Remote health monitoring method, apparatus, device, and medium

Abstract

The application relates to the technical field of remote data interaction, and in particular to a remote health monitoring method, device, equipment and medium. The method system performs semantic recognition on a design draft, determines component requirement information, extracts design intention in the design draft, then performs quality evaluation on the extracted physiological signal by integrating a signal quality index and signal cycle stability, generates a comprehensive credibility score, and performs physiological signal transmission only when the frame quality score and the credibility score are simultaneously satisfied, so as to screen and exclude low-quality or untrustworthy data, improve the quality of physiological signals transmitted to a master control end, and the application can be applied to a remote health service scene in intelligent medical treatment and a health monitoring scene of a user in an insurance service in financial technology, so as to improve the transmission quality of physiological signals.

Description

Technical Field

[0001] This application relates to the field of remote data interaction technology, and in particular to a remote health monitoring method, device, equipment and medium. Background Technology

[0002] With the aging population and increasing demand for chronic disease management, remote health monitoring technology is becoming more important; however, current remote health monitoring technologies still have many problems.

[0003] Currently, most remote health monitoring methods primarily rely on photoplethysmography (PPG) analysis of specific areas in video footage to calculate physiological information such as heart rate, which is then uploaded to a server or control unit. However, this approach receives a large amount of invalid or low-quality data due to poor acquisition conditions, increasing the burden on network transmission and affecting the overall accuracy of remote health monitoring, potentially generating misleading health alerts. For example, in smart healthcare and elderly care scenarios, when family doctors analyze heart rate using PPG on a person's face via a remote camera, insufficient lighting or user movement causing video blurring can lead to erroneous data being uploaded to the cloud, increasing network traffic and potentially triggering false heart rate alarms, thus affecting diagnostic accuracy. Similarly, in financial insurance services, remote health monitoring can collect user information to match them with optimal insurance products; however, the signal quality of remote health monitoring may be abnormal, affecting the accuracy of the service.

[0004] Therefore, how to comprehensively evaluate the signal quality of physiological signals and automatically select physiological signals that meet the quality standards to complete signal transmission has become an urgent problem to be solved. Summary of the Invention

[0005] This invention provides a remote health monitoring method, device, computer equipment, and medium to solve the technical problem of comprehensively evaluating the signal quality of physiological signals, thereby automatically screening physiological signals that meet quality standards to complete signal transmission.

[0006] Firstly, a remote health monitoring method is provided, wherein the remote health monitoring method is applied to each source end in a remote health monitoring system, the remote health monitoring system further includes a master control end, and all source ends are used to connect to the master control end, including: Acquire continuous video frame data; For any frame image in the video frame data, the region containing a face in the frame image is extracted to obtain the region of interest image corresponding to the frame image; Traverse all frame images to obtain the region of interest image corresponding to each frame image. Sort all region of interest images according to the temporal order of each frame image in the video frame data to obtain the sequence of images of interest. Based on the image sequence of interest, determine the physiological signal characterizing the physiological characteristics, the signal quality index corresponding to the physiological signal, the signal periodic stability, and the frame quality score characterizing the image quality in the image sequence of interest; Based on the signal quality index and signal periodic stability, a reliability score characterizing the reliability of the signal is determined; When the frame quality score is greater than a first preset score and the credibility score is greater than a second preset score, the physiological signal is sent to the main control terminal.

[0007] Secondly, this application provides a remote health monitoring device, which is applied to each source end of a remote health monitoring system. The remote health monitoring system also includes a master control end, and all source ends are used to connect to the master control end, including: The data acquisition module is used to acquire continuous video frame data; The region extraction module is used to extract the region containing a face in any frame image of the video frame data to obtain the region of interest image corresponding to the frame image. The sorting module is used to traverse all frame images, obtain the region of interest image corresponding to each frame image, and sort all region of interest images according to the temporal order of each frame image in the video frame data to obtain the sequence of images of interest. The quality analysis module is used to determine, based on the image sequence of interest, physiological signals characterizing physiological characteristics, signal quality index corresponding to the physiological signals, signal periodic stability, and frame quality scores characterizing the image quality in the image sequence of interest. The credibility scoring module is used to determine a credibility score that characterizes the credibility of the signal based on the signal quality index and the signal periodic stability. The signal transmission module is used to send the physiological signal to the main control terminal when the frame quality score is greater than a first preset score and the credibility score is greater than a second preset score.

[0008] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the aforementioned remote health monitoring method.

[0009] Fourthly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the steps of the aforementioned remote health monitoring method.

[0010] The beneficial effects of this application embodiment compared with the prior art are as follows: Continuous video frame data is collected; for any frame image in the video frame data, the region containing a face in the frame image is extracted to obtain the region of interest image of the corresponding frame image; all frame images are traversed to obtain the region of interest image corresponding to each frame image; according to the time sequence of each frame image in the video frame data, all region of interest images are sorted to obtain a sequence of images of interest; based on the sequence of images of interest, physiological signals representing physiological characteristics, the signal quality index of the corresponding physiological signals, the signal periodicity, and the frame quality score representing the image quality in the sequence of images of interest are determined; based on the signal quality index and the signal periodicity, a reliability score representing the reliability of the signal is determined; when the frame quality score is greater than a first preset score and the reliability score is greater than a second preset score, a physiological signal is sent to the main control terminal. This application can be applied to remote health business scenarios in smart healthcare and health monitoring scenarios for users in insurance and other businesses in fintech. By integrating the signal quality index and signal periodic stability, the extracted physiological signals are evaluated for quality and a comprehensive credibility score is generated. Physiological signal transmission is only allowed when both the frame quality score and the credibility score are met, thereby filtering out low-quality or unreliable data and improving the quality of physiological signals transmitted to the main control terminal. Attached Figure Description

[0011] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 This is a schematic diagram of an application environment for a remote health monitoring method provided in an embodiment of this application; Figure 2 This is a schematic flowchart of a remote health monitoring method provided in an embodiment of this application; Figure 3 This is a schematic flowchart of a remote health monitoring method provided in an embodiment of this application; Figure 4 This is a schematic flowchart of a remote health monitoring method provided in an embodiment of this application; Figure 5 This is a schematic flowchart of a remote health monitoring method provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of a remote health monitoring device provided in one embodiment of this application; Figure 7This is a schematic diagram of the structure of a computer device according to an embodiment of the present invention; Figure 8 This is another structural schematic diagram of a computer device according to one embodiment of the present invention. Detailed Implementation

[0013] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] The remote health monitoring method provided in Embodiment 1 of this application can be applied to, for example... Figure 1 In this application environment, specifically in a remote health monitoring system, the client acts as the user's interactive interface, while the server serves as the control center. Users can set monitoring conditions, submit requests, and send control commands through the client. The server receives these commands and, based on a pre-defined remote health monitoring method, issues specific execution instructions to each source endpoint. All source endpoints are connected to the control center, collaboratively implementing the remote health monitoring function.

[0015] See Figure 2 This is a flowchart illustrating a remote health monitoring method provided in an embodiment of this application. The aforementioned remote health monitoring method can be applied to... Figure 1 The server-side component.

[0016] like Figure 2 As shown, this remote health monitoring method may include the following steps: Step S201: Collect continuous video frame data.

[0017] Step S202: For any frame image in the video frame data, extract the region containing a face in the frame image to obtain the region of interest image corresponding to the frame image.

[0018] Step S203: Traverse all frame images to obtain the region of interest image corresponding to each frame image. Sort all region of interest images according to the temporal order of each frame image in the video frame data to obtain the sequence of images of interest.

[0019] The camera or a mobile terminal with an integrated camera (such as a smartphone, tablet, or laptop) is placed in a suitable position in front of the user. Preferably, the distance between the camera and the user's face should be between 0.5 meters and 1.5 meters to ensure that the facial area has sufficient pixel resolution in the image.

[0020] When the health monitoring application or system is launched, the camera is triggered to start working. The system sends initialization instructions to the camera hardware to configure the acquisition parameters, including but not limited to resolution, frame rate and exposure mode. The camera's optical sensor converts the received light signals into continuous electronic signals.

[0021] The camera's built-in image signal processor performs preprocessing on the electronic signal, including de-mosaicing, white balance, and color correction, to generate raw image data in a standard format. The system driver or application programming interface encapsulates this raw image data into a series of video frames in chronological order and writes them to a designated memory buffer or video stream queue. The acquisition process continues, forming a data set containing multiple frames, which is the video frame data.

[0022] The system calls a pre-trained face detection model (such as a deep learning model) to scan and analyze any frame in the video frame data. The face detection model outputs one or more face bounding boxes. Each bounding box is typically defined by the coordinates of the top-left corner of a rectangular region and its width and height, accurately outlining the detected faces in the image. Based on the face bounding box coordinates, the system crops the corresponding rectangular region from the original frame image. To improve the efficiency and consistency of subsequent processing, the system can scale this cropped region to a uniform fixed size for normalization. A separate image file or memory data block containing only the cropped face region is generated for this frame image; this is the region of interest image corresponding to that frame image.

[0023] The system processes all frame images in the video frame data sequentially or in parallel, repeating step S202 for each frame to ensure that each original frame image generates a corresponding region of interest image.

[0024] During processing, the system maintains a relationship between each region of interest image and its source frame's timestamp or frame number.

[0025] After processing all frames, the system sorts all region-of-interest (ROI) images according to the timestamps or frame numbers associated with each RIO image, in chronological order.

[0026] The sorted list of regions of interest (ROI) images is stored as an ordered data structure (such as an array, linked list, or queue), which is defined as the sequence of ROI images. The sequence of ROI images reflects the continuous flow of visual information about the facial regions over time.

[0027] For example, in smart healthcare scenarios, cameras can capture continuous video frames of hospitalized patients, extract facial regions from each frame in real time, and generate facial image sequences in chronological order. This allows for the automatic analysis of changes in the patient's pain expressions (such as the duration and intensity of micro-expressions), providing medical staff with objective pain assessment data and assisting in the precise dispensing of analgesic drugs.

[0028] For example, in financial insurance services, authorized and trusted cameras can collect images of a user's head with the user's knowledge, thereby obtaining facial images, which can be analyzed to obtain relevant physiological signals.

[0029] Step S204: Based on the image sequence of interest, determine the physiological signal representing the physiological characteristics, the signal quality index corresponding to the physiological signal, the signal periodic stability, and the frame quality score representing the image quality in the image sequence of interest.

[0030] Optionally, determining the physiological signal characterizing the physiological feature may include the following steps: Obtain a preset analysis period and determine the number of image frames in the video frame data based on the analysis period; Based on the analysis time period and the number of image frames, target images are extracted sequentially from the sequence of images of interest to obtain the image set to be analyzed. Based on the analysis period, pixel value analysis is performed on all target images in the image set to be analyzed to obtain a pixel change sequence; Based on the pixel change sequence, physiological signals characterizing physiological features are determined.

[0031] The system first acquires a preset analysis period. This analysis period is the basic time unit for calculating a single physiological signal. Based on this analysis period and the video acquisition frame rate known in step S201, the number of image frames corresponding to this period is calculated.

[0032] From the entire sequence of images of interest obtained in step S203, regions of interest (ROIs) of consecutive frames are extracted in chronological order. These consecutive frame images are called target images, and their set constitutes the image set to be analyzed. Extraction typically begins at the start of the sequence or the end of the previous analysis window to achieve sliding window analysis of the video stream.

[0033] The system iterates through each frame of the target image in the image set to be analyzed. For each frame, it calculates one or more predefined pixel statistics. The most common method is to calculate the average value of the entire region of interest image across a specific color channel. For example, converting the image from a color space to a color space more sensitive to changes in blood volume, such as the green channel or chromaticity space.

[0034] Calculate the average brightness of all pixels in a selected channel (e.g., the green channel) for any given frame. Process all target images chronologically, arranging the average pixel values ​​in chronological order of acquisition time to form a one-dimensional pixel variation sequence. This sequence reflects the subtle periodic fluctuations in the apparent color of facial skin caused by pulsed changes in subcutaneous blood volume. Perform signal processing on the obtained raw pixel variation sequence to enhance periodic physiological information and suppress noise. Processing includes detrending and bandpass filtering.

[0035] The processed signal constitutes a physiological signal that characterizes physiological features. This physiological signal is a time-domain waveform whose periodicity corresponds to vital signs such as heart rate or respiratory rate.

[0036] While or after performing the above physiological signal extraction, the system calculates the signal quality index, signal periodic stability, frame quality score, etc. in parallel.

[0037] Signal quality index refers to a comprehensive assessment of the clarity and reliability of the generated physiological signal itself. Signal periodicity stability refers to the assessment of the temporal regularity of physiological signals (such as heartbeat intervals). Frame quality score refers to the assessment of the quality of each frame of the region of interest in the image set to be analyzed (such as clarity, whether it is blurry, whether there is occlusion, etc.).

[0038] For example, in medical service scenarios, by continuously capturing videos of infants through cameras, extracting facial regions from each frame to generate time-series image sequences, and then analyzing the sequences to obtain physiological signals such as heart rate and respiratory waves, and simultaneously calculating signal quality index, period stability and image frame quality score, it is possible to monitor the stability of the infant's vital signs in real time without any contact with the infant's skin, and automatically trigger an alarm to remind medical staff to re-examine when the signal quality is insufficient.

[0039] Optionally, step S204, the method for determining the signal quality index corresponding to the physiological signal, may include the following steps: Extract the brightness data of any consecutive target image in the image set to be analyzed from the video frame data; Based on the brightness data, the average brightness of the target image in any consecutive frame is calculated; Based on the preset region of interest, the region integrity rate of the image set to be analyzed is calculated. Based on the pixel change sequence, the standard deviation of the pixel change peak value is calculated to obtain the target image stability; The signal quality index corresponding to the physiological signal is calculated based on the preset weighting coefficients combined with the average brightness, the region integrity rate, and the target image stability.

[0040] Optionally, the method for determining the periodic stability of the signal may include the following steps: Based on the video frame data, pixel value analysis is performed on all the frame images to obtain a set of pixel changes; Based on the set of pixel changes, the standard deviation of the peak value of the pixel changes is calculated to obtain the signal periodic stability.

[0041] The signal quality index quantifies the reliability of physiological signals by evaluating the reliability of the acquisition environment and the signal itself from multiple dimensions. The system extracts the luminance component of each target image frame from the original video frame data corresponding to the image set to be analyzed. The average luminance of all target images in the entire image set to be analyzed is calculated to obtain the average luminance. This index reflects the overall illumination level of the acquisition environment. Values ​​that are too low or too high are detrimental to the accurate extraction of physiological signals.

[0042] The preset region of interest (ROI) is defined as the standard face bounding box region output in step S202 of the face detection step. For each frame of the target image in the image set to be analyzed, the system detects the actual valid face region pixels. For example, by using a skin color model or facial feature point detection, it determines what percentage of pixels within the preset bounding box actually belong to the face skin region, rather than being occluded (such as hands, hair, glasses) or belonging to the background. The region integrity rate is equal to the number of valid face pixels plus the total number of pixels in the preset ROI. This metric measures the degree to which the target region is presented completely and without occlusion during the analysis period.

[0043] The calculation of target image stability involves first using the pixel change sequence obtained in step S204 to perform peak detection on the sequence and identify the main peaks. Then, the standard deviation of these pixel change peaks is calculated. The smaller the standard deviation, the more regular and stable the signal amplitude fluctuations, and the less susceptible to interference from motion artifacts. This standard deviation is defined as a measure of target image stability.

[0044] The system pre-defines weighting coefficients for each dimension of the indicators, such as average brightness weight, region integrity weight, and target image stability weight. The three indicators are normalized to map their values ​​to the interval between 0 and 1, where 1 represents the optimal range. The normalization function can be set empirically (e.g., for brightness, an optimal range is set, and scores decrease if the range is exceeded).

[0045] Signal Quality Index (SQI). The formula is: SQI = (WL×NL) + (WR×NR) + (WS×NS) Wherein, NL is the average brightness score, NR is the region integrity rate, NS is the target image stability, WL is the average brightness weight, WR is the region integrity rate weight, and WS is the target image stability weight.

[0046] The final SQI value is also a value between 0 and 1. The higher the value, the more reliable the data quality that generated the physiological signal.

[0047] Signal periodicity stability specifically assesses the temporal consistency of the periodicity of physiological signals and is a core indicator of the inherent regularity of signals. First, based on complete video frame data (or its corresponding sequence of all images of interest), pixel value analysis is performed on all frame images in a manner similar to that in step S204, resulting in a pixel variation set covering a longer time range. This pixel variation set contains more comprehensive periodic information.

[0048] The standard deviation of pixel variation peaks is calculated to obtain the signal periodic stability. A peak detection algorithm is applied to the aforementioned pixel variation set to identify all significant peak positions. The time interval between adjacent peaks is calculated, resulting in a series of interpeak intervals. The standard deviation of this series of peaks is then calculated. This standard deviation directly quantifies the magnitude of fluctuations in the physiological signal period. Signal periodic stability is defined as the reciprocal of this standard deviation; a higher reciprocal value indicates a more stable and regular periodicity of the signal.

[0049] For example, in smart healthcare scenarios, by capturing facial videos of patients through home cameras, extracting time-series facial image sequences, and analyzing physiological signals such as heart rate variability, the system will comprehensively calculate the average brightness of the image, the integrity rate of the facial region, and the stability of pixel changes to generate a signal quality index. At the same time, by analyzing the fluctuation of pixel change peaks, the system will assess the stability of the signal cycle, thereby providing doctors with physiological reports with validity annotations, accurately identifying signal distortion segments caused by patient movement and changes in lighting, and ensuring the clinical credibility of home rehabilitation data.

[0050] Step S205: Determine the credibility score characterizing the credibility of the signal based on the signal quality index and the signal periodic stability.

[0051] Step S206: When the frame quality score is greater than the first preset score and the credibility score is greater than the second preset score, the physiological signal is sent to the main control terminal.

[0052] Specifically, the signal quality index and signal periodicity calculated in step S204 are received. These two indicators characterize signal quality from two dimensions: the acquisition environment and signal sharpness, and the inherent periodicity of the signal, respectively.

[0053] The system pre-sets or adaptively acquires normalization functions or scoring maps for signal quality index and signal periodic stability, ensuring that the value range of both is uniformly converted to a comparable scoring range (e.g., both mapped to 0 to 100 points).

[0054] Define a linear weighted sum formula to calculate the credibility score, for example, the linear weighted formula, C = α * NSQI + β * NSPS; Where C is the reliability score; NSQI is the signal quality index score; and NSPS is the signal period stability score. α is the preset signal quality weight, and β is the preset signal period weight, satisfying α + β = 1. The signal quality weight and signal period weight can be adjusted according to the application scenario. For example, in resting heart rate monitoring, period stability may be more important (larger β), while in motion monitoring, signal quality may be more important to resist motion artifacts (larger α).

[0055] A higher credibility score indicates that the physiological signals extracted during the current analysis period are more reliable.

[0056] The system receives frame quality scores from step S204 and reliability scores from step S205. Two quality thresholds are preset: a first preset score serves as the qualification threshold for the frame quality score, ensuring that the original video frames used for analysis are clear and usable; and a second preset score serves as the qualification threshold for the reliability score, ensuring that the physiological signals extracted from these frames are generally reliable.

[0057] If the frame quality score exceeds the first preset score and the reliability score exceeds the second preset score, the system determines that the quality of the physiological signals extracted during the current analysis period is qualified.

[0058] For example, in smart healthcare scenarios, the system captures facial videos of users through mobile phone cameras, extracts time-series facial image sequences, and analyzes physiological signals (such as pulse waves). It then comprehensively evaluates the signal quality index and period stability to generate a signal reliability score. Finally, only when both image frame quality and signal reliability meet the standards will the system upload reliable pulse waveform data to the hospital's cloud control terminal. This automatically excludes invalid measurements caused by user movement or insufficient lighting, ensuring the validity of the uploaded data and assisting doctors in conducting batch remote initial diagnoses.

[0059] In this embodiment, the extracted physiological signal is evaluated by integrating the signal quality index and the signal periodicity, and a comprehensive credibility score is generated. Physiological signal transmission is only carried out when both the frame quality score and the credibility score are satisfied, thereby screening out and excluding low-quality or unreliable data and improving the quality of physiological signals transmitted to the main control terminal.

[0060] See Figure 3 This is a flowchart illustrating a remote health monitoring method provided in an embodiment of this application. Figure 3 As shown, the method for determining the frame quality score characterizing the image quality in the sequence of images of interest in step S204 above may include the following steps: Step S301: Calculate the proportion of the region of interest for the frame image based on the region of interest, and traverse all the frame images to calculate the average proportion of the region of interest.

[0061] Step S302: According to the preset gradient algorithm, perform sharpness analysis on the frame image to obtain the image sharpness.

[0062] Step S303: Obtain image position information of any of the frame images, and determine position change information of any consecutive frames based on the image position information.

[0063] Step S304: Determine the image stability based on the position change information.

[0064] Step S305: Determine a frame quality score that characterizes the image quality in the image sequence of interest based on the average proportion of the region of interest, the image sharpness, and the image stability.

[0065] In this process, the system receives the region of interest (e.g., a face detection bounding box) defined in step S202, along with the processing results for each frame of the image. This region defines the key area in the image used for physiological signal extraction. For each frame of the image, the system calculates the number of valid pixels within this region of interest (e.g., pixels in the skin region confirmed by skin color segmentation and facial feature point recognition). This process excludes occluded or non-target parts.

[0066] For a single frame image, the region of interest (ROI) percentage is equal to the ratio of the number of valid pixels within the ROI to the total number of pixels in the ROI range. By iterating through all frames in the sequence of images of interest (e.g., an analysis period of 300 frames), the ROI percentage for each frame is calculated. Based on the total number of frames and the ROI percentage for each frame, the average ROI percentage is calculated.

[0067] For each frame of the image, especially its region of interest, the system applies preset gradient operators to perform sharpness analysis, including the Laplacian operator and gradient functions. For example, the Laplacian operator is used to calculate the edge intensity of the image; sharper images have stronger edge responses.

[0068] Calculate the total or mean response of the gradient operator within the range of interest.

[0069] For the entire sequence of images of interest, the sharpness values ​​of all frames are averaged to obtain the image sharpness of the sequence. This image sharpness quantifies the richness of detail in the image.

[0070] Extracting features that characterize the target's position in the image from image location information is a preferred implementation method, using the coordinates of facial feature points (such as the corners of the eyes, the tip of the nose, and the corners of the mouth) obtained in step S202. For any frame of image, the system acquires the coordinates of its preset set of feature points, calculates the positional change information of consecutive frames, and for any set of consecutive frames in the image sequence of interest (such as frame t and frame t+1), calculates the average change in feature point coordinates or the displacement of a specific reference point (such as the center point of the face region). By statistically analyzing this displacement between all consecutive frames, a positional change sequence is formed to describe the target's motion trajectory and speed changes in the image throughout the entire analysis period. Statistical analysis is then performed on the positional change sequence. The standard deviation of this positional change sequence can be calculated. The smaller the standard deviation of the positional change sequence, the smoother and more regular the target's movement; the larger the standard deviation of the positional change sequence, the more sudden and violent shaking or jittering occurs.

[0071] Image stabilization is calculated based on the standard deviation of the position change sequence. For example, stability can be assessed by calculating the peak acceleration of position changes between consecutive frames. Image stabilization is crucial for the stable tracking of a target and the extraction of clean physiological signals.

[0072] The system uses a pre-defined normalization function or mapping table to uniformly convert the values ​​of the three indicators—average percentage, image clarity, and image stability—to the same scoring range (e.g., between 0 and 1, with 1 being the optimal value).

[0073] Set weighting coefficients A (average percentage weight), B (image sharpness weight), and C (image stability weight) to satisfy A + B + C = 1. The weights can be adjusted according to the application scenario; for example, in non-contact measurement, the weight of stability may be higher.

[0074] The formula for calculating the frame quality score is as follows: FQS = (A * NA) + (B * NB) + (C* NC) Among them, FQS is the frame quality score; NA is the average percentage; NB is the image sharpness; and NC is the image stability.

[0075] This application determines a frame quality score that characterizes the image quality in the image sequence of interest based on the average proportion of the region of interest, image sharpness, and image stability, thereby quickly filtering and evaluating image frames in the image sequence of interest.

[0076] See Figure 4 This is a flowchart illustrating a remote health monitoring method provided in an embodiment of this application. Figure 4 As shown, before sending the physiological signal to the main control terminal in step S206, the following steps may also be included: Step S401: When sending a signal to the master control terminal, detect the type of the signal to be sent.

[0077] Step S402: If the signal type is not the type corresponding to the physiological signal, the transmission is blocked; if the signal type is the type corresponding to the physiological signal, the physiological signal is output at an output frequency not exceeding a preset frequency, where the preset frequency is less than the sampling frequency of the source end.

[0078] Specifically, after the system passes the judgment condition in step S206 (both frame quality score and credibility score meet the standard), before preparing to send data to the main control terminal, the signal type detection process is triggered.

[0079] The system performs content analysis on the data packets to be packaged and sent, checking the protocol header or preset data identifiers of the data packets. For example, the system defines different data packet header formats or specific flag bits for different types of sensor data (such as physiological signals, device status, and log information).

[0080] Another approach is to analyze the internal structure of the data. For example, physiological signal data often contains regular time-series waveform data or heart / respiratory rate values, rather than structured text or images. The system can infer the signal type through simple pattern recognition (such as checking whether the data is a numerical array of a preset length or whether it conforms to the amplitude range of physiological signals).

[0081] Based on the above analysis, the system determines the signal type of the signal to be transmitted. This type includes at least two major categories: physiological signals and non-physiological signals.

[0082] If the signal type does not correspond to a physiological signal, the system determines it to be an abnormal situation. For example, software errors, malicious attacks, or hardware failures may cause the data packet to be sent to be corrupted or tampered with. In this case, the system will block transmission, i.e., immediately cancel the transmission of the data packet, and optionally generate an error log or alarm message. This serves as a security filter, preventing irrelevant or harmful data from being sent to the main control terminal. If the signal type corresponds to a physiological signal, the system confirms the validity of the data to be sent and continues to execute subsequent output control procedures.

[0083] The preset frequency represents the maximum rate at which data can be sent to the network or the master control unit.

[0084] The preset frequency is set to be lower than the sampling frequency of the source device. For example, a camera (the source device) might capture video at a sampling rate of 30 frames per second, but the heart rate signal extracted from it might have its output frequency limited to once per second (1Hz). The main purpose of this is to avoid high-frequency data streams consuming too many network resources, especially in wireless or bandwidth-constrained environments. Secondly, the host device (such as a cloud server) may process data from multiple devices simultaneously, and limiting the output frequency helps balance the load.

[0085] The system uses a rate limiter (such as a token bucket or leaky bucket algorithm) or a simple timed transmission mechanism to ensure that the physiological signal output frequency does not exceed the preset frequency. After frequency limiting control, the system formally sends the qualified physiological signal data packets to the master control terminal through the communication interface.

[0086] In this embodiment of the application, by adding a type security check and output rate control before data transmission in the remote physiological monitoring system, the accuracy of transmitted data and system stability are ensured.

[0087] See Figure 5 This is a flowchart illustrating a remote health monitoring method provided in an embodiment of this application. Figure 5 As shown, step S202, which involves extracting the region containing a face from the frame image to obtain the region of interest image corresponding to the frame image, may include the following steps: Step S501: According to the preset detection algorithm, the regions containing human faces in the frame image are marked to obtain the face location representation.

[0088] Step S502: Based on the face position representation, blur the areas in the frame image that do not contain a face to obtain a blurred image.

[0089] Step S503: Based on the face position representation, the blurred image is removed to obtain the region of interest image corresponding to the frame image.

[0090] For each frame of received images, the system calls a preset face detection algorithm. This face detection algorithm can use a deep learning-based convolutional neural network model or a traditional feature classifier to scan and analyze the image, identify and locate all regions containing faces.

[0091] For each detected face region, the algorithm outputs a face location representation, which is a set of bounding box coordinates.

[0092] In a preferred embodiment, the face location representation may also include more refined facial feature point coordinates, such as detecting the positions of key points like the eyes, the tip of the nose, and the corners of the mouth.

[0093] Based on the face location representation (such as the face bounding box) obtained in step S501, the system divides the image into face regions and non-face regions. The face region is the area inside the image defined by the bounding box or feature points. For non-face regions (i.e., all pixels in the image other than the face region), the system applies an image processing algorithm to blur them.

[0094] A typical implementation involves applying a Gaussian blur filter. The system calculates a new pixel value based on a Gaussian function for each pixel in the non-face region. This value is determined by a weighted average of its surrounding neighboring pixels, resulting in a smooth, detail-loss-laden blur effect. The degree of blurring can be adjusted using preset parameters (such as the size of the Gaussian kernel and its standard deviation) to ensure that visual information in the non-face regions (such as the background environment, other people's faces, and recognizable text) becomes difficult to discern after processing. After processing, the face regions in the original image remain unchanged, while the non-face regions are blurred, generating a blurred image.

[0095] The system determines the precise face bounding box in the original image based on the same face location representation. Subsequently, the system directly extracts the rectangular region defined by the bounding box from the blurred image obtained in step S502. This operation is equivalent to cropping the face region from the blurred image.

[0096] After cropping and extraction, the system obtains a new, independent image containing only the target face, with its background information blurred or removed. This independent image is the region of interest image corresponding to the current frame.

[0097] In this embodiment, face detection and labeling technology is used, combined with background blurring and region preservation algorithms, to achieve privacy protection and region of interest extraction of faces in images. Faces are automatically recognized and occluded, effectively protecting personal privacy, while background areas without faces are extracted to maintain the validity of the rest of the image.

[0098] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0099] In one embodiment, a remote health monitoring device is provided, which corresponds one-to-one with the remote health monitoring methods described in the above embodiments. For example... Figure 6 As shown, the remote health monitoring device includes a data acquisition module 61, a region extraction module 62, a sorting module 63, a quality analysis module 64, a reliability scoring module 65, and a signal transmission module 66. Detailed descriptions of each functional module are as follows: Data acquisition module 61 is used to acquire continuous video frame data; The region extraction module 62 is used to extract the region containing a face in any frame image of the video frame data to obtain the region of interest image corresponding to the frame image. The sorting module 63 is used to traverse all frame images, obtain the region of interest image corresponding to each frame image, and sort all region of interest images according to the temporal order of each frame image in the video frame data to obtain a sequence of images of interest. The quality analysis module 64 is used to determine, based on the image sequence of interest, physiological signals characterizing physiological characteristics, signal quality index corresponding to the physiological signals, signal periodic stability, and frame quality scores characterizing the image quality in the image sequence of interest. The credibility scoring module 65 is used to determine a credibility score that characterizes the credibility of the signal based on the signal quality index and the signal periodic stability. The signal transmission module 66 is used to send the physiological signal to the main control terminal when the frame quality score is greater than a first preset score and the credibility score is greater than a second preset score.

[0100] Optionally, the quality analysis module 64 includes: The quantity determination unit is used to obtain a preset analysis time period and determine the number of image frames in the video frame data according to the analysis time period. An image extraction unit is used to extract target images from the sequence of images of interest sequentially according to the analysis time period and the number of image frames, to obtain a set of images to be analyzed; The pixel analysis unit is used to perform pixel value analysis on all target images in the image set to be analyzed according to the analysis time period to obtain a pixel change sequence; The pixel sequence analysis unit is used to determine physiological signals characterizing physiological features based on the pixel change sequence.

[0101] Optionally, the quality analysis module 64 further includes: The data extraction unit is used to extract the brightness data of any consecutive target image in the set of images to be analyzed from the video frame data; An average brightness calculation unit is used to calculate the average brightness of the target image in any consecutive frame based on the brightness data. The integrity rate determination unit is used to calculate the region integrity rate of the image set to be analyzed based on a preset region of interest. The target stability calculation unit is used to calculate the standard deviation of the pixel change peak value based on the pixel change sequence to obtain the target image stability. The index determination unit is used to calculate the signal quality index corresponding to the physiological signal based on preset weighting coefficients combined with the average brightness, the region integrity rate, and the target image stability.

[0102] Optionally, the quality analysis module 64 further includes: The pixel change analysis unit is used to perform pixel value analysis on all the frame images based on the video frame data to obtain a pixel change set; The pixel standard deviation calculation unit is used to calculate the standard deviation of the pixel change peak value based on the pixel change set, and obtain the signal periodic stability.

[0103] Optionally, the quality analysis module 64 further includes: The average percentage calculation unit is used to calculate the percentage of the region of interest for the frame image based on the region of interest, and to traverse all the frame images to calculate the average percentage of the region of interest. A sharpness analysis unit is used to perform sharpness analysis on the frame image according to a preset gradient algorithm to obtain the image sharpness. The position change analysis unit is used to acquire image position information of any frame image and determine position change information of any consecutive frames based on the image position information. A change information analysis unit is used to determine the image stability based on the position change information; The frame quality scoring unit is used to determine a frame quality score that characterizes the image quality in the sequence of images of interest based on the average proportion of the region of interest, the image sharpness, and the image stability.

[0104] Optionally, the remote health monitoring device further includes: The signal type determination module is used to detect the type of signal to be sent when sending the signal to the main control terminal before sending the physiological signal to the main control terminal; The frequency output unit is used to block transmission if the signal type is not the type corresponding to the physiological signal, and to output the physiological signal at an output frequency not exceeding a preset frequency if the signal type is the type corresponding to the physiological signal, wherein the preset frequency is less than the sampling frequency of the source end.

[0105] Optionally, the region extraction module 62 includes: The marking unit is used to mark the regions containing human faces in the frame image according to a preset detection algorithm to obtain a representation of the human face location; The blurring unit is used to blur the areas of the frame image that do not contain a face according to the face position representation, so as to obtain a blurred image. The image removal unit is used to remove the blurred image based on the face position representation to obtain the region of interest image corresponding to the frame image.

[0106] This invention provides a remote health monitoring device that integrates signal quality index and signal periodicity to assess the quality of extracted physiological signals and generates a comprehensive reliability score. Physiological signal transmission is only performed when both the frame quality score and the reliability score are met, thereby filtering out low-quality or unreliable data and improving the quality of physiological signals transmitted to the main control terminal.

[0107] Specific limitations regarding remote health monitoring devices can be found in the limitations of remote health monitoring methods described above, and will not be repeated here. Each module in the aforementioned remote health monitoring device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0108] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 7 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with external clients via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a remote health monitoring method on the server side.

[0109] In one embodiment, a computer device is provided, which may be a client, and its internal structure diagram may be as follows: Figure 8 As shown, the computer device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with an external server via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a remote health monitoring method on the client side. In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps: Acquire continuous video frame data; For any frame image in the video frame data, the region containing a face in the frame image is extracted to obtain the region of interest image corresponding to the frame image; Traverse all frame images to obtain the region of interest image corresponding to each frame image. Sort all region of interest images according to the temporal order of each frame image in the video frame data to obtain the sequence of images of interest. Based on the image sequence of interest, determine the physiological signal characterizing the physiological characteristics, the signal quality index corresponding to the physiological signal, the signal periodic stability, and the frame quality score characterizing the image quality in the image sequence of interest; Based on the signal quality index and signal periodic stability, a reliability score characterizing the reliability of the signal is determined; When the frame quality score is greater than a first preset score and the credibility score is greater than a second preset score, the physiological signal is sent to the main control terminal.

[0110] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor: Acquire continuous video frame data; For any frame image in the video frame data, the region containing a face in the frame image is extracted to obtain the region of interest image corresponding to the frame image; Traverse all frame images to obtain the region of interest image corresponding to each frame image. Sort all region of interest images according to the temporal order of each frame image in the video frame data to obtain the sequence of images of interest. Based on the image sequence of interest, determine the physiological signal characterizing the physiological characteristics, the signal quality index corresponding to the physiological signal, the signal periodic stability, and the frame quality score characterizing the image quality in the image sequence of interest; Based on the signal quality index and signal periodic stability, a reliability score characterizing the reliability of the signal is determined; When the frame quality score is greater than a first preset score and the credibility score is greater than a second preset score, the physiological signal is sent to the main control terminal.

[0111] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be referred to the relevant descriptions on the server side and client side in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0112] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0113] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0114] It should be noted that any AI models, software tools, or components not belonging to this company appearing in the embodiments of this application are merely illustrative examples and do not represent actual use. All user personal information involved in the embodiments of this application has been authorized (with the knowledge and consent) by the relevant parties or has been fully authorized by all parties, and the executing entity may obtain it through various legal and compliant means. The collection, storage, use, processing, transmission, provision, and disclosure of the information, data, and signals involved all comply with relevant laws and regulations and do not violate public order and good morals.

[0115] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A remote health monitoring method, characterized in that, The remote health monitoring method is applied to each source end in the remote health monitoring system, which also includes a master control end. All source ends are used to connect to the master control end, including: Acquire continuous video frame data; For any frame image in the video frame data, the region containing a face in the frame image is extracted to obtain the region of interest image corresponding to the frame image; Traverse all frame images to obtain the region of interest image corresponding to each frame image. Sort all region of interest images according to the temporal order of each frame image in the video frame data to obtain the sequence of images of interest. Based on the image sequence of interest, determine the physiological signal characterizing the physiological characteristics, the signal quality index corresponding to the physiological signal, the signal periodic stability, and the frame quality score characterizing the image quality in the image sequence of interest; Based on the signal quality index and signal periodic stability, a reliability score characterizing the reliability of the signal is determined; When the frame quality score is greater than a first preset score and the credibility score is greater than a second preset score, the physiological signal is sent to the main control terminal.

2. The remote health monitoring method according to claim 1, characterized in that, The method for determining physiological signals characterizing physiological features includes: Obtain a preset analysis period and determine the number of image frames in the video frame data based on the analysis period; Based on the analysis time period and the number of image frames, target images are extracted sequentially from the sequence of images of interest to obtain the set of images to be analyzed. Based on the analysis period, pixel value analysis is performed on all target images in the image set to be analyzed to obtain a pixel change sequence; Based on the pixel change sequence, physiological signals characterizing physiological features are determined.

3. The remote health monitoring method according to claim 2, characterized in that, The method for determining the signal quality index corresponding to the physiological signal includes: Extract the brightness data of any consecutive target image in the image set to be analyzed from the video frame data; Based on the brightness data, the average brightness of the target image in any consecutive frame is calculated; Based on the preset region of interest, the region integrity rate of the image set to be analyzed is calculated. Based on the pixel change sequence, the standard deviation of the pixel change peak value is calculated to obtain the target image stability; The signal quality index corresponding to the physiological signal is calculated based on the preset weighting coefficients combined with the average brightness, the region integrity rate, and the target image stability.

4. The remote health monitoring method according to claim 2, characterized in that, The method for determining the periodic stability of the signal includes: Based on the video frame data, pixel value analysis is performed on all the frame images to obtain a set of pixel changes; Based on the set of pixel changes, the standard deviation of the peak value of the pixel changes is calculated to obtain the signal periodic stability.

5. The remote health monitoring method according to claim 3, characterized in that, The method for determining the frame quality score characterizing the image quality in the sequence of images of interest includes: Based on the region of interest, the percentage of the region of interest is calculated for each frame image. The average percentage of the region of interest is calculated by traversing all the frame images. The image sharpness is obtained by performing a sharpness analysis on the frame image according to a preset gradient algorithm. Obtain the image position information of any of the frame images, and determine the position change information of any consecutive frames based on the image position information; Based on the position change information, determine the image stability; A frame quality score, characterizing the image quality in the image sequence of interest, is determined based on the average proportion of the region of interest, the image sharpness, and the image stability.

6. The remote health monitoring method according to claim 1, characterized in that, Before sending the physiological signal to the main control terminal, the process also includes... When sending a signal to the main control terminal, the type of the signal to be sent is detected; If the signal type is not the type corresponding to the physiological signal, the transmission is blocked. If the signal type is the type corresponding to the physiological signal, the physiological signal is output at an output frequency not exceeding a preset frequency, where the preset frequency is less than the sampling frequency of the source end.

7. The remote health monitoring method according to any one of claims 1 to 6, characterized in that, The step of extracting the region containing a face from the frame image to obtain the region of interest image corresponding to the frame image includes: According to a preset detection algorithm, the regions containing human faces in the frame image are marked to obtain a representation of the human face location; Based on the face location representation, the regions in the frame image that do not contain faces are blurred to obtain a blurred image; Based on the face position representation, the blurred image is removed to obtain the region of interest image corresponding to the frame image.

8. A remote health monitoring device, characterized in that, The remote health monitoring device is applied to each source end of the remote health monitoring system. The remote health monitoring system also includes a main control end, and all source ends are used to connect to the main control end, including: The data acquisition module is used to acquire continuous video frame data; The region extraction module is used to extract the region containing a face in any frame image of the video frame data to obtain the region of interest image corresponding to the frame image. The sorting module is used to traverse all frame images, obtain the region of interest image corresponding to each frame image, and sort all region of interest images according to the temporal order of each frame image in the video frame data to obtain the sequence of images of interest. The quality analysis module is used to determine, based on the image sequence of interest, physiological signals characterizing physiological characteristics, signal quality index corresponding to the physiological signals, signal periodic stability, and frame quality scores characterizing the image quality in the image sequence of interest. The credibility scoring module is used to determine a credibility score that characterizes the credibility of the signal based on the signal quality index and the signal periodic stability. The signal transmission module is used to send the physiological signal to the main control terminal when the frame quality score is greater than a first preset score and the credibility score is greater than a second preset score.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the remote health monitoring method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the remote health monitoring method as described in any one of claims 1 to 7.

Images