Human body temperature compensation method and device based on monocular vision, equipment and medium

Abstract

The present application relates to the technical field of intelligent care, solves the problem of low accuracy of existing temperature measurement technology in temperature detection of the cared object, and provides a human body temperature compensation method, device, equipment and medium based on monocular vision. The method comprises: photographing at least one target heat source according to a preset distance sequence and a preset temperature range to obtain a thermal infrared temperature sequence; preprocessing the thermal infrared temperature sequence to obtain a compensation relationship between the temperature measurement value and the actual shooting distance; obtaining the actual target distance of the cared object according to the preset visible light image and the actual visible light image of the cared object; obtaining the actual body temperature value of the cared object according to the actual thermal infrared image of the cared object; and obtaining the corrected body temperature value of the cared object according to the actual body temperature value, the actual target distance and the compensation relationship. The present application can improve the temperature detection accuracy of the cared object and realize intelligent care.

Description

Technical Field

[0001] This invention relates to the field of intelligent care technology, and in particular to a method, device, equipment and medium for human body temperature compensation based on monocular vision. Background Technology

[0002] Monitoring the body temperature of those under care is crucial for their health and safety, including infants, adults, and the elderly. Elevated body temperature is often an early symptom of infection or other health problems; for example, fever may indicate viral or bacterial infections such as influenza, pneumonia, or urinary tract infections. Regular temperature monitoring allows for early detection of abnormal temperatures and timely medical intervention, reducing potential health risks and helping parents better care for their children. When an abnormal temperature is detected, parents can take appropriate action or seek medical help promptly. Regular temperature measurement and recording helps establish a health record for the child, providing a reference for future medical decisions. Current human body temperature measurement technologies mainly include infrared spot temperature technology, visible light technology, and thermal imaging + blackbody technology. Infrared spot temperature technology measures the infrared radiation emitted from the human body using an infrared sensor to calculate temperature, but it can only measure the temperature of a single point and is greatly affected by environmental factors. Visible light technology measures body temperature by capturing visible light images of the human body. However, due to spatial differences in image position and angle, especially for subjects moving frequently in complex environments, obtaining accurate body temperature data is difficult. Thermal imaging with a blackbody technique uses a thermal infrared camera to capture infrared radiation emitted by the human body or objects, generating a temperature image. A blackbody is an idealized thermal radiation source with completely predictable radiation characteristics. In a thermal imaging system, a blackbody is placed within the field of view, and the camera is calibrated by measuring its known temperature and radiation intensity to correct and improve temperature measurement accuracy. The blackbody radiation source provides a stable and known amount of radiation, allowing for precise calibration of the thermal imaging camera and ensuring accurate temperature measurement. However, although a blackbody can reduce environmental interference, changes in ambient temperature can still affect the stability of the equipment and the measurement results. Furthermore, thermal infrared images may contain thermal radiation reflections from the environment, affecting measurement accuracy.

[0003] Therefore, in monitoring scenarios, how to accurately detect the body temperature of the person being cared for, provide technical support for health monitoring of the person being cared for, and realize intelligent care is an urgent problem to be solved. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a method, device, equipment and medium for human body temperature compensation based on monocular vision, in order to solve the problem of low accuracy of body temperature detection of the cared-for object in the prior art.

[0005] In a first aspect, embodiments of the present invention provide a method for human body temperature compensation based on monocular vision, the method comprising:

[0006] At least one target heat source is photographed according to a preset distance sequence and a preset temperature range to obtain a thermal infrared temperature sequence, wherein the thermal infrared temperature sequence includes temperature measurement values ​​of the target heat source in several thermal infrared images, and the actual temperature and / or actual shooting distance of the target heat source in each thermal infrared image are different;

[0007] The thermal infrared temperature sequence is preprocessed to obtain the compensation relationship between the temperature measurement value and the actual shooting distance;

[0008] Based on the preset visible light image and the actual visible light image of the object being cared for, the actual target distance of the object being cared for is obtained, wherein the preset visible light image is obtained at a predetermined distance between the object being cared for and the visible light vision module;

[0009] The actual body temperature value of the person being cared for is obtained based on the actual thermal infrared image of the person being cared for. The actual body temperature value of the person being cared for is obtained at the actual shooting distance between the person being cared for and the thermal infrared vision module.

[0010] The corrected body temperature value of the person being cared for is obtained based on the actual body temperature value, the actual target distance, and the compensation relationship.

[0011] Preferably, the step of capturing images of at least one target heat source according to a preset distance sequence and a preset temperature range to obtain a thermal infrared temperature sequence includes:

[0012] The target heat source is photographed according to a preset distance sequence to obtain thermal infrared images at each actual temperature.

[0013] Based on the thermal infrared images, the images are sorted by size according to a preset distance to obtain a thermal infrared temperature sequence.

[0014] Preferably, the preprocessing of the thermal infrared temperature sequence to obtain the compensation relationship between the temperature measurement value and the actual shooting distance includes:

[0015] The difference between the measured temperature and the actual temperature of the target heat source at each preset distance in the preset distance sequence is calculated to obtain the first temperature difference sequence corresponding to each preset distance.

[0016] The first temperature difference sequence is processed using the least squares method or nonlinear curve fitting method to obtain the relationship function between each preset distance and each temperature measurement value.

[0017] The compensation relationship is obtained by performing polynomial fitting on each of the aforementioned relationship functions.

[0018] Preferably, obtaining the actual target distance of the monitored object based on a preset visible light image and an actual visible light image of the monitored object includes:

[0019] Acquire a preset visible light image and an actual visible light image;

[0020] Based on a preset neural network model, a preset visible light image, and an actual visible light image, the monitored object is identified, and a first target detection box and a second target detection box are obtained. The preset neural network model is built based on YOLO.

[0021] The actual target distance is obtained based on the first target detection box, the second target detection box, and the first preset algorithm.

[0022] Preferably, obtaining the actual target distance based on the first target detection box, the second target detection box, and the first preset algorithm includes:

[0023] Based on the first target detection box, obtain the first width and first height of the first target detection box;

[0024] Based on the second target detection box, obtain the second width and the second height of the second target detection box;

[0025] The area of ​​the first detection frame is obtained by calculating the first width and the first height;

[0026] The area of ​​the second detection frame is obtained by calculating the second width and the second height.

[0027] The actual target distance is obtained by calculating the area of ​​the first detection frame and the area of ​​the second detection frame.

[0028] Preferably, the calculation of the actual target distance by calculating the area of ​​the first detection frame and the area of ​​the second detection frame includes:

[0029] The actual visible light image is input into a pre-trained facial landmark detection model to determine whether there are missing landmarks in the actual visible light image.

[0030] When key points are missing, the area of ​​the first detection frame and the area of ​​the second detection frame are calculated to obtain the area ratio;

[0031] Determine the initial distance based on the area ratio;

[0032] The missing key points in the actual visible light image are identified to obtain the key point missing type, and the dynamic compensation factor corresponding to the key point missing type is obtained.

[0033] The initial distance is corrected based on the dynamic compensation factor to obtain the actual target distance.

[0034] Preferably, before obtaining the actual target distance of the monitored object based on the preset visible light image and the actual visible light image of the monitored object, the method further includes:

[0035] The system acquires images of the object being cared for at a predetermined distance from the visible light vision module and at a predetermined distance from the visible light vision module.

[0036] Based on the predetermined distance, facial analysis is performed on the image of the person being cared for to obtain facial analysis results, wherein the facial analysis includes facial key point detection and matching;

[0037] Based on the facial analysis results, when the face of the person being cared for in the image of the person being cared for presents a complete frontal view, the image of the person being cared for is used as the preset visible light image.

[0038] When the face of the person being cared for in the image is not in a complete frontal view, the target image is removed and a new image of the person being cared for is acquired. The incomplete frontal view includes a face that is obscured, a side view, and a face that is not visible.

[0039] Secondly, embodiments of the present invention provide a sleep detection device for a cared-for subject based on thermal infrared images, the device comprising:

[0040] The thermal infrared temperature sequence acquisition module captures images of at least one target heat source based on a preset distance sequence and a preset temperature range to obtain a thermal infrared temperature sequence.

[0041] The preprocessing module preprocesses the thermal infrared temperature sequence to obtain the compensation relationship between the temperature measurement value and the actual shooting distance;

[0042] The distance estimation module obtains the actual measured distance of the monitored object based on the preset visible light image and the actual visible light image of the monitored object;

[0043] The actual body temperature acquisition module acquires the actual body temperature value of the person being cared for based on the actual thermal infrared image of the person being cared for.

[0044] The body temperature compensation module obtains the corrected body temperature value of the person being cared for based on the actual body temperature value, the actual target distance, and the compensation relationship.

[0045] Thirdly, embodiments of the present invention provide an electronic device, including: at least one processor, at least one memory, and computer program instructions stored in the memory, which, when executed by the processor, implement the method of the first aspect described above.

[0046] Fourthly, embodiments of the present invention provide a storage medium storing computer program instructions, which, when executed by a processor, implement the method of the first aspect described above.

[0047] In summary, the beneficial effects of the present invention are as follows:

[0048] This invention provides a method, device, equipment, and medium for human body temperature compensation based on monocular vision. It involves capturing images of at least one target heat source according to a preset distance sequence and a preset temperature range to obtain a thermal infrared temperature sequence. Data acquisition at different distances and temperatures helps the system better adapt to various practical situations and provides a more comprehensive understanding and correction of measurement deviations in the thermal infrared imaging system. The thermal infrared temperature sequence is preprocessed to obtain a compensation relationship between the measured temperature value and the actual shooting distance. By analyzing and processing the thermal infrared temperature sequence, a compensation relationship between the measured temperature value and the actual shooting distance is established. This relationship can be used to correct measurement errors caused by distance variations, reducing the impact of errors on results in actual measurements. Based on a preset visible light image and an actual visible light image of the monitored object, the actual target distance of the monitored object is obtained. This actual target distance allows subsequent body temperature compensation algorithms to... Further adjustments to temperature measurements improve the accuracy of overall body temperature measurements for the monitored individuals. Based on the actual thermal infrared image of the monitored individual, the actual body temperature is obtained. Thermal infrared images are unaffected by changes in the visible light environment, allowing for stable temperature data acquisition under varying lighting conditions and adapting to diverse environments. Furthermore, thermal infrared technology enables non-contact measurement, which is particularly important for the monitored individuals, reducing interference and discomfort. Based on the actual body temperature value, the actual target distance, and the compensation relationship, a corrected body temperature value is obtained. By combining the actual body temperature value with the estimated distance and utilizing the compensation relationship to adjust the actual body temperature value, the compensation relationship corrects measurement errors caused by different distances. This corrects temperature deviations due to changes in measurement distance, thereby reducing system errors and providing a more accurate corrected body temperature value for the monitored individuals. This provides technical support for body temperature monitoring and enables intelligent care. Attached Figure Description

[0049] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments of the present invention will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, and these are all within the protection scope of the present invention.

[0050] Figure 1 This is a schematic flowchart of a human body temperature compensation method based on monocular vision according to an embodiment of the present invention.

[0051] Figure 2 This is a schematic flowchart of a human body temperature compensation method based on monocular vision according to an embodiment of the present invention.

[0052] Figure 3 This is another flowchart of an embodiment of the present invention for a human body temperature compensation method based on monocular vision;

[0053] Figure 4 This is another flowchart of an embodiment of the present invention for a human body temperature compensation method based on monocular vision;

[0054] Figure 5 This is another flowchart of an embodiment of the present invention for a human body temperature compensation method based on monocular vision;

[0055] Figure 6 This is a schematic diagram of the structure of a human body temperature compensation device based on monocular vision according to an embodiment of the present invention;

[0056] Figure 7 This is a schematic diagram of the device according to an embodiment of the present invention. Detailed Implementation

[0057] The features and exemplary embodiments of various aspects of the present invention will now be described in detail. To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only configured to explain the present invention and are not configured to limit the present invention. For those skilled in the art, the present invention can be practiced without some of these specific details. The following description of the embodiments is merely intended to provide a better understanding of the present invention by illustrating examples of the invention.

[0058] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0059] It should be noted that all actions involving the acquisition of signals, information, or data in this invention are carried out in compliance with the relevant data protection laws and regulations of the locality and with authorization from the owner of the relevant device.

[0060] Example 1

[0061] Please see Figure 1 This invention provides a method, apparatus, device, and medium for human body temperature compensation based on monocular vision. The method includes:

[0062] S1. At least one target heat source is photographed according to a preset distance sequence and a preset temperature range to obtain a thermal infrared temperature sequence, wherein the thermal infrared temperature sequence includes temperature measurement values ​​of the target heat source in several thermal infrared images, and the actual temperature and / or actual shooting distance of the target heat source in each thermal infrared image are different.

[0063] The preset distance sequence includes a series of different shooting distances set in advance when capturing thermal infrared images, such as fixed distances from 1 meter, 2 meters to 5 meters. These distances are used to simulate different distances between the camera and the target heat source in actual use, so as to collect temperature data at different distances. The preset temperature range includes the temperature range that the object being photographed (target heat source) needs to cover during the acquisition process, for example, set to 20℃ to 50℃. This range is used to ensure that the captured images can cover the temperature changes that the target heat source may experience under different conditions. The response of the thermal infrared camera may have errors at different distances and temperatures. By shooting at different preset temperature ranges and distances, the temperature measurement accuracy of the thermal infrared system under different conditions can be calibrated, thereby improving the overall measurement accuracy.

[0064] Preferably, see Figure 2 Based on a preset distance sequence and a preset temperature range, at least one target heat source is photographed to obtain a thermal infrared temperature sequence, including:

[0065] S11. Capture the target heat source according to the preset distance sequence to obtain thermal infrared images at each actual temperature;

[0066] Preset distance sequences can help calibrate and verify the performance of thermal infrared cameras, ensuring that they provide reliable image data at various distances. By taking pictures at different distances and comparing the images at different distances with standard or expected thermal distribution images, a more comprehensive understanding of the target's thermal characteristics can be obtained, improving the accuracy of the data.

[0067] Specifically, thermal infrared temperature images measured by a thermal infrared module are acquired at each actual temperature within a preset distance sequence from a single heat source. The single heat source refers to a single heating blackbody, obtained through artificial constant-temperature heating. The actual temperature primarily refers to the body temperature range of the person being monitored; here, eight temperature values ​​(integrity T) are selected (35℃, 36℃, 37℃, 38℃, 39℃, 40℃, 41℃, 42℃). The actual shooting distance is determined based on the safe monitoring distance of the person being monitored, and is set to five values: 0.5 meters, 0.8 meters, 1.1 meters, 1.4 meters, and 1.7 meters. The person being monitored refers to someone requiring continuous monitoring or care, typically including infants, the elderly, patients, or those with mobility impairments.

[0068] S12. Based on the thermal infrared image, sort the distances according to a preset distance sequence to obtain a thermal infrared temperature sequence;

[0069] Temperature sequences sorted by distance can help identify changes in heat sources at different distances and understand the thermal characteristics of the target. At the same time, sorting and analyzing temperature sequences provides clear data comparisons, which helps to verify and calibrate the accuracy of thermal infrared imaging systems.

[0070] Specifically, the thermal infrared images are sorted by size according to a preset distance sequence, resulting in a total of 5*8=40 data points. That is, for each actual shooting distance, the thermal infrared module measures 8 actual temperatures to obtain the thermal infrared temperature sequence R1.

[0071] S2. Preprocess the thermal infrared temperature sequence to obtain the compensation relationship between the temperature measurement value and the actual shooting distance;

[0072] Temperature measurements from thermal infrared images can be affected by the shooting distance. For example, when the target is far away, the temperature resolution in the image may decrease, thus affecting the accuracy of the measurement. By establishing a compensation relationship, temperature measurements at different distances can be standardized, which helps to compare and analyze data at different distances. Preprocessing the thermal infrared temperature sequence can effectively reduce measurement bias caused by distance variations, thereby improving the reliability and validity of the data.

[0073] Preferably, see Figure 3 The process of preprocessing the thermal infrared temperature sequence to obtain the compensation relationship between the temperature measurement value and the actual shooting distance includes:

[0074] S21. Based on the temperature measurement value of the target heat source, obtain a first temperature difference sequence under the preset distance sequence, wherein the first temperature difference sequence is the difference between the temperature measurement value of each target heat source and each preset temperature value under the preset distance sequence;

[0075] The first temperature difference sequence can be used to adjust and calibrate the thermal infrared imaging system. By understanding the measurement error at different distances, the system can be appropriately corrected to improve the overall measurement accuracy. It can also identify a scheme to optimize measurement conditions to obtain more accurate results.

[0076] Specifically, for each temperature measurement value T(i), the difference ΔT(i) = R1(i) - T(i) between the temperature measurement value of the thermal infrared module at each preset distance and each preset temperature value is calculated. For each distance, there are 8 difference values, and a total of 5*8 = 40 difference values ​​are obtained to obtain the first temperature difference value sequence.

[0077] S22. Process each of the first temperature difference sequences according to the least squares method or nonlinear curve fitting method to obtain the relationship function between the preset distance and the temperature measurement value.

[0078] By obtaining the relationship function between each preset distance and the temperature measurement value, the temperature measurement characteristics of the target at different distances can be accurately described. This allows for the calibration of the thermal infrared imaging system and adjustment of system settings to reduce the influence of distance on the temperature measurement value, thereby improving measurement accuracy.

[0079] Specifically, the least squares method is a statistical analysis technique used to fit data by minimizing the squared difference between observed values ​​and model predictions. It is particularly important in regression analysis to determine the best-fit line or curve that makes the predicted values ​​as close as possible to the actual data. The least squares method takes the following form:

[0080] Objective function = ∑(theoretical value - actual value) 2

[0081] The theoretical value is the target function to be fitted, which is of the form of m samples with only one feature: (x i ,y i (i = 1, 2, 3, ..., m), the samples are fitted using a general polynomial of degree n, f(x). a(α0,α1,α2...α nGiven a parameter α0, the least squares method aims to find a set of α0 such that... Minimum, i.e., finding Based on this, the present invention applies the least squares method to perform nonlinear fitting on the ΔT sequence for each temperature, and obtains the function G(T) that best describes the relationship between distance and temperature difference at that temperature. i ,d j ), where T represents temperature and d represents distance.

[0082] Polynomial functions offer high flexibility, capable of capturing complex nonlinear relationships. The order of the polynomial can be adjusted according to the actual data, thus achieving a good fit to various data distributions. Measurement data may contain local fluctuations or noise; polynomial functions can smooth these fluctuations by adjusting the polynomial's order, thereby better capturing the overall trend of the data. A key characteristic is its ability to balance overfitting and underfitting by adjusting the order. If the order is too low, it may fail to fit the data well (underfitting); if the order is too high, it may overfit the noise (overfitting). By selecting the optimal order through an optimization process, a balance can be found between these two factors, resulting in the best fit. The following is a detailed implementation method:

[0083] Start with a low polynomial order n=1 and gradually increase the order. For each temperature Ti, use polynomial form. To perform the fitting, the least squares method is used to calculate the fitting coefficients α0, α1, α2...α. n The fitted error is the mean squared error (MSE), used to evaluate the fitting effect. The goal is to minimize the error between the fitted function and the actual data. The MSE expression is as follows:

[0084]

[0085] The key to this process is selecting the optimal polynomial order. If the polynomial order is too low, the fit may be insufficient, resulting in a large error; while if the order is too high, it may lead to overfitting, which would actually increase the error. Therefore, by calculating the errors of different orders and selecting the polynomial order that minimizes the error, we can ensure that the compensation function can fit the data well without losing its generalization ability due to overfitting, thus ensuring the reliability of the compensation function under different temperatures and distances.

[0086] Gradually increase the order of the polynomial, repeating the fitting process and error calculation. Compare the fitting errors of different orders, record the polynomial order n with the smallest error, and obtain the relationship function between each preset distance and the temperature measurement value within the preset temperature range.

[0087] S23. Perform polynomial fitting on each of the aforementioned relational functions to obtain the compensation relation.

[0088] Based on the relationship function between the preset distances and the temperature measurement values The eight functions G(T1,d), G(T2,d), ... G(T8,d) corresponding to the eight temperature measurements are integrated into a continuous temperature-distance compensation function F'(T,d) by using a polynomial function fitting method. This allows for the acquisition of the compensation relationship between the temperature measurements and the actual shooting distance, enabling accurate compensation and prediction of distance changes at any temperature.

[0089] S3. Based on the preset visible light image and the actual visible light image of the object being cared for, obtain the actual target distance of the object being cared for, wherein the preset visible light image is obtained at a predetermined distance between the object being cared for and the visible light vision module;

[0090] Specifically, in real-world monitoring scenarios, the location of the person being monitored is not static, and the distance between the person and the camera may change. Therefore, estimating the actual distance in real time and adjusting the temperature measurement value can improve the stability and consistency of the measurement results. By obtaining the actual target distance, it is possible to ensure that the temperature measurement system maintains high accuracy and reliability. For special groups such as infants, accurate temperature monitoring is particularly important, as temperature data with smaller errors can better reflect the infant's actual health status.

[0091] Preferably, the process further includes the following step before step S3:

[0092] S301: Obtain the predetermined distance between the object being cared for and the visible light vision module, and the image of the object being cared for collected at the predetermined distance between the object being cared for and the visible light vision module;

[0093] Specifically, the visible light vision module is controlled to acquire images of the monitored object at a set fixed distance. These images are then used for subsequent key point detection and pose determination.

[0094] S302: Based on the predetermined distance, perform facial analysis on the image of the person being cared for to obtain facial analysis results, wherein the facial analysis includes facial key point detection and matching;

[0095] Specifically, firstly, based on the preset distance of the image acquired in S301, a keypoint detection model adapted to that distance is invoked to extract the face region and locate key points in the image of the person being cared for. Then, a matching algorithm is used to compare the detected key points with a standard frontal face template to identify the integrity and frontal pose of the face in the current image. This step not only improves the robustness of keypoint detection but also avoids misjudgments of key points due to different shooting angles, thus laying the foundation for selecting the optimal image.

[0096] S303: Based on the facial analysis results, when the face of the person being cared for in the image of the person being cared for presents a complete frontal view, the image of the person being cared for is used as the preset visible light image;

[0097] Specifically, based on the analysis results of S302, it is determined whether a complete frontal face is detected in the current image, meaning that the number of key points is complete, their positions are evenly distributed, and they highly match the frontal face template. When these conditions are met, it indicates that the current image has good quality and representativeness, and the system uses it as a standard visible light image for subsequent processing, such as modeling, comparison, or distance calibration. By selecting only complete frontal face images as references, errors in subsequent body temperature estimation can be effectively avoided, thereby improving the stability and reliability of the overall detection system.

[0098] S304: When the face of the person being cared for in the image of the person being cared for is in an incomplete frontal state, the target image is removed and a new image of the person being cared for is acquired. The incomplete frontal state includes the face being obscured, the face being in profile, and the face not appearing.

[0099] Specifically, if the analysis results in step S302 indicate that the current image contains an incomplete frontal face, the system will automatically determine the image as invalid and discard it in this step. Specific incomplete states include severely missing facial key points (e.g., eyes, nose, or mouth not detected), obvious side profile (key points distributed to one side), or no facial region detected in the image. After discarding the image, the system automatically triggers the image acquisition module to retake the image until a complete frontal face image meeting the standards is obtained. This strategy prevents subsequent processing errors caused by poor image quality, helps the system continuously output high-quality input data, and ensures the accuracy and consistency of modules such as face recognition, temperature fitting, and distance estimation.

[0100] Preferably, see Figure 4 Based on the preset visible light image and the actual visible light image of the monitored object, the actual target distance of the monitored object is obtained, including:

[0101] S31. Obtain the preset visible light image and the actual visible light image;

[0102] A preset visible light image P1, captured at a predetermined distance, records the body temperature characteristics of the person being cared for under normal body temperature conditions. This provides a fixed benchmark for subsequent compensation calculations in actual measurements. The actual visible light image P2, acquired during the actual measurement process, is taken under real-world conditions, and the actual shooting distance may differ from the preset distance. By comparing P1 and P2, the actual target distance can be obtained, allowing for necessary compensation of the body temperature measurement data of the person being cared for.

[0103] Specifically, the preset visible light image P1 refers to the normal body temperature image of the person being cared for, obtained by the user at a predetermined distance of 1m before use through the visible light vision module; the actual visible light image P2 refers to the normal body temperature image of the person being cared for, captured by the user at the actual shooting distance before use through the visible light vision module.

[0104] S32. Based on the preset neural network model, the preset visible light image, and the actual visible light image, identify the target object being monitored and obtain the first target detection box and the second target detection box. The preset neural network model is built based on YOLO.

[0105] By using a pre-defined neural network model, YOLOv8s, to detect targets in both pre-defined and actual visible light images, the monitored objects are identified. The recognition results, specifically the first target detection box R1 detected in the pre-defined visible light image and the second target detection box R2 detected in the actual visible light image, enable accurate target localization and size comparison. The first target detection box provides standard reference information, while the second target detection box reflects the target's position and size changes under actual measurement conditions. Combining the target detection box data with precise adjustments to the compensation model improves system robustness.

[0106] Specifically, the preset visible light image P1 and the actual visible light image P2 are fed into the YOLOV8s model for detection to identify the monitored object. In the recognition result, the first target detection box R1(x,y,w,h) of the preset visible light image and the second target detection box R2(x,y,w,h) of the actual visible light image are obtained, where x represents the horizontal position of the upper left corner of the target box in the image, y represents the vertical position of the upper left corner of the target box in the image, w represents the width of the target box, and h represents the height of the target box.

[0107] S33. Obtain the actual target distance based on the first target detection box, the second target detection box, and the first preset algorithm;

[0108] Preferably, S33 includes:

[0109] S331: Based on the first target detection box, obtain the first width and first height of the first target detection box;

[0110] S332: Based on the second target detection box, obtain the second width and the second height of the second target detection box;

[0111] S333: Calculate the area of ​​the first detection frame by taking the first width and the first height;

[0112] S334: Calculate the second width and the second height to obtain the area of ​​the second detection frame;

[0113] S335: Calculate the area of ​​the first detection frame and the area of ​​the second detection frame to obtain the actual target distance.

[0114] The width w and height h of the target detection box will differ in the preset image and the actual image due to distance variations. By comparing these two detection boxes, the actual target distance can be calculated.

[0115] Specifically, the first preset algorithm expression for the actual target distance is as follows:

[0116]

[0117] This estimation method uses the proportional change between the first target detection box and the second target detection box to estimate the actual target distance, which serves as the basis for body temperature measurement compensation.

[0118] Preferably, S335 includes:

[0119] S3351: Input the actual visible light image into the pre-trained facial landmark detection model to determine whether there are missing landmarks in the actual visible light image;

[0120] Specifically, by inputting actual visible light images into a pre-trained and robust facial landmark detection model, multiple predefined standard landmarks (such as the corners of the eyes, the tip of the nose, the corners of the mouth, and the chin) are automatically detected within the facial region. The detection results are compared with the standard landmark template to determine whether there are missing landmarks, i.e., the model fails to return valid location information for certain landmarks. This step helps to identify landmark loss problems caused by occlusion (such as hats, hands, baby pacifiers, etc.) or pose changes (such as side profile, pitch), thus providing a basis for subsequent distance correction and improving the robustness and adaptability of the system.

[0121] S3352: When key points are missing, calculate the area of ​​the first detection frame and the area of ​​the second detection frame to obtain the area ratio;

[0122] Specifically, the areas of the first detection box in the reference image (i.e., the area of ​​the standard face frame captured at a preset distance) and the second detection box in the current visible light image (i.e., the area of ​​the currently detected face frame) are first calculated. Then, the area ratio is calculated by scaling the two area values, serving as a quantitative basis for initially estimating the change in the relative distance between the face and the device. The area ratio, as a fundamental feature for distance estimation, can provide a certain degree of distance estimation reference even in the presence of occlusion, and is particularly suitable for cases of slight occlusion or partial loss.

[0123] S3353: Determine the initial distance based on the area ratio;

[0124] Specifically, based on the ratio of the areas of the first and second detection boxes, and combined with a preset reference distance (e.g., one meter when the reference image was captured), the initial distance estimate for the current image is derived using the area proportionality formula. This initial distance is the baseline estimate in the uncorrected state. Although it may contain errors when key points are missing, it still serves as the distance baseline value before the subsequent introduction of compensation factors. This method is simple to calculate, highly real-time, and suitable for real-time operation in embedded or mobile systems, saving computational resources for subsequent compensation processing.

[0125] S3354: Identify missing key points in actual visible light images, obtain the key point missing type, and acquire the dynamic compensation factor corresponding to the key point missing type;

[0126] Specifically, by analyzing the location and region of missing keypoints in actual visible light images, the system identifies the type of missing keypoints, such as missing eye keypoints, nose tip keypoints, mouth corner keypoints, or facial edge keypoints. Based on a preset mapping table of keypoint missing types and corresponding dynamic compensation factors, the system retrieves the most matching compensation factor value. This dynamic compensation factor, with a value less than 1, reflects the degree of reduction in the detection box area that may result from missing keypoints and is used for subsequent distance correction operations. This design considers the differences in estimation results caused by different types of occlusion, improving the system's distance estimation accuracy in complex shooting scenarios.

[0127] S3355: Based on the dynamic compensation factor, the initial distance is corrected to obtain the actual target distance.

[0128] Specifically, the initial distance is used as the base estimate, and then a dynamic compensation factor is used to correct the initial estimate by dividing the initial distance by the compensation factor to obtain the actual target distance. This correction mechanism effectively compensates for the error in the face bounding box area caused by missing key points, making the final estimated distance between the baby and the device closer to reality, thereby further improving the accuracy and reliability of the subsequent body temperature compensation function calculation. Furthermore, in an optional scheme, the distance value can be weighted and fused with the output of a monocular depth estimation model to achieve high-precision distance estimation under multi-source fusion.

[0129] S4. Obtain the actual body temperature value of the person being cared for based on the actual thermal infrared image of the person being cared for, wherein the actual body temperature value of the person being cared for is obtained at the actual shooting distance between the person being cared for and the thermal infrared vision module.

[0130] The actual body temperature value provides real data support for subsequent body temperature compensation, which ensures that the algorithm can effectively cope with various measurement distances and environmental changes. In actual use, it can obtain the actual body temperature value in real time and compensate to obtain the corrected body temperature value, thus improving the real-time performance and accuracy of the body temperature measurement system.

[0131] Specifically, the actual thermal infrared image of the person being cared for is obtained by shooting at a small actual shooting distance using a thermal infrared module, thereby obtaining the actual body temperature value R2 of the person being cared for.

[0132] S5. Based on the actual body temperature value, the actual target distance, and the compensation relationship, obtain the corrected body temperature value of the person being cared for;

[0133] By correcting the actual body temperature value through compensation, the measurement error caused by distance changes is corrected. The influence of actual measurement conditions on the body temperature measurement results of the person being cared for is taken into account. After compensation and adjustment, the true body temperature of the person being cared for can be reflected more accurately.

[0134] Preferably, see Figure 5 The corrected body temperature value of the person being cared for is obtained based on the actual body temperature value, the actual target distance, and the compensation relationship, including:

[0135] S51. Obtain the first temperature deviation based on the actual body temperature value, the actual target distance, and the compensation relationship;

[0136] Based on the actual body temperature value R2 and the actual target distance Dis, substitute R2 and Dis into the compensation relationship F'(T,d) to obtain the first temperature deviation ΔT. expected =F'(T R2 Dis).

[0137] S52. The person being cared for adjusts the actual body temperature value according to the first temperature deviation to obtain the corrected body temperature value of the person being cared for.

[0138] According to the first temperature deviation ΔT expected The actual body temperature value R2 is compensated to obtain the corrected body temperature value T of the person being cared for. true The formula for compensation adjustment is as follows:

[0139] T true =R2+ΔT expected

[0140] Relate R2 to ΔT expected The sum of these values ​​is the corrected body temperature of the person being cared for.

[0141] Example 2

[0142] Please see Figure 6 This invention provides a human body temperature compensation device based on monocular vision, the device comprising:

[0143] Specifically, the human body temperature compensation device based on monocular vision provided in this embodiment of the invention includes: a thermal infrared temperature sequence acquisition module, which captures images of at least one target heat source according to a preset distance sequence and a preset temperature range to obtain a thermal infrared temperature sequence; a preprocessing module, which preprocesses the thermal infrared temperature sequence to obtain a compensation relationship between the temperature measurement value and the actual shooting distance; a distance estimation module, which obtains the actual measurement distance of the monitored object based on a preset visible light image and an actual visible light image of the monitored object; an actual body temperature acquisition module, which obtains the actual body temperature value of the monitored object based on the actual thermal infrared image of the monitored object; and a body temperature compensation module, which obtains the corrected body temperature value of the monitored object based on the actual body temperature value, the actual target distance, and the compensation relationship.

[0144] This device can acquire data at different distances and temperatures, helping the system better adapt to various practical situations and providing a more comprehensive understanding and correction of measurement deviations in thermal infrared imaging systems. By analyzing and processing thermal infrared temperature sequences, a compensation relationship is established between the measured temperature value and the actual shooting distance. This relationship can be used to correct measurement errors caused by distance variations, reducing the impact of errors on results in actual measurements. The actual target distance allows subsequent body temperature compensation algorithms to further adjust the temperature measurement value, improving the overall accuracy of the monitored subject's body temperature measurement. Thermal infrared images are unaffected by changes in the visible light environment, allowing for stable acquisition of body temperature data under different lighting conditions, adapting to various environments. Furthermore, thermal infrared technology allows for non-contact measurement, which is particularly important for monitored subjects, reducing interference and discomfort. By combining the actual body temperature value with the estimated distance and adjusting the actual body temperature value using the compensation relationship, which corrects measurement errors caused by different distances, the device can correct body temperature deviations caused by changes in measurement distance, thereby reducing system errors and providing more accurate corrected body temperature values ​​for monitored subjects. This provides technical support for monitored subject body temperature monitoring and enables intelligent care.

[0145] It should be noted that each module and unit in the human body temperature compensation device based on monocular vision in this embodiment corresponds one-to-one with each step in the human body temperature compensation method based on monocular vision in the aforementioned embodiment. Therefore, the specific implementation of this embodiment can refer to the implementation of the aforementioned human body temperature compensation method based on monocular vision, and will not be repeated here.

[0146] Example 3

[0147] In addition, combined Figure 6 The human body temperature compensation method based on monocular vision described in this embodiment of the invention can be implemented by a human body temperature compensation device based on monocular vision, characterized in that it is used to implement the method as described in any one of claims 1-7. Figure 7 The diagram shows a hardware structure schematic of a human body temperature compensation device based on monocular vision provided by an embodiment of the present invention.

[0148] A human body temperature compensation device based on monocular vision may include a processor and a memory storing computer program instructions.

[0149] Specifically, the processor may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement embodiments of the present invention.

[0150] The memory may include a large-capacity storage device for data or instructions. For example, and not limitingly, the memory may include a hard disk drive (HDD), a floppy disk drive, flash memory, an optical disk drive, a magneto-optical disk drive, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory may include removable or non-removable (or fixed) media. Where appropriate, the memory may be internal or external to a data processing device. In a particular embodiment, the memory is a non-volatile solid-state memory. In a particular embodiment, the memory includes a read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM), or flash memory, or a combination of two or more of these.

[0151] The processor reads and executes computer program instructions stored in the memory to implement any of the monocular vision-based human body temperature compensation methods in the above embodiments.

[0152] In one example, a monocular vision-based human body temperature compensation device may also include a communication interface and a bus. For example, Figure 6 As shown, the processor 401, memory 402, and communication interface 403 are connected through bus 410 and complete communication with each other.

[0153] The communication interface is mainly used to enable communication between various modules, devices, units and / or equipment in the embodiments of the present invention.

[0154] A bus, including hardware, software, or both, couples components of an electronic device together. For example, and not limitingly, a bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, a bus may include one or more buses. While specific buses are described and illustrated in embodiments of the invention, the invention contemplates any suitable bus or interconnect.

[0155] Example 4

[0156] Furthermore, in conjunction with the monocular vision-based human body temperature compensation method described in the above embodiments, this invention can be implemented using a computer-readable storage medium. This computer-readable storage medium stores computer program instructions; when executed by a processor, these computer program instructions implement any of the monocular vision-based human body temperature compensation methods described in the above embodiments.

[0157] In summary, the embodiments of the present invention provide a method, apparatus, device, and medium for human body temperature compensation based on monocular vision.

[0158] It should be clarified that the present invention is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of the present invention.

[0159] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this invention are programs or code segments used to perform the required tasks. The programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0160] It should also be noted that the exemplary embodiments mentioned in this invention describe methods or systems based on a series of steps or apparatus. However, this invention is not limited to the order of the steps described above; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0161] The above description is merely a specific embodiment of the present invention. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the protection scope of the present invention.

Claims

1. A method for human body temperature compensation based on monocular vision, characterized in that, The method includes: At least one target heat source is photographed according to a preset distance sequence and a preset temperature range to obtain a thermal infrared temperature sequence, wherein the thermal infrared temperature sequence includes temperature measurement values ​​of the target heat source in several thermal infrared images, and the actual temperature and / or actual shooting distance of the target heat source in each thermal infrared image are different; The thermal infrared temperature sequence is preprocessed to obtain the compensation relationship between the temperature measurement value and the actual shooting distance; Based on the preset visible light image and the actual visible light image of the object being cared for, the actual target distance of the object being cared for is obtained, wherein the preset visible light image is obtained at a predetermined distance between the object being cared for and the visible light vision module; The actual body temperature value of the person being cared for is obtained based on the actual thermal infrared image of the person being cared for. The actual body temperature value of the person being cared for is obtained at the actual shooting distance between the person being cared for and the thermal infrared vision module. The corrected body temperature value of the person being cared for is obtained based on the actual body temperature value, the actual target distance, and the compensation relationship.

2. The human body temperature compensation method based on monocular vision according to claim 1, characterized in that, The step of capturing images of at least one target heat source based on a preset distance sequence and a preset temperature range to obtain a thermal infrared temperature sequence includes: The target heat source is photographed according to a preset distance sequence to obtain thermal infrared images at each actual temperature. Based on the thermal infrared images, the images are sorted by size according to a preset distance to obtain a thermal infrared temperature sequence.

3. The method for human body temperature compensation based on monocular vision according to claim 1, characterized in that, The step of preprocessing the thermal infrared temperature sequence to obtain the compensation relationship between the temperature measurement value and the actual shooting distance includes: The difference between the measured temperature and the actual temperature of the target heat source at each preset distance in the preset distance sequence is calculated to obtain the first temperature difference sequence corresponding to each preset distance. The first temperature difference sequence is processed using the least squares method or nonlinear curve fitting method to obtain the relationship function between each preset distance and each temperature measurement value. The compensation relationship is obtained by performing polynomial fitting on each of the aforementioned relationship functions.

4. The method for human body temperature compensation based on monocular vision according to claim 1, characterized in that, The step of obtaining the actual target distance of the monitored object based on the preset visible light image and the actual visible light image of the monitored object includes: Acquire a preset visible light image and an actual visible light image; Based on a preset neural network model, a preset visible light image, and an actual visible light image, the monitored object is identified, and a first target detection box and a second target detection box are obtained. The preset neural network model is built based on YOLO. The actual target distance is obtained based on the first target detection box, the second target detection box, and the first preset algorithm.

5. The human body temperature compensation method based on monocular vision according to claim 4, characterized in that, The step of obtaining the actual target distance based on the first target detection box, the second target detection box, and the first preset algorithm includes: Based on the first target detection box, obtain the first width and first height of the first target detection box; Based on the second target detection box, obtain the second width and the second height of the second target detection box; The area of ​​the first detection frame is obtained by calculating the first width and the first height; The area of ​​the second detection frame is obtained by calculating the second width and the second height. The actual target distance is obtained by calculating the area of ​​the first detection frame and the area of ​​the second detection frame.

6. A method for human body temperature compensation based on monocular vision according to any one of claims 5, characterized in that, The calculation of the actual target distance by the area of ​​the first detection frame and the area of ​​the second detection frame includes: The actual visible light image is input into a pre-trained facial landmark detection model to determine whether there are missing landmarks in the actual visible light image. When key points are missing, the area of ​​the first detection frame and the area of ​​the second detection frame are calculated to obtain the area ratio; Determine the initial distance based on the area ratio; The missing key points in the actual visible light image are identified to obtain the key point missing type, and the dynamic compensation factor corresponding to the key point missing type is obtained. The initial distance is corrected based on the dynamic compensation factor to obtain the actual target distance.

7. A method for human body temperature compensation based on monocular vision according to any one of claims 1-6, characterized in that, Before obtaining the actual target distance of the monitored object based on the preset visible light image and the actual visible light image of the monitored object, the method further includes: The system acquires images of the object being cared for at a predetermined distance from the visible light vision module and at a predetermined distance from the visible light vision module. Based on the predetermined distance, facial analysis is performed on the image of the person being cared for to obtain facial analysis results, wherein the facial analysis includes facial key point detection and matching; Based on the facial analysis results, when the face of the person being cared for in the image of the person being cared for presents a complete frontal view, the image of the person being cared for is used as the preset visible light image. When the face of the person being cared for in the image is not in a complete frontal view, the target image is removed and a new image of the person being cared for is acquired. The incomplete frontal view includes a face that is obscured, a side view, and a face that is not visible.

8. A human body temperature compensation device based on monocular vision, characterized in that, The device includes: The thermal infrared temperature sequence acquisition module captures images of at least one target heat source based on a preset distance sequence and a preset temperature range to obtain a thermal infrared temperature sequence. The preprocessing module preprocesses the thermal infrared temperature sequence to obtain the compensation relationship between the temperature measurement value and the actual shooting distance; The distance estimation module obtains the actual measured distance of the monitored object based on the preset visible light image and the actual visible light image of the monitored object; The actual body temperature acquisition module acquires the actual body temperature value of the person being cared for based on the actual thermal infrared image of the person being cared for. The body temperature compensation module obtains the corrected body temperature value of the person being cared for based on the actual body temperature value, the actual target distance, and the compensation relationship.

9. An electronic device, characterized in that, include: At least one processor, at least one memory, and computer program instructions stored in the memory, which, when executed by the processor, implement the method as described in any one of claims 1-7.

10. A storage medium storing computer program instructions thereon, characterized in that, The method as described in any one of claims 1-7 is implemented when the computer program instructions are executed by the processor.

Images