Human skin temperature extraction method and thermal comfort classification method

Abstract

The invention relates to the technical field of thermal comfort evaluation, in particular to a human skin temperature extraction method and a thermal comfort classification method. According to the method, the infrared thermal imaging and the visible light image are subjected to registration processing, then the human body region of interest is identified from the visible light image and mapped to the infrared thermal imaging, and therefore the human body region of interest is positioned on the infrared thermal imaging; and then extracting the skin temperature of the region of interest of the human body on infrared thermal imaging, constructing a human body thermal comfort prediction model by utilizing different combinations of the skin temperature and the environmental parameters, and finally obtaining the thermal comfort of the human body for classification by using the optimal performance model. Each region of interest of the human body can be accurately positioned from the visible light image, so that the defect that the region of interest is difficult to position through infrared thermal imaging can be made up through the visible light image, the skin temperature of the region of interest is accurately extracted, and accurate classification of thermal comfort is realized through the accurate skin temperature.

Description

Technical Field

[0001] This invention relates to the field of thermal comfort assessment technology, specifically to a method for extracting human skin temperature and a method for classifying thermal comfort. Background Technology

[0002] Human skin temperature is related to thermal comfort, therefore, skin temperature needs to be extracted when analyzing thermal comfort. Infrared thermal imaging technology can extract skin temperature; however, due to the differences in skin temperature across different regions of interest (ROIs) and the inability of infrared thermal imaging technology to precisely locate each ROI, the extracted skin temperature cannot be matched to each ROI, and the skin temperature of each ROI affects human thermal comfort.

[0003] In summary, existing technologies have reduced the classification results for thermal comfort.

[0004] Therefore, existing technologies still need to be improved and enhanced. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method for extracting human skin temperature and a method for classifying thermal comfort, which solves the problem that existing technologies reduce the classification results of thermal comfort.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] In a first aspect, the present invention provides a method for extracting human skin temperature, comprising:

[0008] Acquire infrared thermal images and visible light images of the human body, and perform registration processing on the infrared thermal images and the visible light images;

[0009] Identify the region of interest of the human body from the visible light image;

[0010] The region of interest of the human body is mapped onto the registered infrared thermal image, and the skin temperature of the region of interest of the human body is obtained based on the infrared thermal image. The skin temperature is used for thermal comfort classification of the human body.

[0011] In one implementation, the infrared thermal image and the visible light image are registered, including:

[0012] Common key points of the human body were identified from the infrared thermal imaging and the visible light images, respectively.

[0013] Based on the key points of the human body, the infrared thermal image and the visible light image are registered.

[0014] In one implementation, identifying the region of interest of the human body from the visible light image includes:

[0015] Regions of interest (ROIs) for the face, hands, arms, legs, and neck are identified from the visible light image, and these regions are designated as human body ROIs.

[0016] In one implementation, obtaining the skin temperature of the region of interest in the human body based on the infrared thermal imaging includes:

[0017] Statistically analyze the pixel values ​​of the region of interest of the human body on the infrared thermal image;

[0018] The skin temperature of the human body region of interest is obtained based on the pixel values ​​of the region of interest.

[0019] Secondly, embodiments of the present invention also provide a thermal comfort classification method, including:

[0020] Based on the human skin temperature extraction method described above, the skin temperature of the region of interest in the human body is obtained;

[0021] Human body parameter information and environmental parameters of the human body are obtained, and the skin temperature and environmental parameters are classified based on the human body parameter information to obtain classification results. The classification results are used to characterize the thermal comfort level of the human body under the influence of the skin temperature and environmental parameters.

[0022] In one implementation, the skin temperature and environmental parameters are classified based on the human body parameter information to obtain a classification result, including:

[0023] The XGBoost model is applied to the human body parameter information, the skin temperature, and the environmental parameters to obtain the classification results output by the XGBoost model.

[0024] In one implementation, the skin temperature, the human body parameter information, and the environmental parameter are respectively the filtered skin temperature, human body parameter information, and environmental parameter; wherein the filtering index is the thermal comfort contribution rate.

[0025] Thirdly, embodiments of the present invention also provide a human skin temperature extraction device, wherein the device comprises the following components:

[0026] The registration module is used to acquire infrared thermal images and visible light images of the human body, and to perform registration processing on the infrared thermal images and the visible light images.

[0027] The region of interest (ROI) identification module is used to identify the region of interest of the human body from the visible light image;

[0028] The temperature extraction module is used to map the region of interest of the human body onto the registered infrared thermal image, and to obtain the skin temperature of the region of interest of the human body based on the infrared thermal image. The skin temperature is used for thermal comfort classification of the human body.

[0029] Fourthly, embodiments of the present invention also provide a terminal device, wherein the terminal device includes a memory, a processor, and a human skin temperature extraction program stored in the memory and executable on the processor, wherein when the processor executes the human skin temperature extraction program, it implements the steps of the human skin temperature extraction method described above.

[0030] Alternatively, the terminal device may include a memory, a processor, and a thermal comfort classification program stored in the memory and executable on the processor. When the processor executes the thermal comfort classification program, it implements the steps of the thermal comfort classification method described above.

[0031] Fifthly, embodiments of the present invention also provide a computer-readable storage medium storing a human skin temperature extraction program, wherein when the human skin temperature extraction program is executed by a processor, the steps of the human skin temperature extraction method described above are implemented.

[0032] Alternatively, the computer-readable storage medium stores a thermal comfort classification program, which, when executed by a processor, implements the steps of the thermal comfort classification method described above.

[0033] Beneficial Effects: This invention first performs registration processing on infrared thermal imaging and visible light images. Registration processing, also known as alignment processing, involves aligning the coordinates of identical points in the infrared thermal imaging and visible light images. Then, regions of interest (ROIs) are identified from the visible light image and mapped onto the infrared thermal image, thereby locating the ROIs on the infrared thermal image. The skin temperature of each ROI is then extracted from the infrared thermal image, and finally, this skin temperature is used to classify thermal comfort. Since each ROI can be accurately located from the visible light image, it compensates for the difficulty in locating ROIs in infrared thermal imaging. This allows for the accurate extraction of skin temperature from each ROI, enabling precise classification of thermal comfort. The relationship between skin temperature and thermal comfort varies among the different ROIs. Classification based solely on skin temperature or environmental parameters can lead to inaccurate classification. Therefore, by selecting different combinations of skin temperature and environmental parameters for each ROI and using various prediction models, the accuracy of thermal comfort classification can be improved. Attached Figure Description

[0034] Figure 1 This is an overall flowchart of the present invention;

[0035] Figure 2 This is a flowchart of the temperature extraction process in an embodiment of the present invention;

[0036] Figure 3 This is a schematic diagram of key human body points in an embodiment of the present invention;

[0037] Figure 4 This is a registration diagram in an embodiment of the present invention;

[0038] Figure 5 This is a schematic diagram showing the comparison before and after registration in an embodiment of the present invention;

[0039] Figure 6 This is a schematic diagram illustrating the numbering of facial feature points in an embodiment of the present invention;

[0040] Figure 7 This is a schematic diagram of the hand feature point numbering in an embodiment of the present invention;

[0041] Figure 8 This is a schematic diagram of the numbering of key human body points in an embodiment of the present invention;

[0042] Figure 9 This is a schematic diagram of thermal comfort grading in an embodiment of the present invention;

[0043] Figure 10 A structural diagram of the human skin temperature extraction device provided by the present invention;

[0044] Figure 11 This is a block diagram illustrating the internal structure of a terminal device provided in an embodiment of the present invention. Detailed Implementation

[0045] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments and accompanying drawings. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0046] Research has found that human skin temperature is related to thermal comfort; therefore, skin temperature needs to be extracted when analyzing thermal comfort. Infrared thermal imaging technology can extract skin temperature, but because skin temperature varies across different regions of interest (ROIs) and infrared thermal imaging cannot precisely locate each ROI, the extracted skin temperature cannot be matched to each ROI, and the skin temperature of each ROI affects human thermal comfort.

[0047] To address the aforementioned technical problems, this invention provides a method for extracting human skin temperature and a method for classifying thermal comfort, which solves the problem that existing technologies reduce the classification results of thermal comfort.

[0048] Example 1: This example provides a method for extracting human skin temperature. This method can be applied to a terminal device, which can be a terminal product with image processing capabilities, such as a computer. In this example, as... Figure 1 As shown, the method for extracting human skin temperature specifically includes the following steps:

[0049] S100: Acquire infrared thermal imaging and visible light images of the human body, and perform registration processing on the infrared thermal imaging and the visible light images;

[0050] S200, Identify the region of interest of the human body from the visible light image;

[0051] S300, the region of interest of the human body is mapped onto the registered infrared thermal image, and the skin temperature of the region of interest of the human body is obtained based on the infrared thermal image, and the skin temperature is used for thermal comfort classification of the human body.

[0052] In this embodiment, step S100 uses an infrared thermal imager (in this embodiment, the infrared thermal imager is a FLIR thermal imager) to collect infrared thermal images of the entire skin area of ​​the human body. That is, the infrared thermal imager can sense the infrared radiation formed on the skin surface and then form an infrared thermal image based on the infrared radiation. The infrared thermal imager captures the skin temperature distribution of the subject in real time, generating an infrared thermal image with a resolution of up to 640×512 pixels. The temperature measurement accuracy reaches ±0.1℃, fully meeting the accuracy requirements of thermal comfort research.

[0053] Visible light images of the human body are acquired using a GoPro Hero 10 visible light camera, which is synchronized with an infrared thermal imager. This camera boasts a high resolution of 5568×4176 pixels, a compact design for easy experimental setup, and supports remote control shooting.

[0054] Both infrared thermal imagers and visible light cameras are non-contact measuring instruments, which improves the convenience of measurement.

[0055] In this embodiment, step S100 employs the following steps to perform registration processing between infrared thermal imaging and visible light images: identifying common human body key points from the infrared thermal imaging and the visible light images respectively; and performing registration processing on the infrared thermal imaging and the visible light images based on the human body key points.

[0056] like Figure 2As shown, human key points are detected in both infrared thermal imaging and visible light images. The human key points in this embodiment are as follows: Figure 3 As shown, where Figure 3 The "a" icon marks the key points of the human body detected when the person is in a seated posture. Figure 3 The icon 'b' marks the key points of the human body detected when the human body is in a standing posture. As shown in Table 1, the number of key points of the human body in this embodiment is 25.

[0057] Table 1

[0058]

[0059]

[0060] The OpenPose model was used to detect the aforementioned 25 human body key points. As one of the advanced human pose estimation models, OpenPose's core advantage lies in its ability to accurately identify and locate 25 key points of the human body. These key points systematically cover the main functional parts in the human kinetic chain, including the head, neck, shoulder joint, elbow joint, wrist joint, hip joint, knee joint, and ankle joint, and based on this, a human skeleton model with a complete topological structure was constructed.

[0061] The aforementioned 25 key human body points constitute the approximate skeletal structure of the human body. The OpenPose model can detect these 25 key human body points in both infrared thermal imaging and visible light images. This means that the same key points can be detected in both images. Using these identical key points, registration between the infrared thermal imaging and visible light images can be achieved. Registration involves aligning the spatial coordinates of the same body part in the two images using the same key points. Specifically, the image registration process solves for the geometric transformation parameters (including translation, rotation, scaling, etc.) between the infrared thermal imaging and visible light images, enabling precise spatial correspondence of the key human body points in the two images. First, the 25 key human body points identified by OpenPose's human pose recognition are used as feature points. Then, the 25 key human body points from the two images are used as 25 pairs of matching feature points. The least squares method is used to calculate the affine transformation matrix. This matrix accurately represents the geometric transformation relationships such as rotation, scaling, and translation between the images, thereby constructing a spatial mapping model between the visible light image and the thermal imaging image (the mapping relationship between the visible light image and the thermal imaging image is as follows). Figure 4 (As shown). After obtaining the affine transformation matrix, it is applied to the original visible light image to finally obtain a visible light image that is precisely registered with the infrared thermal imaging. Figure 5 Image a is the original visible light image. Figure 5 Image b is the original thermal imaging image. Figure 5Figure c shows the registered visible light image. The registration method in this embodiment performs well in terms of registration accuracy. This study adopts a batch processing method that registers each pair of thermal imaging images and visible light images one by one. The technology is implemented on the Python platform, mainly relying on the OpenCV library to complete the image batch processing operations, and using the matplotlib library for visualization analysis.

[0062] In this embodiment, the regions of interest in the human body in step S200 include the region of interest in the face, the region of interest in the hands, the region of interest in the arms, the region of interest in the legs, and the region of interest in the neck.

[0063] The OpenPose recognition algorithm was used to detect visible light images to identify 68 facial feature points as shown in Table 2. These 68 facial feature points are systematically distributed in five core areas: the peri-eye region (including the pupil center and eyelid contour), the eyebrow region, the nose contour, the mouth region (including the lip contour), and the outer contour of the face.

[0064] The OpenPose facial recognition algorithm employs an independent convolutional neural network architecture. Through multi-level feature extraction and fusion mechanisms, it effectively enhances the algorithm's robustness in detecting facial features under complex backgrounds, varying lighting conditions, and multi-angle facial poses. Specifically, regardless of whether the human body is standing or sitting, the OpenPose algorithm can detect facial feature points.

[0065] Table 2

[0066]

[0067] The facial regions of interest (ROIs) are determined using the aforementioned 68 facial feature points. The distribution and labeling of these 68 facial feature points on the face are as follows: Figure 6 As shown. The facial regions of interest include the forehead region of interest, the nose region of interest, the cheek region of interest, the eye region of interest, and the mouth region of interest.

[0068] The region of interest for the eyes includes the region of interest for the left eye and the region of interest for the right eye. The region of interest for the right eye is defined by the five feature points on the right eyebrow in Table 2 (i.e., ...). Figure 6 The feature points numbered 17, 18, 19, 20, and 21 in the middle eye and four feature points on the right eye contour (i.e., Figure 6 The region of interest for the right eye is formed by connecting the nine feature points (numbered 39, 40, 41, and 36) in a clockwise direction. The region of interest for the left eye is... Figure 6The region formed by connecting the feature points numbered 22, 23, 24, 25, 26, 45, 46, 47, and 42 in Table 2 in a clockwise direction.

[0069] The area of ​​interest for the nose is Figure 6 The region formed by connecting the feature points numbered 27, 35, 34, 33, 32, and 31 in Table 2 in a clockwise direction.

[0070] The region of interest (ROI) for the cheek includes the left cheek ROI and the right cheek ROI, with the left cheek ROI being... Figure 6 The region of interest for the right cheek is formed by connecting the feature points numbered 1, 40, 31, 4, 3, and 2 in Table 2 in a clockwise direction. Figure 6 The region formed by connecting the feature points numbered 47, 15, 14, 13, 12, and 35 in Table 2 in a clockwise direction.

[0071] The area of ​​interest of the mouth Figure 6 The region formed by connecting the feature points numbered 48 to 59 in Table 2 in a clockwise direction.

[0072] The method for constructing the region of interest (ROI) on the forehead includes: determining the line connecting the center points of the two eyebrows, determining the vertical distance from the line to the highest point of the nose, and defining the rectangle with the line as its length, the endpoint of the line as its lower vertex, and the vertical distance as its width as the region of interest on the forehead.

[0073] The region of interest (ROI) for the hand includes the palm ROI and the finger ROI. The palm ROI is constructed by using OpenPose to identify features on the hand such as... Figure 7 The 21 feature points shown will Figure 7 The polygonal regions formed by feature points numbered 0, 1, 2, 5, 9, 13, and 17 are designated as regions of interest (ROIs) for the hand. The names of these 21 feature points are shown in Table 3. OpenPose uses a deep convolutional neural network to process hand images and incorporates multi-stage inference optimization during detection to ensure accurate capture and tracking of hand keypoints even in complex backgrounds and under varying lighting conditions.

[0074] Table 3

[0075]

[0076] The aforementioned feature points correspond to the main outline of the palm and can effectively reflect the temperature distribution characteristics of the palm area.

[0077] The temperature value of the finger ROI is determined by calculating the average temperature of the key points of each finger. By averaging the temperature of all key points of each finger, a more representative finger temperature value can be obtained.

[0078] The region of interest for the arm includes the region of interest for the left arm and the region of interest for the right arm, such as... Figure 8 As shown, 25 key points were identified on the human body, with the region of interest on the left arm being [data missing]. Figure 8 The midpoint of the line connecting the two key points numbered 6 and 7, where Figure 8 Keypoint 6 represents the left elbow, and keypoint 7 represents the left wrist. The region of interest for the right arm is... Figure 8 The midpoint of the line connecting the two key points numbered 3 and 4, where Figure 8 Key point number 3 represents the right elbow, and key point number 4 represents the right wrist. These two midpoints are located in the middle area of ​​the left and right arms, respectively, and can effectively capture the temperature change trend of the arms.

[0079] The region of interest for the legs includes the region of interest for the left calf and the region of interest for the right calf, with the region of interest for the right calf being... Figure 8 The midpoint of the line connecting keypoints 10 and 11 represents the center of the right calf, allowing for precise extraction of temperature data for that region. Keypoint 10 represents the right knee, and keypoint 11 represents the right ankle. The region of interest for the left calf is... Figure 8 The midpoint of the line connecting the two key points numbered 13 and 14 is where key point 13 represents the left knee and key point 14 represents the left ankle.

[0080] Neck region of interest Figure 8 The midpoint of the line connecting keypoints 0 and 1 is used, where keypoint 0 represents the nose and keypoint 1 represents the neck. This midpoint represents the core area of ​​the neck and effectively reflects its temperature distribution characteristics. Choosing the midpoint as the region of interest not only avoids interference from clothing but also comprehensively reflects the temperature distribution characteristics of major parts of the human body.

[0081] In this embodiment, step S300, mapping the regions of interest (ROIs) of the human body to the registered infrared thermal image, involves locating each ROI in the infrared thermal image to obtain the skin temperature of each ROI. The skin temperature of each ROI is determined by the infrared thermal image of that ROI.

[0082] T scene =S·Res

[0083] In the formula, S represents the pixel value of the region of interest in the human body on infrared thermal imaging, Res represents the conversion value between pixel value and temperature, and in this embodiment, Res is 0.04 Kelvin, T scene This represents skin temperature. The formula above allows for the accurate conversion of pixel values ​​in thermal imaging into actual skin temperature values.

[0084] Example 2 provides a thermal comfort classification method for determining the thermal comfort level of a human body, including the following specific steps: acquiring human body parameter information and environmental parameters of the human body, and classifying the skin temperature and environmental parameters based on the human body parameter information to obtain a classification result, wherein the classification result is used to characterize the thermal comfort level of the human body under the influence of the skin temperature and environmental parameters.

[0085] The thermal comfort classification method in this embodiment is applied in the following scenarios:

[0086] When constructing a teaching building for a school, the thermal comfort of the teachers and students using the building needs to be considered. The ventilation environment of the teaching building should be adjusted based on the thermal comfort classification results to improve thermal comfort. The specific process is as follows: Before teachers and students move into the teaching building, their thermal comfort is first predicted based on their individual skin temperature, individual parameter information, and the pre-designed environmental parameters of the teaching building. If the predicted thermal comfort classification result is uncomfortable, the pre-designed environmental parameters are adjusted, and the thermal comfort classification result is re-evaluated based on the adjusted environmental parameters until a comfortable classification result is achieved. The environmental parameters at this point are used as a reference for constructing the teaching building, so that the final constructed teaching building can provide a good thermal comfort environment.

[0087] This embodiment applies the XGBoost model to human body parameter information, specifically skin temperature and environmental parameters, to obtain the classification results output by the XGBoost model. The classification results are represented by TSV, meaning they include cold, neutral, and hot. Cold (TSV < -1) indicates the individual is in a cold state with significantly reduced thermal comfort; neutral (-1 ≤ TSV ≤ 1) indicates the individual is in a thermally neutral state with optimal thermal comfort; and hot (TSV > 1) indicates the individual is in a hot state with significantly decreased thermal comfort. Transforming TSV into a three-class classification problem simplifies the model's output dimension and reduces computational complexity. Furthermore, this classification method is closer to the decision-making needs in practical engineering applications, facilitating guidance for indoor environmental control. The above three-level classification of thermal comfort is as follows: Figure 9 As shown.

[0088] The human body parameters in this embodiment include gender, age, body mass index (BMI), and metabolic rate (MRC). Environmental parameters include ambient temperature, relative humidity, and wind speed. This embodiment filters for gender, age, BMI, skin temperature of each region of interest, ambient temperature, relative humidity, and wind speed based on environmental features such as thermal comfort contribution rate (SHAP). The thermal comfort contribution rate is the SHAP value; a higher SHAP value indicates a greater contribution of that feature to the predicted thermal comfort. Wind speed (SHAP = 0.13), metabolic rate (SHAP = 0.1), and finger temperature (SHAP = 0.08) constitute the three key factors influencing thermal comfort prediction, with significantly higher importance than other features. Relative humidity, BMI, and air temperature are in the second tier, having a certain impact on the prediction results as secondary influencing factors. Based on the SHAP value, this embodiment uses wind speed, metabolic rate, finger temperature, relative humidity, and BMI for thermal comfort classification.

[0089] This embodiment applies the XGBoost model (Extreme Gradient Boosting) to wind speed, metabolic rate, finger temperature, relative humidity, and BMI to obtain classification results for thermal comfort. The XGBoost model achieves an accuracy of 78.5%, and also performs well in other evaluation metrics: precision 84.8%, recall 63.4%, and F1 score 69.4%. This embodiment can also employ other machine learning models besides XGBoost, including logistic regression, SVM, RF, and MLP. In summary, this invention develops a highly efficient automatic skin temperature extraction algorithm by integrating human pose recognition and image registration technologies. This method innovatively achieves precise localization of key human body parts and spatial alignment of multimodal images, significantly improving the accuracy and efficiency of skin temperature data extraction.

[0090] This invention introduces the SHAP interpretable machine learning method into the field of thermal comfort prediction, constructing a thermal comfort prediction optimization model applicable to multiple scenarios. By quantitatively analyzing the contribution of characteristic variables such as skin temperature and environmental parameters to the prediction results, the mechanism of action of each factor on thermal comfort perception is revealed, providing a scientific basis for model optimization.

[0091] This embodiment also provides a human skin temperature extraction device, such as... Figure 10 As shown, the device comprises the following components:

[0092] The registration module 01 is used to acquire infrared thermal images and visible light images of the human body, and to perform registration processing on the infrared thermal images and the visible light images.

[0093] Region of interest identification module 02 is used to identify the region of interest of the human body from the visible light image;

[0094] Temperature extraction module 03 is used to map the human body region of interest onto the registered infrared thermal image, and based on the infrared thermal image, obtain the skin temperature of the human body region of interest, which is used for thermal comfort classification of the human body.

[0095] Based on the above embodiments, the present invention also provides a terminal device, the principle block diagram of which can be as follows: Figure 11 As shown, the terminal device includes a processor, memory, network interface, and display screen 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 stored in the non-volatile storage media. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for extracting human skin temperature. The display screen can be a liquid crystal display (LCD) or an e-ink display.

[0096] Those skilled in the art will understand that Figure 11 The schematic diagram shown is only a partial structural diagram related to the present invention and does not constitute a limitation on the terminal device to which the present invention is applied. The specific terminal device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0097] In one embodiment, a terminal device is provided, comprising a memory, a processor, and a human skin temperature extraction program stored in the memory and executable on the processor. When the processor executes the human skin temperature extraction program, it implements the following operation instructions:

[0098] Acquire infrared thermal images and visible light images of the human body, and perform registration processing on the infrared thermal images and the visible light images;

[0099] Identify the region of interest of the human body from the visible light image;

[0100] The region of interest of the human body is mapped onto the registered infrared thermal image, and the skin temperature of the region of interest of the human body is obtained based on the infrared thermal image. The skin temperature is used for thermal comfort classification of the human body.

[0101] 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 by this invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various 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 RAMbus dynamic RAM (RDRAM), etc.

[0102] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; 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; and these 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.

Claims

1. A method for extracting human skin temperature, characterized in that, include: Acquire infrared thermal images and visible light images of the human body, and perform registration processing on the infrared thermal images and the visible light images; Identify the region of interest of the human body from the visible light image; The region of interest of the human body is mapped onto the registered infrared thermal image, and the skin temperature of the region of interest of the human body is obtained based on the infrared thermal image. The skin temperature is used for thermal comfort classification of the human body.

2. The method for extracting human skin temperature as described in claim 1, characterized in that, The registration process for the infrared thermal image and the visible light image includes: Common key points of the human body were identified from the infrared thermal imaging and the visible light images, respectively. Based on the key points of the human body, the infrared thermal image and the visible light image are registered.

3. The method for extracting human skin temperature as described in claim 1, characterized in that, Identifying regions of interest in the human body from the visible light image includes: Regions of interest (ROIs) for the face, hands, arms, legs, and neck are identified from the visible light image, and these regions are designated as human body ROIs.

4. The method for extracting human skin temperature as described in claim 1, characterized in that, Based on the infrared thermal imaging, the skin temperature of the region of interest in the human body is obtained, including: Statistically analyze the pixel values ​​of the region of interest of the human body on the infrared thermal image; The skin temperature of the human body region of interest is obtained based on the pixel values ​​of the region of interest.

5. A method for classifying thermal comfort, characterized in that, include: Based on the human skin temperature extraction method as described in claim 1, the skin temperature of the human body region of interest is obtained; Human body parameter information and environmental parameters of the human body are obtained, and the skin temperature and environmental parameters are classified based on the human body parameter information to obtain classification results. The classification results are used to characterize the thermal comfort level of the human body under the influence of the skin temperature and environmental parameters.

6. The thermal comfort classification method as described in claim 5, characterized in that, Based on the human body parameter information, the skin temperature and environmental parameters are classified to obtain classification results, including: The XGBoost model is applied to the human body parameter information, the skin temperature, and the environmental parameters to obtain the classification results output by the XGBoost model.

7. The thermal comfort classification method as described in claim 5, characterized in that, The skin temperature, the human body parameter information, and the environmental parameter are the filtered skin temperature, human body parameter information, and environmental parameter, respectively; the filtered index is the thermal comfort contribution rate.

8. A human skin temperature extraction device, characterized in that, The device comprises the following components: The registration module is used to acquire infrared thermal images and visible light images of the human body, and to perform registration processing on the infrared thermal images and the visible light images. The region of interest (ROI) identification module is used to identify the region of interest of the human body from the visible light image; The temperature extraction module is used to map the region of interest of the human body onto the registered infrared thermal image, and to obtain the skin temperature of the region of interest of the human body based on the infrared thermal image. The skin temperature is used for thermal comfort classification of the human body.

9. A terminal device, characterized in that, The terminal device includes a memory, a processor, and a human skin temperature extraction program stored in the memory and executable on the processor. When the processor executes the human skin temperature extraction program, it implements the steps of the human skin temperature extraction method as described in any one of claims 1-4. Alternatively, the terminal device includes a memory, a processor, and a thermal comfort classification program stored in the memory and executable on the processor, wherein when the processor executes the thermal comfort classification program, it implements the steps of the thermal comfort classification method as described in any one of claims 5-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a human skin temperature extraction program, which, when executed by a processor, implements the steps of the human skin temperature extraction method as described in any one of claims 1-4. Alternatively, the computer-readable storage medium stores a thermal comfort classification program, which, when executed by a processor, implements the steps of the thermal comfort classification method as described in any one of claims 5-7.

Images