accuracy based on a wearable sensor device, does not require the user to wear various sensor devices, does not bring a foreign body sensation to the user, can realize contactless recognition of a sleep posture, improves the convenience of sleep posture recognition, and improves the sleep posture recognition accuracy by obtaining a personalized sleep posture characteristic reference value of the target object.

[0022] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0024] Figure 1 A flowchart of a sleep posture recognition method provided by the first embodiment of the present application;

[0025] Figure 2 A schematic diagram of the relative position of the radar device and the target object related to the first embodiment of the present application;

[0026] Figure 3 A radar echo signal schematic diagram related to the first embodiment of the present application;

[0027] Figure 4 A target feature map schematic diagram related to the first embodiment of the present application;

[0028] Figure 5 A flowchart of a sleep posture recognition method provided by the second embodiment of the present application;

[0029] Figure 6A sleep posture recognition device structure schematic diagram provided for the third embodiment of the present application;

[0030] Figure 7 A device structure schematic diagram provided for the fourth embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to enable persons skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by persons skilled in the art without creative labor should fall within the scope of protection of the present application.

[0032] It should be noted that the terms "first preset condition", "second preset condition" and the like in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0033] Embodiment one

[0034] Figure 1 A flowchart of a sleep posture recognition method provided for the first embodiment of the present application, the present embodiment can be applicable to the case of recognizing the sleep posture of a target object according to the electromagnetic echo signal of a radar. The method can be executed by a sleep posture recognition device, which can be realized in the form of hardware and / or software, and can be configured on a computer device, which can be a notebook, a desktop computer, a smart tablet and the like. As shown in the figure, the method comprises: Figure 1

[0035] S110, obtaining a first electromagnetic echo signal received by a radar device within a first preset time length.

[0036] The radiation range of the radar device includes the sleep area of the target object. The target object is a user who needs to be recognized for sleep posture. For example, a schematic diagram of the relative position between the radar device and the target object is shown in the figure Figure 2 Figure 2 ​​As shown, the radar device is installed at a position associated with a target bed, and a target object lies on the target bed when the target object is identified in a sleep posture. The electromagnetic echo signal is a reflected electromagnetic wave corresponding to a detection electromagnetic wave emitted by the radar device. The first preset time length is a pre-set time, for example, the first preset time length is 8 hours. The first electromagnetic echo signal is an electromagnetic echo signal in the first preset time length.

[0037] Specifically, the signal transmitter of the radar device periodically sends a detection electromagnetic wave signal to the sleep area of the target object, and the electromagnetic signal received by the signal receiver of the radar device after the detection electromagnetic wave signal is scattered by the target object is an electromagnetic echo signal. In order to improve the accuracy of the real-time sleep posture detection result of the target user, the sleep posture features corresponding to the target object can be learned in advance, and based on this, the first electromagnetic echo signal of the target user in the first preset time length is obtained before the target object is detected in real time. The sleep posture, so as to extract the personalized data features of the target user based on the first electromagnetic echo signal.

[0038] For example, the first preset time length is 8 hours, the target object starts to sleep in the radar device radiation area at 10 o'clock in the evening of the first day, and ends to sleep at 6 o'clock in the morning of the second day. The electromagnetic echo data collected by the radar device in the period from 10 o'clock in the evening of the first day to 6 o'clock in the morning of the second day is the first electromagnetic echo data.

[0039] In this embodiment, if the detection electromagnetic wave signal emitted by the radar device is a pulse signal, the transmitter of the radar device can emit a periodic pulse sequence, and this period can be defined as a pulse repetition interval. The receiver of the radar device can accept the radar echo signal corresponding to the detection electromagnetic wave signal. The radar echo signal schematic diagram is shown in Figure 3 As shown in Figure 3 The radar echo signal is one-to-one corresponding to the detection electromagnetic wave signal, so the radar echo signal is also a periodic pulse sequence. Each pulse sequence of the radar echo signal is stored in a row, for example, the first pulse of the radar echo signal is placed in the first row, and the second pulse of the radar echo signal is placed in the second row, and so on. In general, the dimension viewed in the row direction is defined as the fast time dimension, and since the data sampling interval between rows is often greater than the pulse duration, the dimension viewed in the column direction is defined as the slow time dimension.

[0040] In the embodiment, after obtaining the electromagnetic echo signal received by the radar device, the electromagnetic echo signal needs to be preprocessed to remove the clutter in the electromagnetic echo signal and retain the pure electromagnetic echo signal. Based on this, the received electromagnetic echo signal is first subjected to Fourier transform in the fast time dimension, and then static clutter suppression is performed to obtain the pure electromagnetic echo signal. In subsequent processing, the preprocessed electromagnetic echo signal is processed, so in the subsequent embodiments, the electromagnetic echo signal mentioned is the preprocessed electromagnetic echo signal.

[0041] S120, determine the prone reference feature value and the supine reference feature value corresponding to the target object based on the first electromagnetic echo signal.

[0042] The prone reference feature value is a characteristic quantity when the target object's sleep posture is a prone posture. The supine reference feature value is a characteristic quantity when the target object's sleep posture is a supine posture.

[0043] Specifically, in the first preset time length, the target user corresponds to multiple different sleep postures, so the first electromagnetic echo signal is first divided into multiple electromagnetic echo signal segments based on the time of the user turning over, and the posture feature data corresponding to each electromagnetic echo data segment is extracted. For each electromagnetic echo data segment, based on the side-lying feature data in the posture feature data for distinguishing the side-lying posture and the non-side-lying posture, it is determined whether the sleep posture corresponding to this signal segment is a side-lying posture. Further, the signal segment with the non-side-lying posture in the first electromagnetic echo signal is extracted as a prone-supine echo signal segment, and then the prone-supine echo signal segment is subjected to feature sorting to determine the prone data feature corresponding to the signal segment with the prone posture and the supine data feature corresponding to the signal segment with the supine posture. The prone reference feature value is determined based on the prone data feature, and the supine reference feature value is determined based on the supine data feature.

[0044] In the embodiment, the manner of determining the posture feature data is the same for each electromagnetic echo signal segment, which is exemplarily described by taking one of the electromagnetic echo signal segments as an example. To extract the posture feature data corresponding to the electromagnetic echo signal segment, first, a target feature map corresponding to the electromagnetic echo signal segment is determined. The target feature map is used to represent the correspondence between the distance of the target object and the radar device, the Doppler frequency value and the signal amplitude value, so the target feature map is also called a range-Doppler map. For the electromagnetic echo signal segment, the target feature map can be obtained by performing Fourier transform along the slow time dimension. The electromagnetic echo signal can be represented by a matrix, which can be referred to as a radar echo matrix. The row vector of the radar echo matrix corresponds to the fast time dimension, and the column vector corresponds to the slow time dimension. The Fourier transform matrix can be obtained by performing Fourier transform on each column vector of the radar echo matrix, that is, performing Fourier transform on each column vector of the radar echo matrix.

[0045]

[0046] wherein x(k, n) is the data in a time window, and the time window is a column vector, so x(k, n) is the data of a column vector. k ∈ [1, K], K is the length of the time window, ω is the transform frequency, and w(m) is a Hamming window function.

[0047] The dimension of the Fourier transform matrix obtained after the Fourier transform is the same as that of the radar echo matrix. For example, the radar echo matrix is a 1000*500 matrix, and the Fourier transform matrix is also a 1000*500 matrix. Further, the Fourier transform matrix is converted into a target feature map by a drawing program. The value corresponding to each feature point in the target feature map is the amplitude of each element of the Fourier transform matrix. After the target feature map is generated, the target feature map is denoised to obtain a pure target feature map with higher signal-to-noise ratio, which can be represented as Here, the denoising process can use mean filtering, Gaussian filtering, median filtering, bilateral filtering, etc., which is not specifically limited here.

[0048] In the target feature map, the horizontal coordinate is the distance of the target object and the radar device, and the vertical coordinate is the Doppler frequency value. The feature points in the target feature map are the amplitudes of the elements of the Fourier transform matrix. Exemplarily, the generated target feature map is shown in FIG. 2. Figure 4Wherein, (A) is the target feature map corresponding to the supine posture, (B) is the target feature map corresponding to the lateral posture, and (C) is the target feature map corresponding to the lateral posture. In actual application, the target feature map is a color image, and the target feature map is composed of a large number of feature points, each feature point corresponding to a different amplitude. The greater the amplitude, the redder the color of the corresponding feature point, and the smaller the amplitude, the bluer the color of the corresponding feature point.

[0049] After obtaining the target feature map, the function expression form corresponding to the posture feature of the target feature map is further extracted, and the function expression column vector amplitude information G1(n), the first shape feature information G2(n), the second shape feature information G3(n), and the row vector amplitude information G4(n) corresponding to the amplitude information and the shape information of the target feature map are extracted. Further, based on the specific feature data in the target feature map and the function expression form G1(n), G2(n), G3(n), and G4(n) corresponding to the posture feature, a plurality of posture feature data corresponding to the electromagnetic echo signal segment are determined. The determined posture feature data can include but are not limited to: the second non-zero column feature point amplitude F1, the third non-zero column feature point amplitude F2, the second largest non-zero column feature point maximum amplitude F3, and the third largest non-zero column feature point maximum amplitude F4, the second non-zero column head size F5, the third non-zero column head size F6, the first threshold micro-motion frequency distribution F7, the second threshold micro-motion frequency distribution F8, the first column number amplitude distribution F9, the second column number amplitude distribution F 10 , and the maximum Doppler frequency F 11 of the body trunk. Wherein, F1-F6 are lateral feature data; F7-F 11 are supine feature data.

[0050] S130, when it is detected that the target object changes posture after a first preset time length, a second electromagnetic echo signal received by the radar device within a second preset time length is obtained, and based on the second electromagnetic echo signal, current posture feature data corresponding to the target object is determined.

[0051] Wherein, the second preset time length is used to distinguish from the first preset time length. The second preset time length is a pre-set time length, for example, the second preset time length is 1 minute. The second electromagnetic echo signal is the electromagnetic echo signal of the second preset time length. The current posture feature data is used to represent the specific feature quantity corresponding to the sleep posture of the target object at the current time.

[0052] Specifically, the personalized reference feature of the target object is acquired after the first preset time length. After that, when a posture change of the target object is detected, a second electromagnetic echo signal in a second preset time length starting from the posture change time is acquired. A current target feature map is generated according to the second electromagnetic echo signal, and current amplitude information and current shape information are extracted from the current target feature map, and then current posture feature data corresponding to the target object is determined based on the current amplitude information and the current shape information.

[0053] In S140, the current sleep posture category corresponding to the target object is determined based on the current posture feature data, the prone reference feature value and the supine reference feature value.

[0054] The current sleep posture category includes a side-lying posture category, a prone posture category or a supine posture category. The current posture feature data includes current side-lying feature data and current prone-supine feature data.

[0055] Optionally, S140 specifically includes the following steps: based on the current side-lying feature data, it is determined whether the current sleep posture category corresponding to the target object is a side-lying posture category; if not, the current sleep posture category corresponding to the target object is determined based on the current prone-supine feature data, the prone reference feature value and the supine reference feature value.

[0056] Specifically, based on the current side-lying feature data in the current posture feature data and the pre-trained side-lying classifier, it is determined whether the sleep posture of the target object at the current time is a side-lying posture or a non-side-lying posture. If the result output by the side-lying classifier indicates that the sleep posture of the target object at the current time is a side-lying posture, the current sleep posture category corresponding to the target object is a side-lying posture category. If the result output by the side-lying classifier indicates that the sleep posture of the target object at the current time is a non-side-lying posture, it is necessary to further determine whether the sleep posture of the target object at the current time is a prone posture category or a supine posture category. The specific implementation process can be: calculating the Euclidean distance between the current prone-supine feature data and the prone reference feature value to determine the first similarity between the current prone-supine feature data and the prone reference feature value; calculating the Euclidean distance between the current prone-supine feature data and the supine reference feature value to determine the second similarity between the current prone-supine feature data and the supine reference feature value, and determining whether the current sleep posture category corresponding to the target object is a prone posture category or a supine posture category based on the first similarity and the second similarity.

[0057] In another embodiment, based on the current side-lying feature data in the current posture feature data, it can be determined whether the sleep posture of the target object at the current time corresponds to a side-lying posture or a non-side-lying posture by means of voting. For example, if the second electromagnetic echo signal of 1 minute is obtained, it can be determined whether the sleep posture corresponding to the current side-lying feature data of the second electromagnetic echo signal of each second corresponds to a side-lying posture or a non-side-lying posture, and then the sleep posture of the target object is determined to be a side-lying posture or a non-side-lying posture by means of voting based on 60 prediction results. For example, if 40 results of the 60 prediction results are side-lying postures, the sleep posture of the target object is determined to be a side-lying posture or a non-side-lying posture.

[0058] The technical scheme provided by the embodiment of the present application obtains the first electromagnetic echo signal received by the radar device within the first preset time length, wherein the radiation range of the radar device includes the sleep area of the target object, and determines the prone reference feature value and the supine reference feature value corresponding to the target object based on the first electromagnetic echo signal, so that the personalized reference feature value of the target user can be obtained. When it is detected that the posture of the target object changes after the first preset time length, the second electromagnetic echo signal received by the radar device within the second preset time length is obtained, and the current posture feature data corresponding to the target object is determined based on the second electromagnetic echo signal, and then the current sleep posture category corresponding to the target object is determined based on the current posture feature data, the prone reference feature value and the supine reference feature value. The present scheme solves the technical problem of low accuracy of sleep posture recognition based on a wearable sensor device, and does not require the user to wear various sensor devices, so as to avoid the foreign body sensation of the user, realize contactless recognition of the sleep posture, improve the convenience of sleep posture recognition, and improve the accuracy of sleep posture recognition by obtaining the personalized sleep posture feature reference value of the target object.

[0059] Embodiment two

[0060] Figure 5 The flowchart of the sleep posture recognition method provided by the embodiment two of the present application is based on the above-mentioned embodiments, and the steps of "determining the prone reference feature value and the supine reference feature value corresponding to the target object based on the first electromagnetic echo signal" and "determining the current posture feature data corresponding to the target object based on the second electromagnetic echo signal" are further optimized. The embodiment of the present application can be combined with each optional scheme in one or more of the above-mentioned embodiments. As shown in the figure, the method comprises: Figure 5

[0061] S210, obtaining the first electromagnetic echo signal received by the radar device within the first preset time length; wherein the radiation range of the radar device includes the sleep area of the target object.

[0062] ​S220, dividing the first electromagnetic echo signal into at least one electromagnetic echo signal segment, and determining posture feature data of each electromagnetic echo signal segment.

[0063] The posture feature data includes side-lying feature data and prone-tilt feature data. On the basis of the above embodiment, the posture feature data includes a second non-zero column feature point amplitude F1, a third non-zero column feature point amplitude F2, a second largest non-zero column feature point maximum amplitude F3 and a third largest non-zero column feature point maximum amplitude F4, a second non-zero column head size F5, a third non-zero column head size F6, a first threshold micro-motion frequency distribution F7, a second threshold micro-motion frequency distribution F8, a first column number amplitude distribution F9, a second column number amplitude distribution F 10 , and a body trunk maximum Doppler frequency F 11 F1-F6 are side-lying feature data; F7-F 11 are prone-tilt feature data.

[0064] In the embodiment, dividing the first electromagnetic echo signal into at least one electromagnetic echo signal segment can be understood as a process of separating human states, that is, the "motion" and "lying" states of the human body can be separated by a body motion index estimation, and the "motion" is taken as a segmentation point to divide the first electromagnetic echo signal data into multiple electromagnetic echo signal segments, and the human body maintains the same posture in a segment of data.

[0065] Optionally, in step S220, the posture feature data of each electromagnetic echo signal segment is determined, specifically including the following steps:

[0066] (1) For each electromagnetic echo signal segment, a target feature map is generated based on the electromagnetic echo signal;

[0067] In the embodiment, the specific implementation process of step (1) has been described in detail in the above embodiment, which will not be repeated here.

[0068] (2) Extracting amplitude information and shape information in the target feature map;

[0069] Next, how to extract the amplitude information and shape information of the image from the target feature map will be described in detail. The amplitude information specifically includes column vector amplitude information G1(n) and row vector amplitude information G4(n). The shape information specifically includes first shape feature information G3(n) and second shape feature information G3(n).

[0070] Optionally, the amplitude information in the target feature map is extracted, specifically including: determining a column feature amplitude corresponding to each column vector in the target feature map based on amplitude values corresponding to each element in the column vector, and determining column vector amplitude information of the target feature map based on the column feature amplitudes; determining a row feature amplitude corresponding to each row vector in the target feature map based on amplitude values corresponding to each element in the row vector, and determining row vector amplitude information of the target feature map based on the row feature amplitudes.

[0071] In the embodiment, for each column vector in the target feature map, the amplitude values corresponding to each element in the column vector are added as the column feature amplitude corresponding to the column vector, and an array composed of the column feature amplitudes is taken as the column vector amplitude information of the target feature map. For the row vector amplitude information in the target feature map, the amplitude values corresponding to each element in the row vector are also added as the row feature amplitude corresponding to the row vector, and an array composed of the row feature amplitude information is taken as the row vector amplitude information of the target feature map. If the target feature map is represented as the column vector amplitude information of the target feature map can be represented as: the row vector amplitude information of the target feature map can be represented as:

[0072] For example, if the Fourier transform matrix corresponding to the target feature map is represented as:

[0073] 0 0 0 0 0 0 0 1 2 0 0 0 0 2 3 3 3 3 0 3 4 4 2 4 0 0 5 6 5 5 0 2 9 8 7 5 0 1 6 6 6 8 0 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

[0074] On the basis of the above examples, the sum of the amplitude values corresponding to all elements in each column vector is 0, 12, 29, 27, 23, 25 and 0, respectively, and the column vector amplitude information of the target feature map can be represented as G1(n) = {0, 12, 29, 27, 23, 25}. The sum of the amplitude values corresponding to all elements in each row vector is 0, 3, 14, 17, 21, 31, 27, 3, 0 and 0, respectively, and the row vector amplitude information of the target feature map can be represented as G4(n) = {0, 3, 14, 17, 21, 31, 27, 3, 0, 0}.

[0075] Optionally, the shape information in the target feature map is extracted, specifically including the following steps:

[0076] 1) determining a shape feature value corresponding to each element in the target feature map based on the amplitude value corresponding to the element and a preset amplitude threshold.

[0077] wherein the preset amplitude threshold is a preset amplitude value, for example, the preset amplitude threshold is 0.

[0078] In the embodiment, the magnitude value corresponding to each element in the target feature map is compared with the preset magnitude threshold. If the magnitude value corresponding to the element in the target feature map is greater than the preset magnitude threshold, the shape feature value corresponding to the element position is the row index; if the magnitude value corresponding to the element in the target feature map is less than the preset magnitude threshold, the shape feature value corresponding to the element position is zero. For example, assuming that the preset magnitude threshold is 0, the shape feature value corresponding to each element in the target feature map can be represented as:

[0079]

[0080] 2) Determine the shape feature matrix composed of shape feature values, and perform norm operation on each column vector in the shape feature matrix to determine the first shape feature information corresponding to each column.

[0081] On the basis of the above example, after determining the shape feature value corresponding to each element in the target feature map, a new matrix can be obtained, which is taken as the shape feature matrix. If P n = [Q(1, n), Q(2, n), …, Q(H, n)], T , the first shape feature information corresponding to each column can be represented as G2(n) = ||P n ||0. The first shape feature information is used to represent how many points in each column vector have values.

[0082] 3) Determine the second shape feature information corresponding to each column based on the maximum shape feature value and the minimum shape feature value of each column in the shape feature matrix.

[0083] In the embodiment, the maximum shape feature value max η Q(η,n) and the minimum shape feature value min η Q(η,n) of each column vector in the shape feature matrix are determined, and the second shape feature information corresponding to each column can be represented as: G3(n) = max η Q(η,n) - min η Q(η,n).

[0084] (3) Determine the posture feature data of the electromagnetic echo signal segment based on the amplitude information and the shape information.

[0085] In the embodiment, after the extracted amplitude information G1(n) and G4(n), and shape information G2(n) and G3(n), ten posture feature data can be extracted according to G1(n), G2(n), G3(n), G4(n). The extracted ten posture feature data can be divided into four groups, and the following will describe how the ten posture feature data are determined. In addition, the posture feature data include the maximum Doppler frequency of the body trunk in addition to the extraction of ten features according to G1(n), G2(n), G3(n), G4(n).

[0086] The first group: different body part strong feature point amplitude features include the second non-zero column feature point amplitude F1, the third non-zero column feature point amplitude F2, the second largest non-zero column feature point maximum amplitude F3, and the third largest non-zero column feature point maximum amplitude F4, and the specific extraction formula is as follows:

[0087]

[0088]

[0089]

[0090]

[0091] Wherein, N2 is the second non-zero column vector in the target feature map, and N3 is the third non-zero column vector in the target feature map. The second and third largest amplitudes of G1(n) respectively. On the basis of the above example, if G1(n) = {0, 12, 29, 27, 23, 25}, then is 27, is 25.

[0092] The second group: head size features include the second non-zero column head size F5 and the third non-zero column head size F6, and the specific extraction formula is as follows:

[0093]

[0094]

[0095] Wherein, the vector The vector Here, Γ(l), Θ(l) respectively represent the following:

[0096]

[0097]

[0098] Wherein, ε is a threshold value, G 2m , G 3mThe maximum values of G2(l), G3(l) respectively, l', l" are the column numbers corresponding to G 2m , G 3m .

[0099] The third group: the body micro-motion frequency distribution features include a first threshold micro-motion frequency distribution F7 and a second threshold micro-motion frequency distribution F8, and the specific extraction formula is as follows:

[0100] F7 = ||C1||0 (11)

[0101] F8 = ||C2||0 (12)

[0102] Wherein, C1, C2 respectively represent the vector [W(1), W(2), …, W(L)] when the threshold ζ is ζ1, ζ2 (ζ1<ζ2) respectively. W(η) here represents the following:

[0103]

[0104] The fourth group: the body feature point amplitude distribution features include a first column number amplitude distribution F9 and a second column number amplitude distribution F 10 , and the specific extraction formula is as follows:

[0105]

[0106]

[0107] Wherein, N0 is the column number corresponding to the first non-zero column vector in the target feature map, represents the column number closest to the radar of the target object, n1, n2 represent the number of column vectors.

[0108] In particular, when determining the posture feature data, the maximum Doppler frequency of the body trunk can also be determined, which does not need to be determined according to the amplitude information and the shape information, but in order to describe all the data contained in the posture feature data, the maximum Doppler frequency of the body trunk is described here. The extraction formula corresponding to the maximum Doppler frequency of the body trunk is as follows:

[0109] F 11 = max(|f +max |,|f -max |) (16)

[0110] Wherein, f +max is the maximum value of the positive Doppler frequency in the target feature map, and f -max is the maximum value of the negative Doppler frequency in the target feature map.

[0111] S230, for each segment of the electromagnetic echo signal segment, based on the side lying feature data, determine whether the sleep posture corresponding to the electromagnetic echo signal segment is a side lying posture.

[0112] In this embodiment, to determine whether the sleep posture corresponding to the electromagnetic echo signal segment is a side lying posture, it can be determined by voting. That is, for each segment of the electromagnetic echo signal segment, it can be divided into multiple sub-data segments, and for each sub-data segment, it is respectively determined whether the sleep posture corresponding to the sub-data segment is a side lying posture or a non-side lying posture based on the side lying feature data and the pre-trained side lying classifier. Further, by comparing the number of side lying posture sub-data segments and the number of non-side lying posture sub-data segments, it is determined whether the sleep posture corresponding to the electromagnetic echo signal segment is a side lying posture.

[0113] S240, from at least one electromagnetic echo signal segment, the electromagnetic echo signal segment with a side lying posture is removed, and the pitch lying echo signal segment is obtained.

[0114] On the basis of S230, the electromagnetic echo signal segment which has been determined as a side lying posture category in the first electromagnetic echo signal is removed, and only the electromagnetic echo signal segment of the non-side lying posture category is reserved, which is used as the pitch lying echo signal segment.

[0115] S250, the pitch lying feature data corresponding to each pitch lying echo signal segment is sorted, and the pitch lying reference feature value and the supine reference feature value corresponding to the target object are determined.

[0116] In this embodiment, the way of sorting the pitch lying feature data is not specifically limited, as long as the pitch lying data feature corresponding to the sleep posture of the pitch lying posture and the supine data feature corresponding to the sleep posture of the supine posture can be determined. Further, the pitch lying reference feature value is determined based on the pitch lying data feature, and the supine reference feature value is determined based on the supine data feature.

[0117] S260, based on the second electromagnetic echo signal, a current target feature map is generated, and current amplitude information and current shape information in the current target feature map are extracted.

[0118] In this embodiment, the way of generating the current target feature map based on the second electromagnetic echo signal is the same as the way of generating the target feature map based on the first electromagnetic echo signal in the above embodiment, which will not be described here. After generating the current target feature map, the way of extracting the current amplitude information and the current shape information therefrom is the same as that of "extracting the amplitude information and the shape information in the target feature map" in step S220. The current column vector amplitude information H1(n), the current first shape feature information H2(n), the current second shape feature information H3(n) and the current row vector amplitude information H4(n) can be extracted.

[0119] S270. Based on the current amplitude information and the current shape information, determine the current pose feature data corresponding to the target object.

[0120] In this embodiment, the current pose feature data corresponding to the target object can be determined based on the specific feature data in the current target feature map and H1(n), H2(n), H3(n), and H4(η). The determined current pose feature data may include, but is not limited to: the amplitude M1 of the current second non-zero column feature point, the amplitude M2 ​​of the current third non-zero column feature point, the maximum amplitude M3 of the current second largest non-zero column feature point, the maximum amplitude M4 of the current third largest non-zero column feature point, the head size M5 of the current second non-zero column, the head size M6 of the current third non-zero column, the current first threshold micro-motion frequency distribution M7, the current second threshold micro-motion frequency distribution M8, the amplitude distribution of the current first column quantity M9, and the amplitude distribution of the current second column quantity M1. 10 and the current maximum Doppler frequency M of the body trunk 11 .

[0121] S280. Based on the current posture feature data, prone reference feature value, and supine reference feature value, determine the current sleeping posture category corresponding to the target object.

[0122] The technical solution provided in this invention, when determining the prone and supine reference feature values ​​corresponding to a target object, firstly generates a target feature map based on a first electromagnetic echo signal, then extracts amplitude and shape information from the target feature map, and thus determines the posture feature data of the electromagnetic echo signal segment based on the amplitude and shape information, thereby determining the prone and supine reference feature values ​​corresponding to the target object based on the posture feature data. In this invention, after obtaining the first electromagnetic echo signal, the corresponding target feature map is determined. The electromagnetic echo signal can be converted into an image. The sleeping posture features extracted from the image contain more information than those extracted directly from the pulse signal. Generally speaking, the more features that can distinguish sleeping postures, the higher the accuracy of sleep posture recognition, thereby improving the accuracy of sleep posture recognition.

[0123] Example 3

[0124] Figure 6 This is a schematic diagram of a sleep posture recognition device provided in Embodiment 3 of the present invention. The device can execute the sleep posture recognition method provided in this embodiment of the invention. The device includes: a reference signal acquisition module 310, a reference feature determination module 320, a feature data determination module 330, and a current sleeping posture determination module 340.

[0125] The reference signal acquisition module 310 is configured to acquire a first electromagnetic echo signal received by a radar device within a first preset time length; wherein a radiation range of the radar device comprises a sleep area of a target object;

[0126] The reference feature determination module 320 is configured to determine a prone reference feature value and a supine reference feature value corresponding to the target object based on the first electromagnetic echo signal;

[0127] The feature data determination module 330 is configured to, when it is detected that the target object has a posture change after the first preset time length, acquire a second electromagnetic echo signal received by the radar device within a second preset time length, and determine current posture feature data corresponding to the target object based on the second electromagnetic echo signal.

[0128] The current sleep posture determination module 340 is configured to determine a current sleep posture category corresponding to the target object based on the current posture feature data, the prone reference feature value and the supine reference feature value.

[0129] On the basis of the above technical solutions, the reference feature determination module 320 comprises:

[0130] The sleep posture reference feature determination submodule is configured to divide the first electromagnetic echo signal into at least one electromagnetic echo signal segment, and determine posture feature data of each electromagnetic echo signal segment; wherein the posture feature data comprises side-lying feature data and prone-supine feature data.

[0131] The side-lying posture determination submodule is configured to, for each electromagnetic echo signal segment, determine whether a sleep posture corresponding to the electromagnetic echo signal segment is a side-lying posture based on the side-lying feature data.

[0132] The prone-supine data segment acquisition submodule is configured to exclude electromagnetic echo signal segments with a side-lying posture from the at least one electromagnetic echo signal segment, and acquire prone-supine echo signal segments.

[0133] The reference feature value determination submodule is configured to perform feature sorting on the prone-supine feature data corresponding to each of the prone-supine echo signal segments, and determine a prone reference feature value and a supine reference feature value corresponding to the target object.

[0134] On the basis of the above technical solutions, the sleep posture reference feature determination submodule comprises:

[0135] The feature map determination unit is configured to, for each electromagnetic echo signal segment, generate a target feature map based on the electromagnetic echo signal.

[0136] The feature information acquisition unit is configured to extract amplitude information and shape information in the target feature map.

[0137] a feature data acquisition unit configured to determine posture feature data of the electromagnetic echo signal segment based on the amplitude information and the shape information.

[0138] In the above technical solutions, the feature information acquisition unit comprises:

[0139] an amplitude information acquisition subunit configured to determine column feature amplitudes corresponding to each column vector of the target feature map based on amplitude values corresponding to each element in each column vector, and determine column vector amplitude information of the target feature map based on the column feature amplitudes; and determine row feature amplitudes corresponding to each column vector based on amplitude values corresponding to each element in each row vector of the target feature map, and determine row vector amplitude information of the target feature map based on the row feature amplitudes.

[0140] a shape information acquisition subunit configured to determine shape feature values corresponding to each element in the target feature map based on the amplitude values corresponding to each element and a preset amplitude threshold value, determine a shape feature matrix composed of the shape feature values, perform norm operation on each column vector in the shape feature matrix to determine first shape feature information corresponding to each column, and determine second shape feature information corresponding to each column based on maximum and minimum shape feature values of each column in the shape feature matrix.

[0141] In the above technical solutions, the feature data determination module 330 comprises:

[0142] a current feature map determination sub-module configured to generate a current target feature map based on the second electromagnetic echo signal;

[0143] a current feature information determination sub-module configured to extract current amplitude information and current shape information in the current target feature map;

[0144] a current feature data determination sub-module configured to determine current posture feature data corresponding to the target object based on the current amplitude information and the current shape information.

[0145] In the above technical solutions, the current sleep posture determination module 340 comprises:

[0146] a side-lying category determination sub-module configured to determine whether a current sleep posture category corresponding to the target object is a side-lying posture category based on the current side-lying feature data;

[0147] a prone-supine posture determination sub-module configured to determine the current sleep posture category corresponding to the target object based on the current prone-supine feature data, a prone reference feature value and a supine reference feature value if the current sleep posture category corresponding to the target object is a non-side-lying posture category.

[0148] The technical scheme provided by the embodiment of the present application comprises the following steps: obtaining a first electromagnetic echo signal received by a radar device within a first preset time length, wherein a radiation range of the radar device comprises a sleep area of a target object; determining a prone reference characteristic value and a supine reference characteristic value corresponding to the target object based on the first electromagnetic echo signal, so as to obtain a personalized reference characteristic value of the target user; when it is detected that the target object has a posture change after the first preset time length, obtaining a second electromagnetic echo signal received by the radar device within a second preset time length, and determining current posture characteristic data corresponding to the target object based on the second electromagnetic echo signal; and then determining a current sleep posture category corresponding to the target object based on the current posture characteristic data, the prone reference characteristic value and the supine reference characteristic value. The present scheme solves the technical problem of low sleep posture recognition accuracy based on a wearable sensor device, does not require the user to wear various sensor devices, does not bring a foreign body sensation to the user, can realize contactless recognition of a sleep posture, improves the convenience of sleep posture recognition, and improves the sleep posture recognition accuracy by obtaining a personalized sleep posture characteristic reference value of the target object.

[0149] The sleep posture recognition device provided by the embodiments of the present disclosure can execute the sleep posture recognition method provided by any of the embodiments of the present disclosure, and has the corresponding function modules and beneficial effects of the execution method.

[0150] It should be noted that each unit and module included in the above device is only divided according to the function logic, but is not limited to the above division, as long as the corresponding function can be realized; in addition, the specific name of each functional unit is only for convenient mutual distinction, and does not limit the protection scope of the embodiments of the present disclosure.

[0151] Embodiment four

[0152] Figure 7 A structural schematic diagram of a device provided for the fourth embodiment of the present application. The device 10 is intended to represent various forms of digital computers, such as laptops, desktops, tablets, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The device can also represent various forms of mobile devices, such as personal digital processors, cellular telephones, smart phones, wearable devices (such as headsets, glasses, watches, etc.), and other similar computing devices. The components shown here, their connections, and relationships, and their functions, are merely examples and are not intended to limit the implementations of the present application described and / or claimed herein.

[0153] As Figure 7As shown, the device 10 includes at least one processor 11, and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., connected to the at least one processor 11 in communication. The memory stores computer programs executable by the at least one processor 11, which can perform various appropriate actions and processes according to the computer programs stored in the read-only memory (ROM) 12 or loaded into the random access memory (RAM) 13 from the storage unit 18. In the RAM 13, various programs and data required for the operation of the device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 13. An input / output (I / O) interface 15 is also connected to the bus 13.

[0154] Various components in the device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc., an output unit 17, such as various types of displays, speakers, etc., a storage unit 18, such as a magnetic disk, an optical disk, etc., and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the device 10 to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunication networks.

[0155] The processor 11 can be various general and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 performs various methods and processes described above, such as the sleep posture recognition method.

[0156] In some embodiments, the sleep posture recognition method can be implemented as a computer program tangibly embodied in a computer readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed onto the device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the sleep posture recognition method described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to perform the sleep posture recognition method by any other appropriate means, such as by means of firmware.

[0157] The various embodiments of the systems and techniques described above can be implemented in digital electronic circuitry, integrated circuitry, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on a chip (SOC), a load programmable logic device (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0158] Computer programs used to implement the processes of the application can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, special purpose computer, or other programmable processing device to produce a machine, such that the computer program, when executed, implements the functions / acts specified in the flowcharts and / or block diagrams. The computer program can be executed entirely on a machine, partially on a machine, partially on a machine and partially on a remote machine or entirely on a remote machine or server.

[0159] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store computer programs for use by or in connection with an instruction execution system, apparatus, or device. The computer-readable storage medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium will include one or more lines of a program of instructions in a transitory signal, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0160] To provide for interaction with a user, the systems and techniques described here can be implemented on a device having a display (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the device. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0161] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), blockchain network, and the Internet.

[0162] The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. A server can be a cloud server, also known as cloud computing server or cloud host, which is a host product in the cloud computing service system, to solve the defects of large management difficulty and weak business scalability in traditional physical host and VPS service. It should be understood that various forms of flow shown above can be used to reorder, add or delete steps. For example, each step described in the present application can be executed in parallel, sequentially or in different order, as long as the desired results of the technical solutions of the present application can be achieved, which is not limited herein. The above specific embodiments do not constitute a limitation on the scope of protection of the present application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent replacement and improvement within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A sleep posture recognition method, characterized in that, include: Acquire the first electromagnetic echo signal received by the radar device within a first preset time period; wherein the radiation range of the radar device includes the sleeping area of ​​the target object; Based on the first electromagnetic echo signal, determine the prone reference feature value and the supine reference feature value corresponding to the target object; When a change in posture of the target object is detected after a first preset time period, the second electromagnetic echo signal received by the radar device within a second preset time period is acquired. Generate the current target feature map based on the second electromagnetic echo signal; Extract the current amplitude information and current shape information from the current target feature map; Based on the current amplitude information and the current shape information, determine the current pose feature data corresponding to the target object; Based on the current posture feature data, prone reference feature value, and supine reference feature value, the current sleeping posture category corresponding to the target object is determined; The step of determining the prone and supine reference feature values ​​corresponding to the target object based on the first electromagnetic echo signal includes: The first electromagnetic echo signal is divided into at least one electromagnetic echo signal segment; For each of the electromagnetic echo signal segments, a target feature map is generated based on the electromagnetic echo signal; Extract the amplitude and shape information from the target feature map; Based on the amplitude and shape information, the posture feature data of the electromagnetic echo signal segment is determined; wherein, the posture feature data includes lateral recumbent feature data and supine recumbent feature data; For each segment of electromagnetic echo signal, based on the side-lying feature data, it is determined whether the sleeping posture corresponding to the electromagnetic echo signal segment is a side-lying posture. From the at least one electromagnetic echo signal segment, the electromagnetic echo signal segment with the sleeping posture being the side-lying position is removed to obtain the supine echo signal segment. Feature sorting is performed on the prone and supine feature data corresponding to each of the prone and supine echo signal segments to determine the prone reference feature value and supine reference feature value corresponding to the target object.

2. The sleep posture recognition method according to claim 1, characterized in that, The magnitude information includes: column vector magnitude information and row vector magnitude information. Extracting the magnitude information from the target feature map includes: Based on the magnitude value corresponding to each element in each column vector of the target feature map, the column feature magnitude corresponding to each column vector is determined, and the column vector magnitude information of the target feature map is determined based on the magnitude of each column feature. Based on the magnitude value corresponding to each element in each row vector of the target feature map, the row feature magnitude corresponding to each row vector is determined, and the row vector magnitude information of the target feature map is determined based on the row feature magnitude.

3. The sleep posture recognition method according to claim 1, characterized in that, The shape information includes: first shape feature information and second shape feature information, and the extraction of shape information from the target feature map includes: Based on the amplitude value corresponding to each element in the target feature map and the preset amplitude threshold, the shape feature value corresponding to each element is determined; Determine a shape feature matrix composed of the shape feature values, perform norm operation on each column vector in the shape feature matrix, and determine the first shape feature information corresponding to each column; Based on the maximum and minimum shape feature values ​​of each column in the shape feature matrix, the second shape feature information corresponding to each column is determined.

4. The sleep posture recognition method according to claim 1, characterized in that, The current posture feature data includes current lateral recumbent feature data and current prone recumbent feature data; The current sleeping position category includes: side-lying position category, supine position category, and prone position category; determining the current sleeping position category corresponding to the target object based on the current posture feature data, prone reference feature value, and supine reference feature value includes: Based on the current side-lying feature data, determine whether the current sleeping posture category corresponding to the target object is a side-lying posture category; If not, then based on the current prone and supine feature data, prone reference feature value, and supine reference feature value, the current sleeping posture category corresponding to the target object is determined.

5. A sleep posture recognition device, characterized in that, include: A reference signal acquisition module is used to acquire the first electromagnetic echo signal received by the radar device within a first preset time period; wherein, the radiation range of the radar device includes the sleep region of the target object; The reference feature determination module is used to determine the prone reference feature value and the supine reference feature value corresponding to the target object based on the first electromagnetic echo signal. The feature data determination module is used to, when a change in posture of a target object is detected after a first preset time period, acquire the second electromagnetic echo signal received by the radar device within a second preset time period, and generate a current target feature map based on the second electromagnetic echo signal; extract the current amplitude information and current shape information from the current target feature map; and determine the current posture feature data corresponding to the target object based on the current amplitude information and current shape information. The current sleeping posture determination module is used to determine the current sleeping posture category of the target object based on the current posture feature data, prone reference feature value and supine reference feature value; The reference feature determination module includes: The sleeping posture reference feature determination submodule is used to divide the first electromagnetic echo signal into at least one electromagnetic echo signal segment; for each electromagnetic echo signal segment, generate a target feature map based on the electromagnetic echo signal; extract amplitude information and shape information from the target feature map; and determine the posture feature data of the electromagnetic echo signal segment based on the amplitude information and shape information; wherein, the posture feature data includes lateral reclining feature data and supine reclining feature data; The side-lying posture discrimination submodule is used to determine, for each electromagnetic echo signal segment, whether the sleeping posture corresponding to the electromagnetic echo signal segment is a side-lying posture based on the side-lying feature data. The supine data segment acquisition submodule is used to remove electromagnetic echo signal segments with the sleeping posture being the side-lying posture from the at least one electromagnetic echo signal segment to obtain the supine echo signal segment. The reference feature value determination submodule is used to perform feature sorting on the prone and supine feature data corresponding to each of the prone and supine echo signal segments, and to determine the prone reference feature value and supine reference feature value corresponding to the target object.

6. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the sleep posture recognition method according to any one of claims 1-4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the sleep posture recognition method according to any one of claims 1-4.

Images