Binocular infrared wide-area calorific value detection method based on distance and environmental background compensation

Abstract

The invention provides a binocular infrared wide-area calorific value detection method based on distance and environmental background compensation, which comprises the following steps: acquiring a fused infrared radiation image and a visible light binocular image, and introducing an atmospheric transmissivity calculation mechanism based on an infrared radiation transmission physical model, pixel-by-pixel compensation is carried out on infrared radiation attenuation of different distances and environment conditions; the infrared radiation image is subjected to background segmentation, an environment background area of a non-heat source target is extracted, an environment background radiation field covering a monitoring range is reconstructed, and a radiation correction value of the surface of the heat source target is calculated according to the atmospheric transmissivity, the environment background radiation field and a radiation value of a pixel corresponding to the heat source target in the infrared radiation image. Therefore, the actually measured temperature is obtained. According to the method, differentiated distance compensation and environment compensation can be carried out on wide-area and multi-target scenes, temperature measurement errors under complex environment conditions are effectively reduced, and the accuracy and stability of infrared heat value detection in wide-range monitoring application are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of wide-area thermal state monitoring, and in particular to a binocular infrared wide-area thermal value detection method based on distance and environmental background compensation. BACKGROUND

[0002] With the increasing demand for non-contact temperature detection in the fields of power inspection, industrial equipment monitoring and public safety, infrared temperature measurement technology is widely used in multi-target thermal state monitoring scenarios. However, under the condition of wide-area monitoring, the infrared radiation is easily affected by the environmental background radiation and the change of atmospheric conditions during propagation, resulting in deviation of the temperature measurement result from the true temperature of the target, and it is difficult to meet the application requirements of increasing temperature measurement accuracy and stability.

[0003] The existing infrared temperature measurement method is usually based on the radiation value of the target pixel in the infrared image to directly invert the temperature, or only through the introduction of fixed atmospheric compensation parameters for simple correction, which has certain applicability in the scene of close distance and relatively simple background. However, in the wide-area monitoring scene, the distance between the monitoring target and the infrared sensor is quite different, and the atmospheric absorption and scattering effect changes significantly with the distance, so a single compensation parameter cannot accurately describe the radiation attenuation at different spatial positions. At the same time, in complex environments, there are generally background heat sources and environmental reflected radiation, and the traditional method often lacks effective modeling of the spatial distribution of environmental background radiation, which is easy to misjudge the environmental radiation as target radiation, thereby reducing the temperature measurement accuracy, and even causing false alarm or missing report.

[0004] Therefore, there is an urgent need for a binocular infrared wide-area thermal value detection method based on distance and environmental background compensation, which can implement differentiated distance compensation and environmental compensation for monitoring targets at different spatial positions, weaken the interference of atmospheric attenuation and environmental background radiation on the temperature measurement result, and realize accurate and efficient detection of thermal value under wide-area and multi-target monitoring conditions. SUMMARY

[0005] Therefore, the present application provides a binocular infrared wide-area thermal value detection method based on distance and environmental background compensation to solve the technical problem that the existing infrared temperature measurement method has insufficient temperature measurement accuracy and reliability under wide-area monitoring conditions due to large target distance difference and serious influence of environmental background radiation and atmospheric attenuation.

[0006] To achieve the above technical purpose, the present application adopts the following technical scheme: In a first aspect, the present application provides a binocular infrared wide-area thermal value detection method based on distance and environmental background compensation, comprising: acquiring infrared radiation images and visible light binocular images in a monitoring range at the same time, and mapping the binocular image coordinate system to the infrared coordinate system based on joint calibration parameters; Stereo matching is performed on the binocular images in the full field of view to obtain a pixel disparity map, and a depth distance map corresponding to the infrared pixels is obtained according to the pixel disparity map and a coordinate system mapping relationship; The infrared radiation image is subjected to background segmentation, and a background region outside the heat source target is extracted, and an environmental background radiation field covering the monitoring range is reconstructed based on the radiation values of the background region; Current atmospheric environmental parameters are acquired, combined with the depth distance map, and based on an infrared radiation transmission physical model, pixel-by-pixel atmospheric transmittance is calculated; According to the atmospheric transmittance, the environmental background radiation field and the radiation values of the heat source target corresponding pixels in the infrared radiation image, the radiation correction value of the surface of the heat source target is calculated, and the actual temperature of the heat source target is obtained according to the radiation correction value.

[0007] Further, the infrared radiation image is subjected to background segmentation, and a background region outside the heat source target is extracted, including: The pixel radiation values of the infrared radiation image are subjected to histogram statistical analysis, and the main peak representing the background region and the secondary peak representing the heat source target in the histogram are identified; A dynamic radiation threshold separating the heat source target region and the background is calculated by using a double-peak valley bottom method; The region with a pixel radiation value greater than the dynamic radiation threshold is divided into a heat source target region to be removed, and the remaining region is defined as an initial background region; The initial background region is subjected to connected component analysis, and isolated noise points or false hot spots with an area less than a preset threshold are removed, thereby completing the segmentation of the background region.

[0008] Further, the environmental background radiation field covering the monitoring range is reconstructed based on the radiation values of the background region, including: The pixel points of the background region are subjected to radiation value sampling to obtain an environmental radiation sample set; For each to-be-measured pixel point of the heat source target, background pixel points in the neighborhood of the to-be-measured pixel point are searched in the environmental radiation sample set; According to the radiation values of the background pixel points in the neighborhood and the Euclidean distances of the background pixel points from the to-be-measured pixel point, an inverse distance weighted interpolation algorithm is used to calculate an environmental radiation estimated value of the to-be-measured pixel point; Based on the radiation values of the background region pixel points and the environmental radiation estimated values of the heat source target region, a two-dimensional environmental background radiation field matrix is constructed to form continuous environmental radiation distribution data in the full field of view.

[0009] Further, the method for calculating pixel-by-pixel atmospheric transmittance based on the infrared radiation transmission physical model includes: Atmospheric temperature and relative humidity in the monitoring range are acquired to determine the atmospheric attenuation coefficient; Based on the transmittance model, the atmospheric transmittance of each pixel is calculated according to the atmospheric attenuation coefficient and the depth distance map, and expressed by the formula: in, Represents pixel coordinates Atmospheric transmittance at that location. Indicates the atmospheric attenuation coefficient. This indicates the depth distance of the pixel.

[0010] Furthermore, the calculation of the radiation correction value of the heat source target surface includes: The remaining radiation value is obtained by subtracting the atmospheric path radiation term from the received total infrared radiation value to achieve distance compensation; wherein, the atmospheric path radiation term represents the atmospheric path radiation value from the heat source target to the infrared detector. The radiation difference is obtained by subtracting the environmental reflected radiation term from the remaining radiation value to achieve background environmental compensation; wherein the environmental reflected radiation term is obtained based on atmospheric transmittance, preset target emissivity and environmental background radiation value; Divide the radiation difference by the product of atmospheric transmittance and target emissivity to obtain the corrected true radiation value of the target surface.

[0011] Furthermore, before calculating the radiation correction value for the target surface of the heat source, the following steps are also included: Based on the geometric features of the heat source target region and joint calibration parameters, the position of the heat source target is mapped to the visible light image to determine the corresponding target candidate region; Feature extraction is performed on the target candidate region, and the K-nearest neighbor method is used to determine the material attribute classification of the target candidate region; The emissivity of the target heat source region is obtained by classifying and matching according to material properties, and then substituted into the radiation balance equation for calculation.

[0012] Furthermore, obtaining the actual temperature of the heat source target based on the radiation correction value includes: Substituting the pixel-by-pixel radiation correction value into the radiation-temperature conversion formula, the absolute temperature of the heat source target surface is obtained by inversion. The absolute temperature value is converted into the required engineering units as the actual temperature output of the heat source target.

[0013] Furthermore, the step of performing stereo matching on the binocular images across the entire field of view to obtain a pixel disparity map, and obtaining a depth distance map corresponding to the infrared pixels based on the pixel disparity map and the coordinate system mapping relationship, includes: A semi-global matching algorithm is used to perform stereo matching operations on the left and right visible light images in the binocular images across the entire field of view to obtain the pixel disparity of each visible light pixel: Based on the focal length, baseline length, and pixel disparity of the binocular camera, the disparity map is converted into a depth distance map in the visible light coordinate system; Based on the coordinate system mapping relationship, the depth distance map is reprojected and interpolated to convert the depth information in the visible light coordinate system into a depth distance map with one-to-one correspondence between infrared pixels.

[0014] Secondly, the present invention also provides a binocular infrared wide-area calorimetry detection device based on distance and environmental background compensation, comprising: The image acquisition module is used to acquire infrared radiation images and visible light binocular images within the monitoring range at the same time, and to map the binocular image coordinate system to the infrared coordinate system based on joint calibration parameters; The distance calculation module is used to perform stereo matching of binocular images in the full field of view to obtain a pixel disparity map. Based on the pixel disparity map and the coordinate system mapping relationship, a depth distance map corresponding to the infrared pixels is obtained. The background extraction module is used to segment the infrared radiation image, extract the background area outside the heat source target, and reconstruct the environmental background radiation field covering the monitoring range based on the radiation value of the background area. The atmospheric transmittance calculation module is used to obtain the current atmospheric environment parameters, combine them with the depth distance map, and calculate the atmospheric transmittance pixel by pixel based on the infrared radiation transmission physical model. The environment and distance compensation module is used to calculate the radiation correction value of the surface of the heat source target based on the atmospheric transmittance, the ambient background radiation field and the radiation value of the corresponding pixel of the heat source target in the infrared radiation image, and to obtain the actual temperature of the heat source target based on the radiation correction value.

[0015] Thirdly, the present invention also proposes a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the binocular infrared wide-area thermal value detection method based on distance and environmental background compensation described in the above technical solution.

[0016] Compared with existing technologies, the binocular infrared wide-area calorimetry detection method based on distance and environmental background compensation proposed in this invention has the following advantages: (1) By performing background segmentation on the infrared radiation image, the background area outside the heat source target is extracted, and the environmental background radiation field covering the monitoring range is reconstructed based on the radiation value of the background area. The influence of environmental radiation on the infrared temperature measurement results is explicitly modeled in the form of spatial distribution. Thus, environmental background compensation is introduced in the process of heat source target radiation inversion, which can effectively weaken the interference of environmental reflected radiation and background thermal radiation on the temperature measurement results and improve the accuracy and stability of heat source target temperature detection under complex environmental conditions.

[0017] (2) By using visible light binocular images to obtain spatial distance information of targets within the monitoring range, and combining the current atmospheric environmental parameters, atmospheric transmittance is calculated pixel by pixel based on the infrared radiation transmission physical model, thereby realizing atmospheric attenuation compensation for targets at different spatial locations; by introducing a transmittance model that varies with distance, the systematic error caused by using uniform atmospheric parameters in traditional infrared temperature measurement methods is avoided, and the infrared radiation attenuation caused by the difference in target distance can be effectively compensated, thereby improving the temperature measurement accuracy of distant targets and targets with uneven spatial distribution.

[0018] (3) This invention performs pixel-by-pixel modeling of depth distance, atmospheric transmittance, and environmental background radiation across the entire field of view, and completes heat source target radiation correction and temperature inversion on this basis, enabling synchronous and continuous calorific value detection for large-scale, multi-target scenes. This method does not rely on single-point or local area assumptions and is suitable for wide-area monitoring applications in complex and open environments. It can significantly expand the monitoring coverage while maintaining temperature measurement accuracy, improving the system's practicality in scenarios such as inspection, security, and large-scale thermal anomaly detection. Attached Figure Description

[0019] Figure 1 A schematic flowchart of the binocular infrared wide-area calorific value detection method based on distance and environmental background compensation provided by the present invention; Figure 2 This is a schematic diagram of the structure of the binocular infrared wide-area calorific value detection device based on distance and environmental background compensation provided by the present invention. Detailed Implementation

[0020] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.

[0021] Please see Figure 1 This embodiment provides a binocular infrared wide-area thermal value detection method based on distance and environmental background compensation, including: Step S101: Acquire infrared radiation images and visible light binocular images within the monitoring range at the same time, and map the binocular image coordinate system to the infrared coordinate system based on the joint calibration parameters; Step S102: Perform stereo matching on the binocular images in the full field of view to obtain a pixel disparity map, and obtain a depth distance map corresponding to the infrared pixels based on the pixel disparity map and the coordinate system mapping relationship. Step S103: Perform background segmentation on the infrared radiation image, extract the background area outside the heat source target, and reconstruct the environmental background radiation field covering the monitoring range based on the radiation value of the background area; Step S104: Obtain the current atmospheric environment parameters, combine them with the depth distance map, and calculate the atmospheric transmittance pixel by pixel based on the infrared radiative transmission physical model; Step S105: Calculate the radiation correction value of the surface of the heat source target based on the atmospheric transmittance, the environmental background radiation field and the radiation value of the corresponding pixel of the heat source target in the infrared radiation image, and obtain the actual temperature of the heat source target based on the radiation correction value.

[0022] The method in this embodiment accurately acquires the pixel-by-pixel depth distance of the target through binocular stereo ranging, and then calculates the accurate atmospheric transmittance, achieving fine compensation for distance attenuation of the radiation signal during transmission due to distance and atmospheric absorption. By modeling and reconstructing the background radiation field and incorporating it into the radiation balance equation for calculation, interference caused by the reflection of surrounding environmental thermal radiation from the target surface is effectively eliminated, ensuring the authenticity of the temperature measurement data. The entire process requires no manual intervention or adjustment, greatly improving the intelligence, automation level, and robustness of the temperature measurement system, making it suitable for complex and wide-area industrial inspection scenarios.

[0023] As a specific implementation, firstly, the infrared imaging unit and the visible light binocular imaging unit simultaneously image the monitoring range within the same time window, acquiring infrared radiation images and left and right visible light images at corresponding moments to ensure consistency of different imaging data in the time dimension and avoid time misalignment caused by target movement or environmental changes. Then, based on pre-acquired joint calibration parameters, a spatial geometric correspondence between the infrared imaging unit and the visible light binocular imaging unit is established. The joint calibration parameters describe the relative positional relationship and imaging geometry between the two imaging units, mapping the spatial line of sight corresponding to the pixel position in the visible light binocular image to the imaging line of sight of the infrared imaging unit.

[0024] Based on this, the pixel coordinates in the visible light binocular image coordinate system are transformed to the infrared image coordinate system according to the aforementioned spatial geometric correspondence, enabling a one-to-one correspondence between the spatial information obtained from the visible light binocular image and the pixel positions in the infrared radiation image. Through this processing, data alignment between different imaging systems under the same spatial reference is achieved, providing a foundation for subsequent joint calculations based on spatial distance and infrared radiation information.

[0025] In a preferred embodiment, step S102, which involves performing stereo matching on the binocular images across the entire field of view to obtain a pixel disparity map, and obtaining a depth distance map corresponding to the infrared pixels based on the pixel disparity map and the coordinate system mapping relationship, includes: A semi-global matching algorithm is used to perform stereo matching operations on the left and right visible light images in the binocular images across the entire field of view to obtain the pixel disparity of each visible light pixel. : Focal length based on binocular camera Baseline length and pixel parallax The disparity map is converted into a depth-distance map in the visible light coordinate system using the triangulation principle of binocular vision. This can be expressed as a formula: Depth-distance map based on coordinate system mapping Reprojection and interpolation are performed to convert the depth information in the visible light coordinate system into a depth-distance map corresponding to each infrared pixel. .

[0026] By employing a semi-global matching algorithm to perform stereo matching of left and right visible light images across the entire field of view, the stability and continuity of disparity calculation can be ensured while maintaining matching accuracy. This results in reliable pixel disparity maps even in complex textures and large-scale scenes. Compared to local matching methods, this approach effectively reduces mismatches and disparity discontinuities, providing more accurate foundational data for subsequent distance calculations.

[0027] In a preferred embodiment, step S103 involves background segmentation of the infrared radiation image to extract the background region outside the heat source target, including: Histogram statistical analysis was performed on the pixel radiance values ​​of infrared radiation images to identify the main peak representing the background region and the secondary peak representing the heat source target in the histogram. The dynamic radiation threshold that separates the heat source target area from the background is calculated using the bimodal valley method. Regions with pixel radiation values ​​greater than the dynamic radiation threshold are classified as heat source target regions to be removed, and the remaining regions are defined as initial background regions. Connectivity analysis is performed on the initial background region to remove isolated noise points or false hotspots with an area smaller than a preset threshold, thus completing the segmentation of the background region.

[0028] Histogram analysis maximizes the use of global information in an image, ensuring that the threshold is positioned at the most reasonable boundary between the heat source and background radiation distributions, thereby achieving accurate separation between the target and background regions. Dynamic threshold settings automatically adapt to changes in background temperature (e.g., daytime vs. nighttime, sunny vs. cloudy) within the monitored scene, eliminating the need for manual adjustments and ensuring both segmentation accuracy and system automation.

[0029] As a preferred embodiment, reconstructing the environmental background radiation field covering the monitoring area based on the radiation values ​​of the background region includes: Radiometric values ​​are sampled from pixels in the background region to obtain an environmental radiation sample set; For each pixel to be measured in the heat source target, the background pixels in the neighborhood of the pixel to be measured are searched in the environmental radiation sample set. In practical applications, a fixed number of points search (e.g., finding the nearest K points) is usually more robust, which can ensure that even in sparse background areas, there are still enough sample points for interpolation calculation, avoiding the inability to reconstruct certain areas due to lack of background information.

[0030] Based on the radiation values ​​of background pixels in the neighborhood and their Euclidean distance to the pixel to be measured, the estimated environmental radiation value of the pixel to be measured is calculated using the inverse distance weighted interpolation (IDW) algorithm. Based on the radiation values ​​of pixels in the background region and the estimated environmental radiation values ​​of the target heat source region, a two-dimensional environmental background radiation field matrix is ​​constructed to form continuous environmental radiation distribution data across the entire field of view.

[0031] This continuous field Ultimately, it will interact with the heat source target's own radiation. This is combined to calculate the environmental reflected radiation term in order to achieve accurate environmental reflection compensation.

[0032] In a preferred embodiment, the method for calculating pixel-by-pixel atmospheric transmittance based on the infrared radiative transfer physical model in step S104 includes: Obtain the atmospheric temperature and relative humidity within the monitoring range, and determine the atmospheric attenuation coefficient; Based on Beer-Lambert Law, a transmittance model is constructed. The atmospheric transmittance of each pixel is calculated using the atmospheric attenuation coefficient and depth distance map, expressed by the formula: in, Represents pixel coordinates Atmospheric transmittance at that location. Indicates the atmospheric attenuation coefficient. This indicates the depth distance of the pixel.

[0033] It should be noted that the atmospheric attenuation coefficient was obtained by fitting the infrared atmospheric transmission simulation software MODTRAN, and is expressed by the formula: in, Atmospheric temperature, This refers to relative humidity.

[0034] In a preferred embodiment, step S105, calculating the radiation correction value of the heat source target surface, includes: The remaining radiation value is obtained by subtracting the atmospheric path radiation term from the received total infrared radiation value to achieve distance compensation; wherein, the atmospheric path radiation term represents the atmospheric path radiation value from the heat source target to the infrared detector. The radiation difference is obtained by subtracting the environmental reflected radiation term from the remaining radiation value to achieve background environmental compensation; wherein the environmental reflected radiation term is obtained based on atmospheric transmittance, preset target emissivity and environmental background radiation value; Divide the radiation difference by the product of atmospheric transmittance and target emissivity to obtain the corrected true radiation value of the target surface.

[0035] Specifically, the radiation correction method is based on the single-pixel radiation balance equation, which describes that the signal received by the infrared sensor is composed of the target's own radiation, environmental reflection, and atmospheric path radiation, which are balanced superpositions. That is, for any pixel in the field of view... The total radiation value received by the sensor It can be represented as: in, The surface emissivity of the target object; The true radiation value of the target surface; The value of the reconstructed ambient background radiation field at this point is used to compensate for the ambient radiation reflected from the target surface; This is atmospheric path radiation, which is approximately equal to the blackbody radiance corresponding to atmospheric temperature. This indicates atmospheric transmittance.

[0036] By subtracting the atmospheric path radiation from the total signal using the above method, the interference of heat emitted by the air medium itself on the measurement is eliminated. Dividing by atmospheric transmittance compensates for the loss of target signal due to atmospheric absorption and attenuation during long-distance transmission, ensuring accurate intensity regardless of target distance, thus achieving distance compensation. Subtracting the environmental reflection radiation term effectively solves the problem of background heat reflected from the target surface. This allows the system to accurately distinguish between heat emitted by the target itself and reflected heat, guaranteeing the accuracy of temperature measurements in complex, high-reflectivity environments.

[0037] In a preferred embodiment, before calculating the radiation correction value of the heat source target surface, the method further includes: Based on the geometric features of the heat source target region and joint calibration parameters, the position of the heat source target is mapped to the visible light image to determine the corresponding target candidate region; Feature extraction is performed on the target candidate region, and the K-nearest neighbor method is used to determine the material attribute classification of the target candidate region; The emissivity of the target heat source region is obtained by classifying and matching according to material properties, and then substituted into the radiation balance equation for calculation.

[0038] By introducing target material identification based on visible light images and adaptively matching emissivity parameters before radiometric correction, the systematic errors caused by using a fixed emissivity can be avoided, thereby significantly improving the accuracy and reliability of temperature inversion for heat source targets of different materials in complex scenarios.

[0039] In some embodiments, obtaining the actual temperature of the heat source target based on the radiation correction value includes: Substituting the pixel-by-pixel radiation correction value into the radiation-temperature conversion formula, the absolute temperature of the heat source target surface is obtained by inversion. The absolute temperature value is converted into the required engineering units as the actual temperature output of the heat source target.

[0040] This embodiment also provides a binocular infrared wide-area calorimetry detection system 200 based on distance and environmental background compensation, including: Image acquisition module 201 is used to acquire infrared radiation images and visible light binocular images within the monitoring range at the same time, and to map the binocular image coordinate system to the infrared coordinate system based on joint calibration parameters; The distance calculation module 202 is used to perform stereo matching of the binocular image in the full field of view to obtain a pixel disparity map, and to obtain a depth distance map corresponding to the infrared pixels based on the pixel disparity map and the coordinate system mapping relationship. Background extraction module 203 is used to segment the infrared radiation image, extract the background area outside the heat source target, and reconstruct the environmental background radiation field covering the monitoring range based on the radiation value of the background area. Atmospheric transmittance calculation module 204 is used to obtain current atmospheric environmental parameters, combine them with depth distance map, and calculate pixel-by-pixel atmospheric transmittance based on infrared radiation transmission physical model; The environment and distance compensation module 205 is used to calculate the radiation correction value of the surface of the heat source target based on the atmospheric transmittance, the ambient background radiation field and the radiation value of the corresponding pixel of the heat source target in the infrared radiation image, and to obtain the actual temperature of the heat source target based on the radiation correction value.

[0041] This invention also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the binocular infrared wide-area thermal value detection method based on distance and environmental background compensation described in the above technical solution.

[0042] The computer-readable storage medium and computing device provided in the above embodiments of the present invention can be implemented with reference to the specific description of the binocular infrared wide-area calorific value detection method based on distance and environmental background compensation as described above, and have similar beneficial effects as the method described above, which will not be repeated here.

[0043] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A binocular infrared wide-area calorimetry detection method based on distance and environmental background compensation, characterized in that, include: The infrared radiation image and visible light binocular image within the monitoring range are acquired at the same time, and the binocular image coordinate system is mapped to the infrared coordinate system based on the joint calibration parameters; Stereo matching is performed on the binocular images across the entire field of view to obtain a pixel disparity map. Based on the pixel disparity map and the coordinate system mapping relationship, a depth distance map corresponding to the infrared pixels is obtained. Background segmentation is performed on the infrared radiation image to extract the background area outside the heat source target, and the environmental background radiation field covering the monitoring range is reconstructed based on the radiation value of the background area. Obtain current atmospheric environmental parameters, combine them with depth distance maps, and calculate pixel-by-pixel atmospheric transmittance based on the infrared radiative transmission physical model; Based on atmospheric transmittance, environmental background radiation field, and the radiation values ​​of the corresponding pixels of the heat source target in the infrared radiation image, the radiation correction value of the heat source target surface is calculated, and the actual temperature of the heat source target is obtained based on the radiation correction value.

2. The method according to claim 1, characterized in that, The step of performing background segmentation on the infrared radiation image to extract the background region outside the heat source target includes: Histogram statistical analysis was performed on the pixel radiance values ​​of infrared radiation images to identify the main peak representing the background region and the secondary peak representing the heat source target in the histogram. The dynamic radiation threshold that separates the heat source target area from the background is calculated using the bimodal valley method. Regions with pixel radiation values ​​greater than the dynamic radiation threshold are classified as heat source target regions to be removed, and the remaining regions are defined as initial background regions. Connectivity analysis is performed on the initial background region to remove isolated noise points or false hotspots with an area smaller than a preset threshold, thus completing the segmentation of the background region.

3. The method according to claim 1, characterized in that, The reconstruction of the environmental background radiation field covering the monitoring area based on the radiation values ​​of the background region includes: Radiometric values ​​are sampled from pixels in the background region to obtain an environmental radiation sample set; For each pixel to be tested in the heat source target, search for background pixels in the neighborhood of the pixel to be tested within the environmental radiation sample set; Based on the radiation values ​​of background pixels in the neighborhood and their Euclidean distance to the pixel to be measured, the estimated environmental radiation value of the pixel to be measured is calculated using an inverse distance weighted interpolation algorithm. Based on the radiation values ​​of pixels in the background region and the estimated environmental radiation values ​​of the target heat source region, a two-dimensional environmental background radiation field matrix is ​​constructed to form continuous environmental radiation distribution data across the entire field of view.

4. The method according to claim 1, characterized in that, The method for calculating pixel-by-pixel atmospheric transmittance based on the infrared radiative transfer physical model includes: Obtain the atmospheric temperature and relative humidity within the monitoring range, and determine the atmospheric attenuation coefficient; Based on the transmittance model, the atmospheric transmittance of each pixel is calculated according to the atmospheric attenuation coefficient and the depth distance map, and expressed by the formula: in, Represents pixel coordinates Atmospheric transmittance at that location. Indicates the atmospheric attenuation coefficient. This indicates the depth distance of the pixel.

5. The method according to claim 1, characterized in that, The calculation of the radiation correction value of the target surface of the heat source includes: The remaining radiation value is obtained by subtracting the atmospheric path radiation term from the received total infrared radiation value to achieve distance compensation; wherein, the atmospheric path radiation term represents the atmospheric path radiation value from the heat source target to the infrared detector. The radiation difference is obtained by subtracting the environmental reflected radiation term from the remaining radiation value to achieve background environmental compensation; wherein the environmental reflected radiation term is obtained based on atmospheric transmittance, preset target emissivity and environmental background radiation value; Divide the radiation difference by the product of atmospheric transmittance and target emissivity to obtain the corrected true radiation value of the target surface.

6. The method according to claim 5, characterized in that, Before calculating the radiation correction value for the target surface of the heat source, the following steps are also included: Based on the geometric features of the heat source target region and joint calibration parameters, the position of the heat source target is mapped to the visible light image to determine the corresponding target candidate region; Feature extraction is performed on the target candidate region, and the K-nearest neighbor method is used to determine the material attribute classification of the target candidate region; The emissivity of the target heat source region is obtained by classifying and matching according to material properties, and then substituted into the radiation balance equation for calculation.

7. The method according to claim 1, characterized in that, The process of obtaining the actual temperature of the heat source target based on the radiation correction value includes: Substituting the pixel-by-pixel radiation correction value into the radiation-temperature conversion formula, the absolute temperature of the heat source target surface is obtained by inversion. The absolute temperature value is converted into the required engineering units as the actual temperature output of the heat source target.

8. The method according to claim 1, characterized in that, The step of performing stereo matching on the binocular images across the entire field of view to obtain a pixel disparity map, and obtaining a depth distance map corresponding to the infrared pixels based on the pixel disparity map and the coordinate system mapping relationship, includes: A semi-global matching algorithm is used to perform stereo matching operations on the left and right visible light images in the binocular images across the entire field of view to obtain the pixel disparity of each visible light pixel: Based on the focal length, baseline length, and pixel disparity of the binocular camera, the disparity map is converted into a depth distance map in the visible light coordinate system; Based on the coordinate system mapping relationship, the depth distance map is reprojected and interpolated to convert the depth information in the visible light coordinate system into a depth distance map with one-to-one correspondence between infrared pixels.

9. A binocular infrared wide-area calorimetry detection device based on distance and environmental background compensation, characterized in that, include: The image acquisition module is used to acquire infrared radiation images and visible light binocular images within the monitoring range at the same time, and to map the binocular image coordinate system to the infrared coordinate system based on joint calibration parameters; The distance calculation module is used to perform stereo matching of binocular images in the full field of view to obtain a pixel disparity map. Based on the pixel disparity map and the coordinate system mapping relationship, a depth distance map corresponding to the infrared pixels is obtained. The background extraction module is used to segment the infrared radiation image, extract the background area outside the heat source target, and reconstruct the environmental background radiation field covering the monitoring range based on the radiation value of the background area. The atmospheric transmittance calculation module is used to obtain the current atmospheric environment parameters, combine them with the depth distance map, and calculate the atmospheric transmittance pixel by pixel based on the infrared radiation transmission physical model. The environment and distance compensation module is used to calculate the radiation correction value of the surface of the heat source target based on the atmospheric transmittance, the ambient background radiation field and the radiation value of the corresponding pixel of the heat source target in the infrared radiation image, and obtain the actual temperature of the heat source target based on the radiation correction value.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the binocular infrared wide-area thermal value detection method based on distance and environmental background compensation as described in any one of claims 1-8.

Images