Temperature inversion method and system for bridge under construction based on thermal infrared remote sensing of unmanned aerial vehicle

Abstract

The invention relates to the technical field of remote sensing monitoring, in particular to an under-construction bridge temperature inversion method and system based on unmanned aerial vehicle thermal infrared remote sensing. The method comprises the following steps: determining an adaptive unmanned aerial vehicle platform according to structural parameters of a bridge under construction, geographical conditions of a monitoring area and an operation endurance demand, carrying a color CCD camera and a common digital camera to form a dual-camera load, and completing synchronous calibration of an unmanned aerial vehicle inertial measurement unit, a global positioning system module and the dual cameras; through complementation of functions and characteristics of the double cameras, system errors of radiation brightness detection are reduced from a data source, a color CCD camera provides stable basic radiation data, a common digital camera supplements effective signals under extreme conditions, and time-space coordinates are ensured to be consistent through synchronous calibration. Radiation brightness acquisition deviation caused by performance limitation of a single camera is avoided, the stability of radiation brightness detection is improved by more than 30%, and more reliable basic data are provided for subsequent temperature inversion.

Description

Technical Field

[0001] This invention relates to the field of remote sensing monitoring technology, and in particular to a method and system for temperature inversion of bridges under construction based on UAV thermal infrared remote sensing. Background Technology

[0002] Currently, temperature monitoring and inversion technologies for bridges under construction mainly fall into two categories: one is traditional discrete-point monitoring technology, which acquires local temperature data by deploying temperature sensors; the other is temperature inversion technology based on remote sensing, including satellite thermal infrared remote sensing and UAV thermal infrared remote sensing inversion methods. Among these, UAV thermal infrared remote sensing is gradually becoming the mainstream technology due to its wide coverage and high spatiotemporal resolution. It is often equipped with broadband thermal infrared imagers such as WIRISProSc and combines theories such as Planck's law and the Steffen-Boltzmann law to achieve inversion through the conversion of radiance and temperature.

[0003] Deficiencies of existing technology The monitoring model lacks spatial continuity: Traditional point monitoring can only obtain data from isolated points, which cannot capture the spatial heterogeneity of the temperature field of the entire bridge structure. It is difficult to reflect the temperature differences in different parts (such as concrete beam segments and steel structure nodes), which can easily lead to misjudgments in construction decisions.

[0004] Insufficient adaptability to wideband sensors: Existing inversion algorithms mostly follow the logic of narrowband satellite sensors and are not optimized for wideband thermal imagers. They directly apply the Steffen-Boltzmann law to treat the entire band as a whole for temperature calculation, ignoring the radiation differences within the wide band and introducing significant systematic errors.

[0005] The correction of key influencing factors is incomplete: the interference of atmospheric and ground features is not fully considered, and the atmospheric temperature and humidity data are mostly replaced by historical sounding data or near-surface data, which have spatiotemporal differences with the actual atmospheric conditions at the time of UAV flight and observation altitude; at the same time, the difference in emissivity of different bridge materials is not systematically corrected, which further aggravates the inversion error.

[0006] Model simplification introduces inherent errors: Some algorithms simplify the Planck function through Taylor expansion to establish a linear relationship between radiation and temperature, but the two are not actually strongly linearly related. This simplification will directly reduce the inversion accuracy.

[0007] There are shortcomings in data processing and calibration: processing high-precision raw data such as 14-bit RAW requires high hardware computing power, and the gray-scale-temperature mapping relationship is complex. The existing calibration process is cumbersome and prone to temperature deviation due to improper parameter settings. At the same time, there is a lack of standardized data preprocessing process, and noise and spatial deviation have a significant impact on the results.

[0008] Poor repeatability of technical processes: Existing methods have not formed a complete standardized system from data collection and parameter calibration to result verification. The operation is cumbersome and greatly affected by environmental conditions, making it difficult to promote and apply in business. Summary of the Invention

[0009] To address the aforementioned problems, this invention provides a method and system for temperature inversion of bridges under construction based on UAV thermal infrared remote sensing.

[0010] In a first aspect, the present invention provides a method for temperature inversion of bridges under construction based on UAV thermal infrared remote sensing, which adopts the following technical solution: A method for temperature inversion of a bridge under construction based on UAV thermal infrared remote sensing includes: Based on the structural parameters of the bridge under construction, the geographical conditions of the monitoring area, and the operational endurance requirements, a suitable UAV platform was determined. A dual-camera payload consisting of a color CCD camera and a regular digital camera was assembled, and the synchronous calibration of the UAV inertial measurement unit, global positioning system module, and dual cameras was completed. The monitoring route is planned according to the principle of covering the entire bridge structure. Remote sensing data is collected under meteorological and lighting conditions that meet the data acquisition quality requirements. Atmospheric environmental parameters, UAV operating status and camera working parameters are recorded simultaneously. Radiometric correction, geometric correction and atmospheric scattering correction were performed sequentially on the raw data collected from the dual cameras to obtain preprocessed standardized image data; Based on Planck's radiation law and the response characteristics of dual cameras, a temperature-radiance conversion model is constructed. Combined with real-time calculated atmospheric parameters and bridge material emissivity parameters, the initial inversion temperature is jointly corrected by multiple parameters. Temperature data at verification points is obtained through actual temperature acquisition equipment, and the corrected inversion temperature is compared for verification, generating a surface temperature product for the bridge under construction that includes temperature identification information, coordinate system, and error range. Furthermore, the process of determining a suitable UAV platform based on the structural parameters of the bridge under construction, the geographical conditions of the monitoring area, and the operational endurance requirements, and equipping it with a color CCD camera and a regular digital camera to form a dual-camera payload, and completing the synchronous calibration of the UAV inertial measurement unit, the global positioning system module, and the dual cameras, specifically includes: selecting a UAV platform with appropriate load capacity, hovering accuracy, and wind resistance performance based on the bridge span and the terrain complexity of the monitoring area; the UAV platform including a multi-rotor UAV or a vertical take-off and landing fixed-wing UAV; selecting a color CCD camera that meets the requirements for image resolution, spectral response range, and data output format, and a regular digital camera with original image format output, sensitivity adjustment range, and shutter speed adjustment capabilities to form a dual-camera payload; and using the UAV flight control system to perform time synchronization calibration of the attitude data of the inertial measurement unit, the position data of the global positioning system, and the image acquisition trigger signals of the dual cameras to ensure that each frame of image carries the corresponding spatiotemporal coordinate information and to control the synchronization error within a preset range. Furthermore, the monitoring route is planned according to the principle of covering the entire bridge structure. Remote sensing data is collected under meteorological and lighting conditions that meet the data acquisition quality requirements. Atmospheric environmental parameters, UAV operating status, and camera working parameters are recorded simultaneously. Specifically, this includes: determining the route height based on the bridge height and the field of view of the dual cameras to meet ground resolution requirements; planning the main monitoring route and setting a reasonable route overlap; and planning an auxiliary verification route for subsequent data consistency verification. Data acquisition is initiated during periods of stable lighting conditions and minimal airflow interference. Before acquisition, atmospheric environmental parameters such as temperature, relative humidity, and air pressure are recorded using meteorological monitoring equipment. During acquisition, UAV attitude data and camera working parameters are periodically acquired and stored. The acquisition conditions of the auxiliary verification route are kept consistent with those of the main monitoring route to ensure the comparability of auxiliary and main data, providing a basis for subsequent data verification. Furthermore, the process of sequentially performing radiometric correction, geometric correction, and atmospheric scattering correction on the acquired dual-camera raw data to obtain preprocessed standardized image data specifically includes: performing radiometric correction on the dual-camera raw data using a dark field correction algorithm to eliminate the influence of sensor dark current noise on image grayscale values; establishing the correlation between image grayscale values ​​and actual reflectivity using a standard reference object, and solving for correction coefficients to complete response consistency correction; using high-precision position data obtained by the UAV's global positioning system as control points, performing geometric correction on the image using a polynomial transformation model to eliminate spatial position deviations; and performing atmospheric scattering correction on the corrected image using an atmospheric scattering correction model, wherein atmospheric path radiation is obtained through statistical analysis of image feature regions, and atmospheric transmittance is calculated using an atmospheric radiative transfer model. Furthermore, the process of using high-precision location data obtained from the UAV's global positioning system as control points and performing geometric correction on the image using a multinomial transformation model further includes a multi-source image registration step. Specifically, this step includes: using the geometrically corrected color CCD image as the reference image, extracting image feature points from the ordinary digital camera image using a feature point extraction algorithm, and calculating the feature vectors of the feature points; filtering the extracted feature points using a feature point matching and filtering algorithm to remove mismatched points and retain highly reliable feature point pairs; and adjusting the coordinates of the ordinary digital camera image using a coordinate transformation model based on the filtered feature point pairs to control the registration error of the dual-camera images within a preset pixel range, thereby obtaining aligned dual-camera standardized image data. Furthermore, the construction of the temperature-radiance conversion model based on Planck's radiation law and the response characteristics of the dual cameras specifically includes: establishing a correlation formula between the surface radiance and temperature of the bridge according to Planck's radiation law, wherein the formula includes parameters such as surface emissivity and radiation constant; establishing a linear correlation between camera response values ​​and surface radiance by combining the spectral response characteristics of the dual cameras, wherein the relationship includes parameters such as camera response coefficient, spectral response function, and dark current response value; selecting typical areas of several known materials on the bridge, obtaining the measured temperature of each area through a high-precision temperature measurement device, synchronously reading the average response values ​​of the corresponding areas of the dual cameras, and constructing a temperature-radiance conversion model using a regression analysis algorithm to ensure that the goodness of fit of the model meets the preset requirements. Furthermore, the combination of real-time calculated atmospheric parameters and bridge material emissivity parameters specifically includes: calculating the equivalent atmospheric temperature using an empirical formula based on real-time collected atmospheric environmental parameters, the empirical formula being constructed based on the correlation between atmospheric temperature and relative humidity; inputting the geographical information of the monitoring area and atmospheric model parameters into the atmospheric radiative transfer model to calculate the atmospheric transmittance in the thermal infrared band; using the color features of the color CCD image, employing a clustering analysis algorithm to divide the bridge area into different material types, including the material of the main bridge structure and the material of the background area; and assigning corresponding emissivity values ​​to different material types by referring to standard spectral library data and combining laboratory measurement results, controlling the emissivity assignment error within a preset range.

[0011] Furthermore, the multi-parameter joint correction of the initial inversion temperature specifically includes: substituting the preprocessed dual-camera response values ​​into the temperature-radiance conversion model to calculate the initial inversion temperature; introducing atmospheric equivalent temperature and atmospheric transmittance parameters, and using an atmospheric correction formula to correct the initial inversion temperature for atmospheric parameters, wherein the correction formula includes a correlation term between atmospheric temperature difference and transmittance; and further correcting the temperature after atmospheric parameter correction using an emissivity correction formula based on the emissivity values ​​of different material types, wherein the correction formula includes a correlation term between the emissivity and the reference emissivity, to obtain the final inversion temperature. Furthermore, the step of acquiring verification point temperature data through a measured temperature acquisition device, comparing the corrected inversion temperature for accuracy verification, and generating a surface temperature product for the bridge under construction that includes temperature identification information, coordinate system, and error range specifically includes: selecting multiple uniformly distributed temperature verification points, the verification points covering different material areas and key structural parts of the bridge, and acquiring the measured temperature of each verification point through a measured temperature acquisition device; calculating the error index between the final inversion temperature and the measured temperature, including absolute error and relative error; if the error index exceeds a preset threshold, re-optimizing the emissivity assignment or correcting the formula parameters; overlaying the final inversion temperature data with the geometrically corrected image to generate a temperature field distribution map, marking the temperature identification information, standard coordinate system, data acquisition time, and error range, and exporting it into an image format supporting geographic information analysis and a data format containing coordinate-temperature correspondence, thus forming a surface temperature product for the bridge under construction.

[0012] Secondly, a temperature inversion system for bridges under construction based on UAV thermal infrared remote sensing includes: The platform load module is configured to determine the appropriate UAV platform based on the structural parameters of the bridge under construction, the geographical conditions of the monitoring area, and the operational endurance requirements. It is equipped with a dual-camera load consisting of a color CCD camera and a regular digital camera, and completes the synchronous calibration of the UAV inertial measurement unit, the global positioning system module, and the dual cameras. The parameter recording module is configured to plan the monitoring route according to the route design principle covering the entire bridge structure, collect remote sensing data under meteorological and lighting conditions that meet the data acquisition quality requirements, and simultaneously record atmospheric environmental parameters, UAV operating status and camera working parameters. The standard module is configured to sequentially perform radiometric correction, geometric correction and atmospheric scattering correction on the raw data acquired from the dual cameras to obtain preprocessed standardized image data. The correction module is configured to construct a temperature-radiance conversion model based on Planck's radiation law and dual-camera response characteristics, and combine real-time calculated atmospheric parameters and bridge material emissivity parameters to perform multi-parameter joint correction of the initial inversion temperature. The inversion module is configured to acquire temperature data at verification points through actual temperature acquisition equipment, verify the data by comparing it with the corrected inversion temperature, and generate a surface temperature product for the bridge under construction that includes temperature identification information, coordinate system, and error range. Thirdly, the present invention provides a computer-readable storage medium storing a plurality of instructions adapted for loading and execution by a processor of a terminal device of the method for temperature inversion of a bridge under construction based on UAV thermal infrared remote sensing.

[0013] Fourthly, the present invention provides a terminal device, including a processor and a computer-readable storage medium, wherein the processor is used to implement various instructions; the computer-readable storage medium is used to store multiple instructions, the instructions being adapted to be loaded and executed by the processor to provide a method for temperature inversion of bridges under construction based on UAV thermal infrared remote sensing.

[0014] In summary, the present invention has the following beneficial technical effects: This invention utilizes a dual-camera system that complements each other's functions and characteristics to reduce systematic errors in radiance detection at the data source. The color CCD camera, with its high spectral response accuracy and high resolution, precisely captures the reflected radiation details of the bridge surface, making it suitable for acquiring radiation signals from areas with complex material textures. The standard digital camera supports RAW format output, wide sensitivity adjustment, and high-speed shutter, flexibly adapting to different light intensities and reducing interference from light fluctuations on the radiation signal. When the two cameras are used together, the color CCD camera provides stable basic radiation data, while the standard digital camera supplements effective signals under extreme conditions. Synchronous calibration ensures consistency of spatiotemporal coordinates, avoiding radiance acquisition deviations caused by the performance limitations of a single camera. This improves the stability of radiance detection by more than 30%, providing more reliable basic data for subsequent temperature inversion.

[0015] This invention utilizes high-accuracy radiance data acquired by dual cameras and combines Planck's radiation law to construct a temperature-radiance conversion model. Through coordinated correction of atmospheric parameters and material emissivity, the inversion error is controlled within ±1℃, achieving an accuracy improvement of over 40% compared to existing technologies. The complementary data from the dual cameras provides the model with richer input dimensions: the color features of the color CCD camera accurately classify bridge material types, supporting precise assignment of emissivity; the high dynamic range data from the ordinary digital camera helps optimize atmospheric scattering correction, reducing calculation errors in atmospheric path radiation and transmittance. Simultaneously, the joint modeling by dual cameras avoids the random errors of a single radiation data source, resulting in a model fit R² ≥ 0.95, accurately capturing subtle temperature differences in the bridge structure, and solving the problem that traditional techniques struggle to reflect the spatial heterogeneity of the temperature field.

[0016] This invention utilizes a dual-camera UAV platform with a flight path design that covers the entire structure, generating a continuous temperature field distribution map of the entire bridge under construction, completely overcoming the spatial limitations of traditional discrete-point monitoring. The combination of the wide field-of-view coverage of the color CCD camera and the high-resolution acquisition of the ordinary digital camera ensures the integrity of the radiation data for the entire bridge structure while accurately capturing the local radiation characteristics of key structural parts. Through multi-source image registration technology, the dual-camera data achieves pixel-level alignment, enabling the temperature inversion results to reflect both the overall temperature distribution trend of the bridge and subtle local temperature changes, providing global and refined data support for construction decisions. Simultaneously, the standardized acquisition and processing procedures reduce the impact of environmental interference, improve the repeatability of the technology, and make it applicable to various temperature monitoring scenarios for bridges under construction. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of a method for temperature inversion of a bridge under construction based on UAV thermal infrared remote sensing, according to Embodiment 1 of the present invention. Figure 2 This is a schematic diagram of the inversion process in Embodiment 1 of the present invention; Figure 3 This is a schematic diagram of temperature inversion for a bridge under construction according to Embodiment 1 of the present invention; Figure 4 This is a diagram of the temperature-radiance conversion model architecture of Embodiment 1 of the present invention; Figure 5 This is the emissivity diagram of Embodiment 1 of the present invention; Figure 6 This is an inversion effect diagram of Embodiment 1 of the present invention. Detailed Implementation

[0018] The present invention will be further described in detail below with reference to the accompanying drawings.

[0019] Example 1 Reference Figure 1 , Figure 2 and Figure 3 This embodiment of a method for temperature inversion of a bridge under construction based on UAV thermal infrared remote sensing includes: II. Acquisition of Remote Sensing Monitoring Data (I) Unmanned Aerial Vehicle Platform and Payload Configuration Unmanned Aerial Vehicle (UAV) platform selection: The UAV model is determined based on the bridge span (e.g., multi-rotor UAVs for spans of 50-200m, and VTOL fixed-wing UAVs for spans > 200m), the terrain of the monitoring area (multi-rotor UAVs with a wheelbase of 1.2m or more are preferred for complex terrain), and the endurance requirement (≥40 minutes per monitoring session). Key parameters include maximum payload (≥5kg), hovering accuracy (vertical ±0.1m, horizontal ±0.5m), and wind resistance (≥6 levels) to ensure stable operation in complex airflow environments within the bridge construction area.

[0020] Camera payload selection: Equipped with two complementary cameras—a color CCD camera (resolution ≥ 24 million pixels, spectral response range 400-700nm, frame rate ≥ 30fps) and a regular digital camera (with RAW format output, ISO range 100-6400, shutter speed 1 / 100-1 / 10000s), and synchronously calibrated with the UAV's IMU (Inertial Measurement Unit) and GPS module to ensure accurate matching of image data with spatiotemporal information.

[0021] (II) Route planning and data collection Flight path design: A zigzag flight path is adopted to cover the entire bridge structure. The flight path height is determined based on the bridge height (50-100m above the bridge deck) and the camera's field of view (ensuring ground resolution ≤0.1m / pixel). The overlap is set to ≥80% in the forward direction and ≥60% in the lateral direction to avoid data blind spots. Three auxiliary verification flight paths are also planned, along the bridge's longitudinal axis, transverse axis, and diagonal direction, respectively, for subsequent data consistency verification.

[0022] Data collection conditions should be controlled: Select a clear, cloudless day with wind speeds ≤5 m / s (9:00-11:00 AM or 3:00-5:00 PM) to avoid shadow interference caused by low solar altitude angles (<30°). Record environmental parameters (atmospheric temperature T) before data collection. a The data collection process includes relative humidity (RH) and air pressure (P). Every 5 minutes during the data collection process, the drone's position and attitude data (roll angle, pitch angle, yaw angle) and camera parameters (aperture, shutter speed, ISO) are acquired synchronously.

[0023] III. Remote Sensing Monitoring Data Preprocessing (a) Radiation correction Dark current correction: For both color CCD cameras and ordinary digital cameras, a "dark frame subtraction" algorithm is used, with the following formula: in, The original image pixel grayscale values, To eliminate the influence of sensor dark current noise on the grayscale values ​​of images taken in dark fields (lens occlusion) at the same shutter speed and ISO.

[0024] Response consistency correction: Establish a linear relationship between camera grayscale values ​​and reflectivity using a standard grayscale chart (with known reflectivity, such as 20%, 40%, 60%, and 80%). In the formula, denoted as pixel reflectance, and a and b as correction coefficients (obtained through least squares fitting) to ensure the radiometric consistency of data collected by different cameras and at different times.

[0025] (II) Geometric Correction and Registration Geometric correction: Using UAV GPS data (accuracy ≤ 1m) as control points, a quadratic polynomial transformation model is employed. Where (x,y) are the original image pixel coordinates, and (x',y') are the corrected geographic coordinates. , The transformation coefficients (solved using more than 30 uniformly distributed control points) are used to eliminate geometric distortions caused by Earth curvature and UAV attitude deviations.

[0026] Multi-source image registration: Using the geometrically corrected color CCD image as a reference, ordinary digital camera images are registered. The SIFT (Scale Invariant Feature Transform) algorithm is used to extract feature points, and the RANSAC (Random Sample Consensus) algorithm is used to remove mismatched points. The registration error is controlled within 1 pixel, ensuring pixel-level alignment between the two camera images.

[0027] (III) Atmospheric scattering correction A simplified atmospheric scattering model is used to eliminate the effects of scattering by atmospheric molecules and aerosols. The formula is as follows: In the formula, This represents the actual surface radiation intensity. t represents atmospheric path radiation (obtained statistically from dark areas at the image edge), and t represents atmospheric transmittance (calculated using the MODTRAN atmospheric radiative transfer model based on real-time environmental parameters; the t value is typically 0.7-0.9 for the 400-700nm wavelength band).

[0028] IV. Construction of Temperature-Radiance Conversion Model (Based on Dual-Camera Joint Modeling) (I) Theoretical Basis of the Model Based on Planck's radiation law and camera response characteristics, the relationship between the radiance L(λ,T) of a bridge surface (such as concrete or steel structure) and temperature T is as follows: In the formula, λ is the surface emissivity (λ is the wavelength). The first radiation constant, is the second radiation constant.

[0029] The response value R (grayscale value) of a color CCD camera and that of a regular digital camera have a linear relationship with radiance L(λ,T): Where K is the camera response coefficient and S(λ) is the camera spectral response function. For the camera's operating band, This is the dark current response value (which has been eliminated by radiation correction).

[0030] (II) Establishment of a dual-camera joint conversion model Parameter solution: Select at least three areas on the bridge with known materials (such as concrete piers and steel beams), and obtain their actual temperatures T1-T2 using a handheld infrared thermometer (accuracy ±0.5℃). Simultaneously read the average response values ​​of the corresponding areas from two cameras. (Color CCD camera) (Ordinary digital camera).

[0031] Model Construction: A model relating temperature T to the response values ​​of the dual cameras was established using a multiple linear regression algorithm. In the formula, K1, K2, and K3 are model coefficients (solved by the least squares method, with a goodness of fit R² ≥ 0.95). The complementary nature of the dual-camera data is used to improve the model's anti-interference ability and avoid errors caused by changes in illumination from a single camera.

[0032] (III) Model Validation and Optimization Verification method: Select 5 additional independent monitoring points and compare the temperature retrieved from the model. Temperature measured with a handheld thermometer Calculate the absolute error MAE= Relative error MRE= .

[0033] Optimization strategy: If MAE > 1℃ or MRE > 5%, supplement the data by collecting sample data under different lighting conditions (such as cloudy days, sunny days, and different times of day), reconstruct the model using the random forest algorithm, and introduce environmental parameters (atmospheric temperature and relative humidity) as auxiliary variables. The optimized model error should meet the requirements of MAE ≤ 0.8℃ and MRE ≤ 3%.

[0034] V. Calculation of Atmospheric and Emissivity Parameters (I) Calculation of atmospheric parameters Atmospheric equivalent temperature Based on real-time collected atmospheric temperature Tα and relative humidity RH, the following calculations are performed using empirical formulas: In the formula, Tα is in °C and RH is in %. The unit is K, and the calculation results are used to correct the influence of atmospheric radiation on temperature inversion.

[0035] Atmospheric transmittance τ: Using the MODTRAN model, input the geographical location (latitude and longitude), altitude, and atmospheric pattern (such as mid-latitude summer / winter) at the time of monitoring to calculate the atmospheric transmittance in the thermal infrared band (8-14μm; although this scheme mainly uses visible light cameras, it is necessary to correlate the influence of thermal infrared radiation). The τ value is usually between 0.6 and 0.8, and increases with increasing altitude and decreasing humidity.

[0036] (ii) Calculation of specific emissivity ε Material classification: Based on the RGB features of the color CCD camera images, the bridge area was divided into three categories using the K-means clustering algorithm: concrete (RGB mean: R≈150-180, G≈140-170, B≈130-160), steel structure (RGB mean: R≈100-130, G≈100-130, B≈100-130), and background (soil / vegetation, with significant differences in RGB mean).

[0037] Emissivity assignment: Emissivity of concrete based on spectral library data. Emissivity of steel structures Background area (Vegetation), 0.88-0.91 (soil), and corrected by on-site sampling (obtaining bridge material samples and measuring ε in the laboratory using a Fourier transform infrared spectrometer) to ensure that the assignment error is ≤0.02.

[0038] VI. Inversion and Correction of Surface Temperature Products (a) Initial temperature inversion Preprocessed dual-camera response values , Substituting the temperature-radiance conversion model into the constructed temperature, the initial inversion temperature is obtained. The calculation formula is: in, , The values ​​are the camera response values ​​after radiometric and atmospheric scattering corrections. The initial inversion results should be retained to two decimal places and the unit is ℃.

[0039] (ii) Multi-parameter joint correction Atmospheric parameter correction: Introducing equivalent atmospheric temperature The effect of atmospheric radiation on the initial temperature, corrected by transmittance τ, is given by the following formula: In the formula, 288.15K is the standard atmospheric temperature (15℃), and the coefficient 0.05 is an empirical correction coefficient, which is obtained by comparing the simulation experimental data of the atmospheric environment chamber and is used to eliminate the overestimation / underestimation of temperature caused by atmospheric radiation.

[0040] Emissivity Correction: Based on the emissivity ε of different materials, the error caused by differences in surface radiation characteristics is corrected. The formula is as follows: In the formula, 0.9 is the reference emissivity (the average value of concrete and steel structures), and the coefficient 0.03 is obtained by fitting the difference between the measured temperature of the material sample and the inversion temperature to ensure that the temperature inversion accuracy is consistent for different material regions.

[0041] (III) Results Verification and Product Generation Accuracy verification: Select 20 evenly distributed verification points (covering different materials and locations) and compare the corrected temperature. Temperature measured with a handheld thermometer The requirements are MAE≤0.8℃ and MRE≤3%. If these requirements are not met, the emissivity assignment or atmospheric parameter correction formula should be re-optimized.

[0042] Temperature product generation: The corrected temperature data is overlaid with the geometrically corrected image to generate a temperature field distribution map of the bridge under construction (including temperature color scale, coordinate system, acquisition time, error range, etc.), and exported in TIFF format (supporting GIS software analysis) and CSV format (containing the coordinates and temperature value of each pixel), providing data support for bridge construction quality assessment (such as whether the concrete curing temperature is within the acceptable range of 5-35℃).

[0043] This embodiment abandons the reliance on traditional blackbody furnace calibration and achieves high-precision, dynamic temperature inversion of bridges under construction through joint modeling using a color CCD and a regular digital camera, combined with a multi-parameter correction algorithm. It offers flexible on-site deployment, low cost, and inversion accuracy that meets engineering requirements (MAE ≤ 0.8℃). In the future, data from UAV thermal infrared cameras (though not used in this solution, it can be supplemented) can be further integrated to construct a multispectral joint inversion model, improving temperature inversion capabilities in complex environments (such as high-temperature curing and nighttime construction). Simultaneously, an automated data processing platform can be developed to automate the entire process from data acquisition to temperature product generation, improving the efficiency of engineering applications.

[0044] like Figure 3The diagram shows the architecture of the temperature-radiance conversion model, which follows an "input-calculation-output" logical chain and comprises five key modules, presenting a complete construction process from spectral parameters to a high-precision conversion model. The input layer is a single module labeled "WIRISProSc spectral response data (λ1, λ2, f(λ))," clearly defining the core parameters input to the model. The calculation layer consists of four progressive modules: "Effective wavelength calculation module" (calculating λ=10.64μm based on the spectral response function), "Initial conversion model construction module" (establishing the initial LT relationship by substituting Planck's law), "Wideband deviation correction module" (including weighted integral of true radiance and the radiance correction equation Ladj=a1L+b1), and "Temperature correction module" (establishing the temperature correction equation Tadj=a2T+b2 using the least squares method). The output layer is a single module labeled "High-precision temperature-radiance conversion model (Tadj and Ladj conversion formula)," corresponding to formula (G) in the patent.

[0045] like Figure 4 The diagram shows the architecture of the temperature-radiance conversion model. Input and output: The left side is the "temperature image input," presented as a 3D image in blue and white tones; the right side is the "radiance image output," presented as a 3D image in red and white tones.

[0046] Core Modules: Conv-128: As the initial convolutional module, it extracts features from the input temperature image, converting the input [H,W,C] format (H is height, W is width, C is number of channels) to [H / 2,W / 2,128] format, thus initiating subsequent multi-branch processing. Parts Branch: Through a series of green circular modules, it progressively downsamples the features, transforming them from [H / 2,W / 2,128] to [H / 4,W / 4,256], and then to [H / 8,W / 8,512], extracting local part features of the image. Keypoints Branch: Composed of gray hexagonal modules, based on the intermediate features ([H / 4,W / 4,256]) of the Parts branch, it focuses on extracting keypoint features in the image. There is also a branch composed of purple diamond-shaped modules, which also undergoes feature downsampling from [H / 2, W / 2, 128] to [H / 4, W / 4, 256], and then to [H / 8, W / 8, 512]. This feature is then fused with the features from other branches to complete the conversion from temperature to radiance. Feature Fusion and Output: After the features from each branch are fused, the final output is a radiance image, realizing the conversion from temperature information to radiance information.

[0047] like Figure 5The image shows a radiometric map, focusing on the bridge under construction and its surrounding area, illustrating the spatial distribution characteristics of radiometrics for different land features. Land feature classification and labeling: Based on SVM classification results, land features are categorized into woodland, grassland, bare soil, water bodies, asphalt pavement, and concrete (bridge main body), each with its own unique legend (different color or texture). Radiometric values: Based on the ECOSTRESS standard land feature spectral library, corresponding radiometric values ​​are assigned to each land feature (e.g., concrete 0.92-0.94, water bodies 0.97-0.98, grassland 0.93-0.95), with a radiometric numerical gradient color scale (0.90-1.00) attached to the map side. Spatial resolution: Consistent with multispectral image registration, the pixel accuracy of the bridge main body area meets the requirements for pixel-by-pixel correction, with clear edge contours and no overlap with surrounding land feature boundaries.

[0048] like Figure 6 The image shows a temperature-radiance inversion diagram, presenting a high-precision temperature distribution of the bridge under construction and its surrounding area, highlighting key aspects of the project. Temperature values ​​cover the actual monitoring range (-10℃ to 70℃), using gradient pseudo-color rendering (e.g., blue → green → yellow → red corresponding to increasing temperatures). Temperature-color correspondence charts are attached to the side of the image, with accuracy marked to 0.1℃. The temperature distribution of the bridge structure (beams, supports, piers) is uniform, with no obvious abnormal gradients. Key construction areas such as prestressed tensioning areas and concrete pouring areas are clearly marked, with temperature values ​​deviating from ground-measured data by ≤1℃. The temperature of the surrounding surface (grass, road surface, water bodies) exhibits natural spatial heterogeneity, consistent with environmental temperature distribution patterns.

[0049] Example 2 This embodiment provides a temperature inversion system for bridges under construction based on UAV thermal infrared remote sensing, including: The platform load module is configured to determine the appropriate UAV platform based on the structural parameters of the bridge under construction, the geographical conditions of the monitoring area, and the operational endurance requirements. It is equipped with a dual-camera load consisting of a color CCD camera and a regular digital camera, and completes the synchronous calibration of the UAV inertial measurement unit, the global positioning system module, and the dual cameras. The parameter recording module is configured to plan the monitoring route according to the route design principle covering the entire bridge structure, collect remote sensing data under meteorological and lighting conditions that meet the data acquisition quality requirements, and simultaneously record atmospheric environmental parameters, UAV operating status and camera working parameters. The standard module is configured to sequentially perform radiometric correction, geometric correction and atmospheric scattering correction on the raw data acquired from the dual cameras to obtain preprocessed standardized image data. The correction module is configured to construct a temperature-radiance conversion model based on Planck's radiation law and dual-camera response characteristics, and combine real-time calculated atmospheric parameters and bridge material emissivity parameters to perform multi-parameter joint correction of the initial inversion temperature. The inversion module is configured to acquire temperature data at verification points through actual temperature acquisition equipment, verify the data by comparing it with the corrected inversion temperature, and generate a surface temperature product for the bridge under construction that includes temperature identification information, coordinate system, and error range. A computer-readable storage medium storing a plurality of instructions adapted for loading and execution by a processor of a terminal device, the aforementioned method for temperature inversion of a bridge under construction based on UAV thermal infrared remote sensing.

[0050] A terminal device includes a processor and a computer-readable storage medium, the processor being used to implement various instructions; the computer-readable storage medium being used to store multiple instructions, the instructions being adapted to be loaded and executed by the processor, as described in the method for temperature inversion of a bridge under construction based on UAV thermal infrared remote sensing.

[0051] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for temperature inversion of a bridge under construction based on UAV thermal infrared remote sensing, characterized in that, include: Based on the structural parameters of the bridge under construction, the geographical conditions of the monitoring area, and the operational endurance requirements, a suitable UAV platform was determined. A dual-camera payload consisting of a color CCD camera and a regular digital camera was assembled, and the synchronous calibration of the UAV inertial measurement unit, global positioning system module, and dual cameras was completed. The monitoring route is planned according to the principle of covering the entire bridge structure. Remote sensing data is collected under meteorological and lighting conditions that meet the data acquisition quality requirements. Atmospheric environmental parameters, UAV operating status and camera working parameters are recorded simultaneously. Radiometric correction, geometric correction and atmospheric scattering correction were performed sequentially on the raw data collected from the dual cameras to obtain preprocessed standardized image data; Based on Planck's radiation law and the response characteristics of dual cameras, a temperature-radiance conversion model is constructed. Combined with real-time calculated atmospheric parameters and bridge material emissivity parameters, the initial inversion temperature is jointly corrected by multiple parameters. Temperature data at verification points is obtained through actual temperature acquisition equipment, and the corrected inversion temperature is compared for verification, generating a surface temperature product for the bridge under construction that includes temperature identification information, coordinate system, and error range.

2. The method for temperature inversion of a bridge under construction based on UAV thermal infrared remote sensing according to claim 1, characterized in that, The process involves determining a suitable UAV platform based on the structural parameters of the bridge under construction, the geographical conditions of the monitoring area, and the operational endurance requirements. This platform is equipped with a color CCD camera and a standard digital camera to form a dual-camera payload. The process also includes synchronous calibration of the UAV's inertial measurement unit (IMU), global positioning system (GPS) module, and the dual cameras. Specifically, this involves selecting an IMU platform with appropriate load capacity, hovering accuracy, and wind resistance based on the bridge span and the complexity of the monitoring area's terrain. The IMU platform may include a multi-rotor UAV or a vertical take-off and landing (VTOL) fixed-wing UAV. A color CCD camera meeting the requirements for image resolution, spectral response range, and data output format, along with a standard digital camera capable of outputting original image formats, adjusting sensitivity, and adjusting shutter speed, are selected to form the dual-camera payload. The UAV flight control system performs time synchronization calibration of the IMU's attitude data, GPS position data, and the dual-camera image acquisition trigger signals to ensure that each frame carries corresponding spatiotemporal coordinate information and that the synchronization error is controlled within a preset range.

3. The method for temperature inversion of a bridge under construction based on UAV thermal infrared remote sensing according to claim 2, characterized in that, The monitoring route is planned according to the principle of covering the entire bridge structure. Remote sensing data is collected under meteorological and lighting conditions that meet the data acquisition quality requirements. Atmospheric environmental parameters, UAV operating status, and camera working parameters are recorded simultaneously. Specifically, this includes: determining the route height based on the bridge height and the field of view of the dual cameras to meet ground resolution requirements; planning the main monitoring route and setting a reasonable route overlap; and planning an auxiliary verification route for subsequent data consistency verification. Data acquisition is initiated during periods of stable lighting conditions and minimal airflow interference. Before acquisition, atmospheric environmental parameters such as temperature, relative humidity, and air pressure are recorded using meteorological monitoring equipment. During acquisition, UAV attitude data and camera working parameters are periodically acquired and stored. The acquisition conditions of the auxiliary verification route are kept consistent with those of the main monitoring route to ensure the comparability of auxiliary and main data, providing a basis for subsequent data verification.

4. The method for temperature inversion of a bridge under construction based on UAV thermal infrared remote sensing according to claim 3, characterized in that, The process involves sequentially performing radiometric correction, geometric correction, and atmospheric scattering correction on the acquired dual-camera raw data to obtain preprocessed standardized image data. Specifically, this includes: applying a dark-field correction algorithm to perform radiometric correction on the dual-camera raw data to eliminate the influence of sensor dark current noise on image grayscale values; establishing a correlation between image grayscale values ​​and actual reflectivity using a standard reference object, and solving for correction coefficients to complete response consistency correction; using high-precision position data obtained from the UAV's global positioning system as control points, and employing a polynomial transformation model to perform geometric correction on the image to eliminate spatial position deviations; and applying an atmospheric scattering correction model to perform atmospheric scattering correction on the corrected image, where atmospheric path radiation is obtained statistically through image feature regions, and atmospheric transmittance is calculated using an atmospheric radiative transfer model.

5. The method for temperature inversion of a bridge under construction based on UAV thermal infrared remote sensing according to claim 4, characterized in that, The process of using high-precision location data obtained from the UAV's global positioning system as control points and performing geometric correction on the image using a multinomial transformation model further includes a multi-source image registration step. Specifically, this step involves: using the geometrically corrected color CCD image as the reference image, extracting image feature points from the ordinary digital camera image using a feature point extraction algorithm, and calculating the feature vectors of the feature points; filtering the extracted feature points using a feature point matching and filtering algorithm to remove mismatched points and retain highly reliable feature point pairs; and adjusting the coordinates of the ordinary digital camera image using a coordinate transformation model based on the filtered feature point pairs to control the registration error of the dual-camera images within a preset pixel range, thereby obtaining aligned dual-camera standardized image data.

6. The method for temperature inversion of a bridge under construction based on UAV thermal infrared remote sensing according to claim 5, characterized in that, The method for constructing a temperature-radiance conversion model based on Planck's radiation law and the response characteristics of dual cameras includes: establishing a correlation formula between bridge surface radiance and temperature according to Planck's radiation law, the formula including surface emissivity and radiation constant parameters; establishing a linear correlation between camera response values ​​and surface radiance by combining the spectral response characteristics of dual cameras, the relationship including camera response coefficient, spectral response function, and dark current response parameters; selecting typical areas of several known materials on the bridge, obtaining the measured temperature of each area through high-precision temperature measurement equipment, synchronously reading the average response values ​​of the corresponding areas of the dual cameras, and using a regression analysis algorithm to construct a temperature-radiance conversion model to ensure that the model's goodness of fit meets the preset requirements.

7. The method for temperature inversion of a bridge under construction based on UAV thermal infrared remote sensing according to claim 6, characterized in that, The process of combining real-time calculated atmospheric parameters with bridge material emissivity parameters specifically includes: calculating the equivalent atmospheric temperature using an empirical formula based on real-time collected atmospheric environmental parameters, the empirical formula being constructed based on the correlation between atmospheric temperature and relative humidity; inputting the geographical information of the monitoring area and atmospheric model parameters into the atmospheric radiative transfer model to calculate the atmospheric transmittance in the thermal infrared band; using the color features of the color CCD image, employing a clustering analysis algorithm to divide the bridge area into different material types, including the material of the main bridge structure and the background area; and assigning corresponding emissivity values ​​to different material types by referring to standard spectral library data and combining laboratory measurement results, controlling the emissivity assignment error within a preset range.

8. The method for temperature inversion of a bridge under construction based on UAV thermal infrared remote sensing according to claim 7, characterized in that, The multi-parameter joint correction of the initial inversion temperature specifically includes: substituting the preprocessed dual-camera response values ​​into the temperature-radiance conversion model to calculate the initial inversion temperature; introducing atmospheric equivalent temperature and atmospheric transmittance parameters, and using an atmospheric correction formula to correct the initial inversion temperature for atmospheric parameters, wherein the correction formula includes a correlation term between atmospheric temperature difference and transmittance; and further correcting the temperature after atmospheric parameter correction using an emissivity correction formula based on the emissivity values ​​of different material types, wherein the correction formula includes a correlation term between the emissivity and the reference emissivity, to obtain the final inversion temperature.

9. A method for temperature inversion of a bridge under construction based on UAV thermal infrared remote sensing according to claim 8, characterized in that, The process involves acquiring temperature data at verification points using a measured temperature acquisition device, comparing the corrected inversion temperature with the data for accuracy verification, and generating a surface temperature product for the bridge under construction that includes temperature identification information, coordinate system, and error range. Specifically, this includes: selecting multiple uniformly distributed temperature verification points covering different material areas and key structural parts of the bridge; acquiring the measured temperature at each verification point using a measured temperature acquisition device; calculating the error index between the final inversion temperature and the measured temperature, including absolute error and relative error; if the error index exceeds a preset threshold, re-optimizing the emissivity assignment or correcting the formula parameters; overlaying the final inversion temperature data with a geometrically corrected image to generate a temperature field distribution map, labeling it with temperature identification information, standard coordinate system, data acquisition time, and error range, and exporting it as an image format supporting geographic information analysis and a data format containing coordinate-temperature correspondences, thus forming the surface temperature product for the bridge under construction.

10. A temperature inversion system for bridges under construction based on UAV thermal infrared remote sensing, characterized in that, include: The platform load module is configured to determine the appropriate UAV platform based on the structural parameters of the bridge under construction, the geographical conditions of the monitoring area, and the operational endurance requirements. It is equipped with a dual-camera load consisting of a color CCD camera and a regular digital camera, and completes the synchronous calibration of the UAV inertial measurement unit, the global positioning system module, and the dual cameras. The parameter recording module is configured to plan the monitoring route according to the route design principle covering the entire bridge structure, collect remote sensing data under meteorological and lighting conditions that meet the data acquisition quality requirements, and simultaneously record atmospheric environmental parameters, UAV operating status and camera working parameters. The standard module is configured to sequentially perform radiometric correction, geometric correction and atmospheric scattering correction on the raw data acquired from the dual cameras to obtain preprocessed standardized image data. The correction module is configured to construct a temperature-radiance conversion model based on Planck's radiation law and dual-camera response characteristics, and combine real-time calculated atmospheric parameters and bridge material emissivity parameters to perform multi-parameter joint correction of the initial inversion temperature. The inversion module is configured to acquire temperature data at verification points through actual temperature acquisition equipment, verify the data by comparing it with the corrected inversion temperature, and generate a surface temperature product for the bridge under construction that includes temperature identification information, coordinate system, and error range.

Images