Terahertz polarization non-inductive body temperature measurement method and system

Abstract

The application discloses a terahertz polarized non-inductive body temperature measurement method and system, and belongs to the field of microwave measurement. The application solves the problem that the existing method cannot realize quantitative and accurate inversion of the surface temperature of a large-area or whole human body under the condition that a thin layer of dry non-metallic covering such as clothes and dressing is reserved. A multi-line polarized brightness temperature image is obtained, and actual measured brightness temperature is obtained through preprocessing, a measured Stokes parameter is constructed to determine a polarization angle, a linear polarization degree and a local azimuth angle; an effective emission rate model is established according to the local azimuth angle, taking a local incident angle as a variable, and then predicted brightness temperature is estimated and predicted Stokes parameters and predicted linear polarization degrees are constructed; brightness temperature residual and polarization degree residual are calculated respectively, a double residual joint objective function is constructed, and an optimal incident angle is solved through optimization; and the surface temperature of the human body is inverted by using the predicted brightness temperature under the optimal incident angle, and a temperature distribution diagram is generated. The application is mainly applied to non-contact body temperature measurement.

Description

Technical Field

[0001] This invention belongs to the field of microwave measurement. Background Technology

[0002] With the increasing demand for applications such as non-contact medical screening, wound and burn follow-up, postoperative recovery monitoring, and infection or fever risk assessment, human body surface temperature field imaging has gradually become an important technical direction in the field of medical thermal assessment. Local body surface temperature is influenced by multiple factors, including blood perfusion, tissue metabolism, vascular regulation, environmental heat exchange, and clothing or dressing coverage. Compared to single-point temperature readings, spatially resolved body surface temperature maps can provide thermal distribution information for different anatomical regions such as the head and neck, trunk, and limbs. Existing body surface temperature measurement technologies mainly include contact thermometers, thermocouples, thermistors, and infrared thermal imaging. Among these, contact temperature measurement has high local accuracy but is difficult to achieve whole-body or large-area thermal field assessment; infrared thermal imaging has advantages such as non-contact, fast imaging speed, and high spatial resolution, and has been widely used for measuring the thermal distribution of exposed skin surfaces. However, in scenarios such as when clothing or dressings are in place, or in low-exposure examinations, infrared thermal imaging typically only obtains the temperature of the outer surface of the covering layer and cannot directly reflect the temperature distribution of the human skin surface beneath the covering layer.

[0003] Terahertz polarization imaging can receive the natural thermal radiation signals of the human body and has a certain penetration observation capability under thin, dry, and non-metallic covering conditions, thus providing a new physical approach for human thermal assessment under clothing or dressing conditions. This type of technology does not require active target illumination and has potential advantages such as non-contact, low exposure, large-area imaging, and suitability for repeated measurements. However, current terahertz polarization human imaging research is mostly limited to qualitative observation levels such as brightness-temperature contrast display or security check-style target enhancement, and has not yet developed a quantitative body surface temperature inversion method for medical thermal assessment. For complex curved targets such as the whole human body, a single polarization brightness temperature or fixed emissivity assumption is difficult to distinguish between temperature changes and geometric emissivity changes, and is prone to significant errors in areas with rapid curvature changes such as the torso edges and limbs. The quantitative and accurate inversion of body surface temperature is poor, which limits the transformation of terahertz polarization imaging from qualitative brightness temperature maps to quantitative body surface temperature maps.

[0004] Therefore, there is an urgent need to design a terahertz polarized non-contact body temperature measurement method and system for large-area thermal assessment of the human body under clothing or dressing conditions, so as to convert terahertz brightness temperature images into quantitative body surface temperature distribution maps, and provide a new technical solution for non-contact, low-exposure, and clothing-covered whole-body or large-area body surface thermal assessment. Summary of the Invention

[0005] The purpose of this invention is to solve the problem that existing methods cannot achieve quantitative and accurate inversion of the surface temperature of a large area or the whole body when thin, dry non-metallic coverings such as clothing and dressings are retained. This invention provides a terahertz polarized non-sensory body temperature measurement method and system.

[0006] Terahertz polarization non-sensory body temperature measurement methods include:

[0007] S1. Acquire the brightness temperature image of the human body under terahertz multi-linear polarization direction, and preprocess it to obtain the measured brightness temperature of each pixel in the region of the human body under each linear polarization direction. ; These are the angles of the linear polarization directions in the ground reference coordinate system. ;

[0008] S2. Construct Stokes parameters under measured conditions using the measured brightness temperature of each pixel in the multi-line polarization direction; determine the polarization angle based on the measured Stokes parameters. and measured linear polarization According to the polarization angle Determine the local azimuth angle ;

[0009] S3, based on local azimuth angle Calculate the local incident angle of each pixel within the area of ​​the human body being measured. At that time, the effective emissivity of each linear polarization direction on the local surface element of the tested human body ; As variables, ;

[0010] S4. Based on effective emissivity Estimate the local incident angle of the human body Predicted brightness temperature in each linear polarization direction ;

[0011] S5. Based on the local incident angle of each human body The predicted brightness temperature of the lower multi-line polarization direction is used to construct the local incident angle of the current pixel in the human body. The Stokes parameters under the given prediction conditions are used to determine the local incident angle of the human body. Predicted linear polarization ;

[0012] S6. Based on the predicted brightness temperature under each linear polarization direction. The deviation between the measured brightness temperature and the actual brightness temperature is used to construct the local incident angle of the human body. Multipolar brightness temperature residual Based on the predicted linear polarization Compared with measured linear polarization Inter-period deviation, constructing local incident angle of the human body polarization residual ;

[0013] S7, according to and Construct a joint objective function with two residuals and optimize it. Select the local incident angle of the human body that minimizes the joint objective function with two residuals. As the optimal angle of incidence ;

[0014] S8. Based on the optimal incident angle Predicted brightness temperature in each linear polarization direction The surface temperature of each pixel within the area where the human body is located is inverted to obtain a surface temperature distribution map of the human body.

[0015] Preferably, in step S7, the joint objective function of the two residuals... ; This represents the linear polarization consistency constraint weight.

[0016] Preferably, the implementation of the Stokes parameters under actual experimental conditions is as follows: ;

[0017] in, , and The Stokes parameters are determined under actual conditions. Indicates the measured brightness temperature intensity. This represents the measured brightness temperature difference between the 0° linear polarization direction and the 90° linear polarization direction. This indicates the measured brightness temperature polarization difference between the 45° linear polarization direction and the 135° linear polarization direction.

[0018] Preferably, the polarization angle is determined based on the Stokes parameters under measured conditions. The implementation method is as follows: ;

[0019] According to polarization angle Determine the local azimuth angle The implementation method is as follows:

[0020] .

[0021] Preferably, in step S3, ;in,

[0022] , ;

[0023] The emissivity of the human body surface under local horizontal polarization direction. This represents the emissivity of the human body surface under local vertical polarization direction. The complex permittivity is... and These represent the surface reflectivity under the local horizontal polarization direction and the local vertical polarization direction, respectively.

[0024] Preferably, in step S4, ;

[0025] in,

[0026] ;

[0027] ;

[0028] ;

[0029] ;

[0030] The local angle of incidence on the human body is... The estimated human body surface temperature at that time This represents the environmental equivalent incident brightness temperature; The emissivity of the human body surface under local horizontal polarization direction. This represents the emissivity of the human body surface under local vertical polarization direction. The complex permittivity is... and These represent the surface reflectivity under the local horizontal polarization direction and the local vertical polarization direction, respectively.

[0031] Preferably, in step S5, the Stokes parameters under the prediction conditions are determined by... , and Composition, and ;

[0032] in,

[0033] , , ;

[0034] Indicates the local angle of incidence on the human body The predicted brightness temperature difference between the 0° linear polarization direction and the 90° linear polarization direction. Indicates the local angle of incidence on the human body The difference in predicted brightness temperature polarization between the 45° and 135° linear polarization directions. Indicates the local angle of incidence on the human body The predicted brightness temperature intensity.

[0035] Preferably, in step S6, .

[0036] Preferably, in step S8, ;in, This represents the equivalent incident brightness temperature of the environment. This indicates that each pixel within the area of ​​the human body being measured is at the optimal incident angle. hour, Effective emissivity of the linear polarization direction on a local surface element of the tested human body. Indicates the optimal angle of incidence. Predicted brightness temperature for each polarization direction.

[0037] A terahertz polarized non-contact body temperature measurement system includes a storage device, a processor, and a computer program stored in the storage device and executable on the processor. The processor executes the computer program to implement the terahertz polarized non-contact body temperature measurement method described above.

[0038] The beneficial effects of this invention are:

[0039] This invention defines a complete inversion process, constructing Stokes parameters using multi-line polarization brightness temperature data, then introducing local incident angle variables, and solving for the optimal incident angle through joint optimization using dual residuals (brightness temperature residual and polarizability residual), ultimately inverting the surface temperature. This method couples polarization information with geometric parameters for optimization, unlike fixed incident angle or single polarization methods. This invention achieves non-contact, passive inversion of human body surface temperature, independent of active radiation sources, and adaptable to complex clothing or obstructed scenarios; through joint optimization, it reduces the sensitivity of model errors to incident angle estimation, improves the spatial consistency and accuracy of temperature inversion, and finally outputs a whole-body temperature distribution map.

[0040] The specific innovations are as follows:

[0041] (1) A quantitative thermal imaging approach of “using multi-polarization passive brightness temperature to retrieve human body surface temperature map” is proposed. Unlike existing infrared thermal imaging, which mainly measures the temperature of exposed skin or the outer surface of the covering layer, and unlike traditional terahertz imaging, which is mainly used for brightness temperature comparison display, this invention uses the natural radiation signal of the human body to obtain multi-polarization brightness temperature images under thin, dry, non-metallic clothing or dressing conditions, and converts them into a large area or whole body surface temperature distribution, providing a new quantitative imaging path for medical thermal assessment under low exposure, non-contact, and covering layer retention conditions.

[0042] (2) Introducing a pixel-level joint inversion mechanism with polarization physical constraints. This invention combines Stokes parameters, linear polarizability, polarization angle, emissivity model, polarization rotation, and equivalent ambient brightness temperature. It utilizes the response differences between multiple linear polarization channels to constrain the local incident angle, local azimuth angle, and effective emissivity of the human body, thereby reducing the influence of human curvature, viewing angle changes, and regional emissivity differences on temperature inversion, and realizing the conversion from "passive brightness temperature image" to "physical surface temperature map".

[0043] In summary, this invention provides a terahertz polarized non-contact body temperature measurement method and system, which can reconstruct the temperature of a large area or the entire human body surface without active irradiation. It is particularly suitable for non-contact thermal assessment under conditions where thin, dry, non-metallic coverings are retained. Even with clothing wear and tear, sub-Celsius temperature measurement accuracy is achieved, demonstrating significantly enhanced engineering applicability and potential for widespread application in thermal assessment of the human body under covered conditions. Attached Figure Description

[0044] Figure 1 This is a flowchart of the terahertz polarization non-sensory body temperature measurement method described in this invention.

[0045] Figure 2 This is a schematic diagram of the structure of a terahertz multipolar imaging system in the prior art;

[0046] Figure 3 This is a schematic diagram of the measured brightness temperature of each pixel in the area of ​​the human body under test under each polarization direction, obtained after preprocessing the brightness temperature image under multiple polarization directions; where K represents the Kelvin temperature unit.

[0047] Figure 4 This diagram illustrates the construction results of the Stokes parameters and polarization characteristics involved in the experiment. (a) shows the brightness temperature intensity term constructed from brightness temperature data under multiple polarization directions. Distribution diagram; (b) shows the brightness temperature difference between the 0° linear polarization direction and the 90° linear polarization direction. Distribution diagram; (c) shows the brightness temperature difference between the 45° linear polarization direction and the 135° linear polarization direction. Distribution diagrams; (d) is the linear polarizability (DoLP) distribution diagram; (e) is the polarization angle (AoP) distribution diagram;

[0048] Figure 5This diagram illustrates the pixel-level joint constraint inversion process used in the experiment. (a) shows the multi-polarized brightness-temperature residual field and closed-form temperature estimation trajectory under different candidate incident angles. The Closed-form trajectory represents the closed-form temperature estimation trajectory, the Reference point represents the reference point, and the Optimal point represents the optimal point. (b) shows the relationship between the candidate incident angle and the estimated temperature. (c) shows the brightness-temperature residual term. Polarization residual and joint objective function A graph showing how the incident angle changes.

[0049] Figure 6 To verify the temperature distribution maps of multiple subjects and multiple perspectives of the human body surface involved in the experiment;

[0050] Figure 7 To verify the consistency assessment of the error statistics of the surface temperature inversion results involved in the experiment with the reference temperature; (a) is a schematic diagram of the statistical matrix of the mean absolute error (MAE) of ROI and the mean deviation error (MBE) of the anatomical region for the four subjects under front-view, back-view and side-view conditions; (b) is a comparison diagram of the consistency between the inverted temperature and the reference temperature under front-view, back-view and side-view conditions; Slope represents the linear fitting slope, and CCC represents the consistency correlation coefficient. Detailed Implementation

[0051] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0053] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but this is not intended to limit the scope of the invention.

[0054] Specific Implementation Method 1: Combination Figure 1 This embodiment describes a terahertz polarization non-sensory body temperature measurement method, which includes:

[0055] S1. Acquire the brightness temperature image of the human body under terahertz multi-linear polarization direction, and preprocess it to obtain the measured brightness temperature of each pixel in the region of the human body under each linear polarization direction. ; These are the angles of the linear polarization directions in the ground reference coordinate system. ;

[0056] S2. Construct Stokes parameters under measured conditions using the measured brightness temperature of each pixel in the multi-line polarization direction; determine the polarization angle based on the measured Stokes parameters. and measured linear polarization According to the polarization angle Determine the local azimuth angle ;

[0057] S3, based on local azimuth angle Calculate the local incident angle of each pixel within the area of ​​the human body being measured. At that time, the effective emissivity of each linear polarization direction on the local surface element of the tested human body ; As variables, ;

[0058] S4. Based on effective emissivity Estimate the local incident angle of the human body Predicted brightness temperature in each linear polarization direction ;

[0059] S5. Based on the local incident angle of each human body The predicted brightness temperature of the lower multi-line polarization direction is used to construct the local incident angle of the current pixel in the human body. The Stokes parameters under the given prediction conditions are used to determine the local incident angle of the human body. Predicted linear polarization ;

[0060] S6. Based on the predicted brightness temperature under each linear polarization direction. The deviation between the measured brightness temperature and the actual brightness temperature is used to construct the local incident angle of the human body. Multipolar brightness temperature residual ;

[0061] Based on the predicted linear polarization Compared with measured linear polarization Inter-period deviation, constructing local incident angle of the human body polarization residual ;

[0062] S7, according to and Constructing a joint objective function with two residuals The solution is optimized, and the local incident angle of the human body is selected to minimize the joint objective function of the two residuals. As the optimal angle of incidence ; that is: ; Let be the set of incident angles. Several discrete angles within the range of 0° to 80° can be selected, or they can be determined based on the system's field of view, the location of the human body region, the degree of image edge severity, or the prior geometric range, as expressed as: ,in, to They represent the first to the second. Local angle of incidence on an individual body;

[0063] S8. Based on the optimal incident angle Predicted brightness temperature in each linear polarization direction The surface temperature of each pixel within the area where the human body is located is inverted to obtain a surface temperature distribution map of the human body.

[0064] The core of this invention lies in its ability to utilize polarization geometry information from terahertz multipolar brightness temperature data, without relying on the removal of clothing or dressings. By constructing Stokes parameters and linear polarizability under measured and predicted conditions, an effective emissivity model with the local incident angle as a variable is established. Furthermore, a dual objective function of brightness temperature residual and polarizability residual is constructed. The optimal incident angle is obtained through optimization, and finally, the predicted brightness temperature at this incident angle is used to invert the human body surface temperature. This process does not require pre-setting the dielectric constant or thickness parameters of the covering; instead, it transforms the differential influence of the covering on each polarization component into a polarization consistency constraint. This allows polarization distortion to be automatically corrected during the optimization process, thereby achieving quantitative inversion of the surface temperature of a large area or even the entire human body while retaining a thin, dry, non-metallic covering, and outputting a temperature distribution map.

[0065] See Figure 2 A schematic diagram of a terahertz multipolarization imaging system in the prior art is given. This imaging system can acquire the brightness temperature image of the human body under the terahertz multipolarization direction. The system includes a reflector 2, a near-field lens 3, a multipolar receiving antenna 4, a terahertz radiometer channel 5, a turntable for switching radiometer polarization 6, a sliding rail scanning device 7, and a data acquisition and processing device 8.

[0066] The human body being tested 1 is used to radiate outward millimeter-wave or Asia-Pacific Hertz natural radiation signals generated by its own thermal state.

[0067] The reflector 2 and the near-field lens 3 are used to form a passive imaging optical path to map the natural radiation signal of the observation area where the human body 1 is located to the receiving end.

[0068] The receiving antenna 4 is used to receive the natural radiation signals of the human body 1 under different fixed linear polarization directions in the ground reference coordinate system.

[0069] The terahertz radiometer channel 5 is connected to the multipolar receiving antenna 4 and is used to convert the received natural radiation signal into brightness temperature under the corresponding polarization direction.

[0070] The turntable 6 used for switching radiometer polarization is used to change the fixed receiving polarization direction in the ground reference coordinate system in order to obtain brightness temperature images of four fixed linear polarization directions: 0°, 45°, 90° and 135°.

[0071] The imaging scanning device 7 is used to realize two-dimensional scanning imaging of the observation area and obtain a two-dimensional multi-polarization brightness temperature image of the human body 1 being tested.

[0072] The data acquisition and processing device 8 is used to acquire, store and process the brightness temperature images of each fixed receiving polarization channel (each linear polarization direction) under the ground reference coordinate system, and output the human body surface temperature distribution map.

[0073] Furthermore, preprocessing the brightness temperature images under each linear polarization direction to obtain the measured brightness temperature of each pixel within the area of ​​the tested human body under each linear polarization direction can be achieved using existing technologies. This can be achieved by sequentially performing radiometric calibration, channel registration, human body region mask extraction, and effective pixel selection on the brightness temperature images under each linear polarization direction. Specifically, radiometric calibration is performed on the brightness temperature images of each polarization channel to convert the natural radiation signal radiated outward from the tested human body 1 into a brightness temperature image. Then, spatial registration is performed on the images of different polarization channels so that the same pixel position corresponds to the same observation area on the surface of the tested human body. Afterward, a mask of the human body region is extracted based on brightness temperature intensity, human body contour, effective imaging range, or auxiliary visible light image to remove background areas and invalid pixels. Finally, effective pixels are selected based on channel noise, edge-mixed pixels, low signal-to-noise ratio areas, and non-human body areas to obtain the measured brightness temperature under multiple polarization directions for subsequent inversion.

[0074] Furthermore, the implementation of the Stokes parameters under actual experimental conditions is as follows:

[0075] ;

[0076] in, , and The Stokes parameters are determined under actual conditions. Indicates the measured brightness temperature intensity. This represents the measured brightness temperature difference between the 0° linear polarization direction and the 90° linear polarization direction. This indicates the measured brightness temperature polarization difference between the 45° linear polarization direction and the 135° linear polarization direction. These represent the measured brightness temperatures at four linear polarization directions: 0°, 45°, 90°, and 135°, respectively, in the ground reference coordinate system.

[0077] , and Used to characterize the intensity and differences in brightness temperature of multipolarization. and Used to provide polarization constraint information related to local surface geometry and polarization rotation of the human body.

[0078] See Figure 4 The polarization angle is determined based on the Stokes parameters under measured conditions. The implementation method is as follows:

[0079] ;

[0080] According to polarization angle Determine the local azimuth angle The implementation method is as follows:

[0081] .

[0082] This preferred method provides analytical inversion formulas for polarization angle and local azimuth angle, converting measured Stokes parameters into spatial orientation information, realizing quantitative conversion from polarization measurement to local geometric attitude, providing key input for local incident angle and emissivity modeling, and avoiding reliance on external 3D scanning equipment.

[0083] Furthermore, in step S3, ;

[0084] in,

[0085] ;

[0086] ;

[0087] The emissivity of the human body surface under local horizontal polarization direction. This represents the emissivity of the human body surface in the local vertical polarization direction. The complex permittivity is... and These represent the surface reflectivity under the local horizontal polarization direction and the local vertical polarization direction, respectively.

[0088] This preferred method establishes a functional relationship between effective emissivity and local incident angle, complex permittivity, and polarization direction (horizontal / vertical), employing a Fresnel reflectivity model. This provides a clear physical basis for emissivity calculation, enabling it to adapt to different local incident angle variations and improving the physical reliability of multi-polarization brightness temperature prediction.

[0089] Furthermore, in step S4, ;in,

[0090] ;

[0091] ;

[0092] ;

[0093] ;

[0094] The local angle of incidence on the human body is... The estimated human body surface temperature at that time This represents the environmental equivalent incident brightness temperature; The emissivity of the human body surface under local horizontal polarization direction. This represents the emissivity of the human body surface in the local vertical polarization direction. The complex permittivity is... and These represent the surface reflectivity under the local horizontal polarization direction and the local vertical polarization direction, respectively.

[0095] This preferred method provides an expression for predicting brightness temperature, which correlates emissivity, ambient incident brightness temperature, and estimated surface temperature, realizing a calculation path from surface temperature assumptions to brightness temperatures of each polarization. It provides forward modeling tools for residual construction and iterative optimization, and explicitly compensates for ambient brightness temperature, reducing environmental radiation interference.

[0096] Furthermore, in step S5, the Stokes parameters under the prediction conditions are determined by... , and Composition, and ;

[0097] in,

[0098] ;

[0099] ;

[0100] ;

[0101] Indicates the local angle of incidence on the human body The predicted brightness temperature difference between the 0° linear polarization direction and the 90° linear polarization direction. Indicates the local angle of incidence on the human body The difference in predicted brightness temperature polarization between the 45° and 135° linear polarization directions. Indicates the local angle of incidence on the human body The predicted brightness temperature intensity.

[0102] The preferred approach defines a method for constructing Stokes parameters under predicted conditions, maintaining the same format as the measured Stokes parameters. This ensures that the predicted and measured polarization characteristics are structurally comparable, facilitating the calculation of subsequent polarizability consistency residuals and avoiding errors caused by inconsistencies in dimensions or definitions.

[0103] Furthermore, in step S8, ;

[0104] in, This represents the equivalent incident brightness temperature of the environment. This indicates that each pixel within the area of ​​the human body being measured is at the optimal incident angle. hour, Effective emissivity of the linear polarization direction on a local surface element of the tested human body. Indicates the optimal angle of incidence. Predicted brightness temperature for each polarization direction.

[0105] Specific implementation method two: Terahertz polarization non-contact body temperature measurement system, including a storage device, a processor, and a computer program stored in the storage device and executable on the processor, wherein the processor executes the computer program to implement the terahertz polarization non-contact body temperature measurement method as described above.

[0106] Verification experiment:

[0107] A W-band passive terahertz linearly polarized radiometer is preferred for image acquisition, operating in the 75 GHz to 110 GHz frequency band with a center frequency around 94 GHz. During acquisition, brightness temperature images of four linearly polarized channels (0°, 45°, 90°, and 135°) are obtained sequentially by rotating the linearly polarized receiving unit or switching the polarization direction, and are denoted as follows: .in, These correspond to 0° and 90° linear polarization channels, respectively. In whole-body measurement scenarios, multi-polarization brightness temperature images of the human body are acquired from the front, back, and side views, respectively. In local measurement scenarios, local multi-polarization brightness temperature imaging can be performed on wound areas, dressing-covered areas, local limb areas, or other target areas.

[0108] Subsequently, the measured multi-polarization brightness temperature images were preprocessed, see [reference needed]. Figure 3 The preprocessing includes radiometer calibration, channel registration, human body region mask extraction, and effective pixel selection.

[0109] The radiometer calibration involves converting the radiometer output signal into a brightness temperature value to obtain a physically meaningful multipolarized brightness temperature image.

[0110] The channel registration is to spatially align brightness temperature images under different linear polarization directions so that the same pixel position corresponds to the same observation area on the surface of the human body being tested.

[0111] The human body region mask extraction is performed by extracting the human body region based on brightness temperature intensity, human body contour, effective imaging range or optional auxiliary image, and removing the background region.

[0112] The effective pixel selection is based on channel noise, edge-mixed pixels, low signal-to-noise ratio regions, and non-human regions to select effective pixels that can be used for temperature inversion.

[0113] After the above preprocessing, four-polarization brightness temperature data are obtained for subsequent body surface temperature inversion, including ,like Figure 3 As shown, Figure 3 In this context, K represents the Kelvin temperature unit.

[0114] Then, Stokes parameters and polarization characteristics are constructed based on the four-polarization brightness temperature data, see [link to documentation]. Figure 4 First, using Calculate the brightness temperature intensity term and linear polarization difference term .in, Common intensity components used to characterize multipolar brightness temperature and Used to characterize the difference components between different linear polarization directions. Subsequently, according to The linear polarizability DoLP and polarization angle AoP are further calculated. DoLP is used to characterize the intensity of the linear polarization response, and AoP is used to characterize the principal direction of linear polarization. Figure 4 In this context, K represents the Kelvin temperature unit; DoLP × 10 -3 DoLP values ​​are expressed in 10. -3 The order of magnitude is indicated by deg, which represents the angle unit "degree".

[0115] Because the human body surface is a complex curved surface, the local incident angle, local azimuth angle, and polarization rotation relationship vary at different locations. Therefore, four-polarization brightness temperature is not just a display result of temperature intensity, but also contains information related to the local surface geometry and polarization response of the human body. By using Stokes parameters, DoLP, and AoP, local geometric and polarization rotation constraints can be provided for subsequent pixel-level temperature inversion, reducing the errors generated by single-polarization brightness temperature or fixed emissivity models in high-curvature regions such as the edges of the human body, limbs, and the lateral edges of the torso.

[0116] Subsequently, pixel-level joint constraint inversion was performed based on the Fresnel emissivity model, polarization rotation model, and environmental reflection brightness temperature model of the human body surface. (See [link to relevant documentation]). Figure 5For each effective pixel, the Fresnel emissivity under local horizontal and local vertical polarization directions is first calculated based on the equivalent complex permittivity of the human body surface and the candidate incident angle. Then, the effective emissivity corresponding to different linear polarization channels is calculated by combining the local azimuth angle and the receiving polarization direction. Finally, the radiative transfer relationship between the single pixel brightness temperature and the human body surface temperature is established based on the multi-polarization brightness temperature, effective emissivity, and equivalent ambient incident brightness temperature.

[0117] In the example, for each candidate incident angle, the corresponding surface temperature estimate is first obtained by least squares estimation; then, a joint objective function is constructed based on the multi-polarization brightness temperature residual and polarizability residual; finally, the candidate incident angle that minimizes the joint objective function is selected as the optimal local incident angle of the pixel, and the temperature estimate corresponding to the candidate angle is used as the final inverted human body surface temperature of the pixel.

[0118] Figure 5 In the diagram, (a) represents the brightness temperature residual field and closed-loop temperature estimation trajectory under different candidate incident angles; (b) represents the relationship curve between the candidate incident angle and the estimated temperature, used to represent the surface temperature estimation results corresponding to different candidate incident angles; and (c) represents the brightness temperature residual term. Polarization residuals and joint objective function The curve showing the change with the incident angle is used to determine the optimal candidate incident angle and the corresponding pixel-level surface temperature inversion result. Through this process, local incident angle, local azimuth angle, effective emissivity, and human body surface temperature can be constrained simultaneously at the pixel level.

[0119] Finally, based on the temperature inversion results of all valid pixels, a two-dimensional or multi-view surface temperature distribution map of the tested human body is reconstructed. (See [link]). Figure 6 Samples 1 to 4 represent the surface temperature inversion results for different tested objects; each sample includes two-dimensional surface temperature distribution maps under front-view, back-view, and side-view conditions. This map illustrates that the present invention can generate continuous human surface temperature distribution results under different tested objects and different observation angles. For multi-polarized brightness temperature images acquired under front-view, back-view, and side-view conditions, corresponding human surface temperature distribution maps can be generated. For whole-body measurement scenarios, temperature indicators for areas such as the head and neck, trunk, upper limbs, and lower limbs can be further extracted based on the division of human anatomical regions; for local measurement scenarios, corresponding regional temperature indicators can be extracted based on wound areas, dressing-covered areas, or target local areas. The regional temperature indicators include the regional average temperature, regional median temperature, regional maximum temperature, regional temperature offset, the proportion of area exceeding a preset surface temperature threshold, and the temperature change between different time points.

[0120] To verify the feasibility of this invention, this example uses four adult subjects wearing thin, dry, non-metallic everyday indoor clothing. Four-polarized brightness temperature images were acquired under forward, rear, and side-view conditions, respectively, and the invention was then applied. Figure 1 Process inlet surface temperature map generation. Figure 6 The surface temperature inversion results of four subjects under forward, rear, and side-view conditions are presented. It can be seen that the present invention can generate continuous human surface temperature distribution maps under different subjects and different observation angles, while preserving the differences in heat distribution in anatomical regions such as the head and neck, trunk, and limbs.

[0121] Furthermore, to evaluate the accuracy of the regional temperature measurement in this invention, the retrieved human body surface temperature map is compared with the thermocouple contact reference temperature at the region of interest (ROI) level. See [link to relevant documentation]. Figure 7 . Figure 7 In the figure, (a) shows the statistical matrix of the mean absolute error (MAE) and mean deviation error (MBE) of the anatomical region for four subjects under forward, rear, and side-view conditions at the ROI level; MAE represents the average absolute error between the retrieved temperature and the reference temperature, and MBE represents the average deviation direction and magnitude of the retrieved temperature relative to the reference temperature. (b) is a consistency comparison diagram between the retrieved temperature and the reference temperature under forward, rear, and side-view conditions; Slope represents the linear fitting slope, and CCC represents the consistency correlation coefficient. This figure is used to illustrate that the human surface temperature results generated by this invention can be consistent with the regional reference temperature and can be used for ROI-level human surface thermal state assessment.

[0122] Specifically, multiple Regions of Interest (ROIs) for the head and neck, torso, and limbs are set under both front-view and back-view conditions, and the same ROIs are set under side-view conditions. The effective pixel temperature within each ROI is averaged to obtain the inversion temperature of that ROI. This inversion temperature is then compared with the average value of repeated thermocouple contact readings at the corresponding location to calculate the ROI-level mean absolute error and region mean bias error. Here, ROI stands for Region of Interest; MAE stands for Mean Absolute Error; and MBE stands for Mean Bias Error. The reference temperature is the average value of contact temperature readings obtained by thermocouples at the corresponding ROI location, and the inversion temperature is the average surface temperature of the ROI obtained from the multi-polarized brightness temperature image.

[0123] In this example, four subjects were subjected to front-view, rear-view, and side-view imaging, respectively, and a total of 12 sets of subject-view data were obtained. Each subject contributed 85 view-related ROIs, and the four subjects obtained a total of 340 sets of ROI-level technical comparisons. Figure 7In the figure, (a) represents the statistical matrix of the mean absolute error of the ROI and the mean deviation error of the anatomical region for four subjects under front-view, back-view, and side-view conditions; (b) represents the consistency comparison results between the inverted temperature and the reference temperature under front-view, back-view, and side-view conditions. The results show that the present invention can achieve sub-degree Celsius regional temperature measurement under multiple subjects and multiple observation angles. The error under front-view and back-view conditions is lower than that under side-view conditions. The relatively higher error under side-view conditions is mainly related to the narrower projection width, the higher proportion of pixels at the human body edge, the faster change in local curvature, and the mixed pixel effect.

[0124] In other examples, the acquisition method of the multi-polarization brightness temperature image is not limited to a rotating linear polarization receiving unit, but can also employ a multi-channel parallel polarization receiving structure, a polarization switching structure, an array imaging structure, a fast scanning structure, or a compressed sampling imaging structure; the imaging frequency band is not limited to the W-band in this embodiment, but can also be selected from other millimeter-wave or Asia-Pacific Hertz bands according to the covering conditions, spatial resolution requirements, and system hardware conditions; the human body area mask can be directly extracted from the brightness temperature image, or can be generated with the assistance of visible light images, depth images, or other auxiliary imaging results; the equivalent environmental incident brightness temperature can be approximated by indoor ambient temperature, or can be obtained through environmental radiation calibration or background modeling; the thermal assessment object can be the whole body area, or it can be a wound, burn, postoperative area, local area of ​​the limb, or dressing-covered area.

[0125] As can be seen from the above specific examples, this invention can utilize terahertz multi-polarization brightness temperature data to achieve quantitative inversion of human body surface temperature under conditions without active irradiation; through Stokes polarization features, Fresnel emissivity model, polarization rotation model, and pixel-level joint constraint inversion, traditional qualitative brightness temperature images can be converted into physically meaningful human body surface temperature distribution maps; at the same time, this method can output ROI-level temperature indicators and multi-view temperature distribution results, providing a technical basis for non-contact assessment of the thermal state of human body surfaces under clothing or dressing conditions.

[0126] While the invention has been described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the invention. Therefore, it should be understood that many modifications can be made to the exemplary embodiments, and other arrangements can be designed without departing from the spirit and scope of the invention as defined by the appended claims. It should be understood that different dependent claims and features described herein can be combined in ways different from those described in the original claims. It is also understood that features described in conjunction with individual embodiments can be used in other described embodiments.

Claims

1. A terahertz polarized non-sensory body temperature measurement method, characterized in that, include: S1. Acquire the brightness temperature image of the human body under terahertz multi-linear polarization direction, and preprocess it to obtain the measured brightness temperature of each pixel in the region of the human body under each linear polarization direction. ; These are the angles of the linear polarization directions in the ground reference coordinate system. ; S2. Construct Stokes parameters under measured conditions using the measured brightness temperature of each pixel in the multi-line polarization direction; determine the polarization angle based on the measured Stokes parameters. and measured linear polarization According to the polarization angle Determine the local azimuth angle ; S3, based on local azimuth angle Calculate the local incident angle of each pixel within the area of ​​the human body being measured. At that time, the effective emissivity of each linear polarization direction on the local surface element of the tested human body ; As variables, ; S4. Based on effective emissivity Estimate the local incident angle of the human body Predicted brightness temperature in each linear polarization direction ; S5. Based on the local incident angle of each human body The predicted brightness temperature of the lower multi-line polarization direction is used to construct the local incident angle of the current pixel on the human body. The Stokes parameters under the given prediction conditions are used to determine the local incident angle of the human body. Predicted linear polarization ; S6. Based on the predicted brightness temperature under each linear polarization direction. The deviation between the measured brightness temperature and the actual brightness temperature is used to construct the local incident angle of the human body. Multipolar brightness temperature residual Based on the predicted linear polarization Compared with measured linear polarization Inter-period deviation, constructing local incident angle of the human body polarization residual ; S7, according to and Construct a joint objective function with two residuals and optimize it. Select the local incident angle of the human body that minimizes the joint objective function with two residuals. As the optimal angle of incidence ; S8. Based on the optimal incident angle Predicted brightness temperature in each linear polarization direction The surface temperature of each pixel within the area where the human body is located is inverted to obtain a surface temperature distribution map of the human body.

2. The terahertz polarization non-sensory body temperature measurement method according to claim 1, characterized in that, In step S7, the joint objective function of the two residuals ; This represents the linear polarization consistency constraint weight.

3. The terahertz polarization non-sensory body temperature measurement method according to claim 1, characterized in that, The implementation of Stokes parameters under actual experimental conditions is as follows: ; in, , and The Stokes parameters are determined under actual conditions. Indicates the measured brightness temperature intensity. This represents the measured brightness temperature difference between the 0° linear polarization direction and the 90° linear polarization direction. This indicates the measured brightness temperature polarization difference between the 45° linear polarization direction and the 135° linear polarization direction.

4. The terahertz polarization non-sensory body temperature measurement method according to claim 3, characterized in that, Determine the polarization angle based on the Stokes parameters under measured conditions. The implementation method is as follows: ; According to polarization angle Determine the local azimuth angle The implementation method is as follows: 。 5. The terahertz polarization non-sensory body temperature measurement method according to claim 1, characterized in that, In step S3, ;in, ， ； The emissivity of the human body surface under local horizontal polarization direction. This represents the emissivity of the human body surface under local vertical polarization direction. The complex permittivity is... and These represent the surface reflectivity under the local horizontal polarization direction and the local vertical polarization direction, respectively.

6. The terahertz polarization non-sensory body temperature measurement method according to claim 1, characterized in that, In step S4, ; in, ； ； ； ； The local angle of incidence on the human body is... The estimated human body surface temperature at that time This represents the environmental equivalent incident brightness temperature; The emissivity of the human body surface under local horizontal polarization direction. This represents the emissivity of the human body surface under local vertical polarization direction. The complex permittivity is... and These represent the surface reflectivity under the local horizontal polarization direction and the local vertical polarization direction, respectively.

7. The terahertz polarization non-sensory body temperature measurement method according to claim 1, characterized in that, In step S5, the Stokes parameters under the prediction conditions are determined by... , and Composition, and ; in, ， ， ； Indicates the local angle of incidence on the human body The predicted brightness temperature difference between the 0° linear polarization direction and the 90° linear polarization direction. Indicates the local angle of incidence on the human body The difference in predicted brightness temperature polarization between the 45° and 135° linear polarization directions. Indicates the local angle of incidence on the human body The predicted brightness temperature intensity.

8. The terahertz polarization non-sensory body temperature measurement method according to claim 1, characterized in that, In step S6, .

9. The terahertz polarization non-sensory body temperature measurement method according to claim 1, characterized in that, In step S8, ;in, This represents the equivalent incident brightness temperature of the environment. This indicates that each pixel within the area of ​​the human body being measured is at the optimal incident angle. hour, Effective emissivity of the linear polarization direction on a local surface element of the tested human body. Indicates the optimal angle of incidence. Predicted brightness temperature for each polarization direction.

10. A terahertz polarized non-contact body temperature measurement system, comprising a storage device, a processor, and a computer program stored in the storage device and executable on the processor, characterized in that, The processor executes a computer program to implement the terahertz polarization non-sensory body temperature measurement method as described in any one of claims 1 to 9.

Images