A method and system for constructing a three-dimensional temperature field of a material surface of an industrial furnace

Abstract

The application discloses a kind of industrial furnace material surface three-dimensional temperature field construction method and system, by the synchronous acquisition visible light image and infrared image in industrial furnace, based on bright dark channel priori principle, estimate the dust transmissivity of visible light image, obtain clear visible light image according to the dust transmissivity of visible light image, according to the dust transmissivity of visible light image, mapping obtains the dust transmissivity of infrared image, and based on the dust transmissivity of infrared image, establish infrared temperature measurement compensation model and based on clear visible light image and infrared temperature measurement compensation model, construct industrial furnace material surface three-dimensional temperature field, solved the technical problem that present industrial furnace material surface three-dimensional temperature field construction precision is low, can realize the high fidelity detection and intuitive display of industrial furnace material surface three-dimensional temperature field, to provide important basis condition for industrial furnace furnace condition judgment, in-furnace reaction control etc..

Description

Technical Field

[0001] This invention mainly relates to the field of metallurgical technology, specifically a method and system for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns. Background Technology

[0002] In industrial furnaces, the surface temperature distribution of materials is an important indicator of combustion and guides material distribution control. Constructing a high-fidelity three-dimensional temperature field on the surface of materials in industrial furnaces can provide crucial temperature feedback for closed-loop control of the reaction process within the furnace, which is of great significance for improving the quality and efficiency of industrial furnace production, as well as reducing energy consumption and emissions.

[0003] In recent years, researchers have conducted extensive studies on temperature detection within industrial furnaces and kilns, upgrading the dimensionality of temperature detection from point to surface, and even three-dimensional temperature distribution. Currently, methods for measuring three-dimensional temperature fields using non-contact instruments are widely used, such as reconstructing the temperature distribution by capturing radiation images of industrial furnaces and kilns using charge-coupled device (CCD) cameras. The radiation intensity represented by each pixel in the radiation image is the sum of the attenuated radiation reaching the CCD target surface from each three-dimensional point along the line of sight. This method is susceptible to interference from foreign objects such as dust in the line of sight, resulting in low accuracy. Acoustic tomography inverts the temperature distribution within the furnace and kiln by calculating the multipath propagation time of sound waves, offering advantages such as a wide measurement range and large measurement space. However, the measurement effect is limited because sound waves are easily affected by changes in the temperature of the medium during propagation. Due to the complexity of three-dimensional temperature distribution and the complex environment of industrial furnaces and kilns, existing measurement devices and methods are insufficient to meet the requirements for constructing three-dimensional temperature fields on the surface of materials in industrial furnaces and kilns, and there are few methods for compensating for high dust levels in the furnace and kiln environment with infrared thermography.

[0004] To address this, the present invention proposes a method for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns. By constructing a visible light and infrared cross-spectral stereo vision detection system, the temperature field on the surface of materials in industrial furnaces and kilns is obtained. At the same time, a dust transmittance estimation model is established to obtain the dust transmittance in the visible light and infrared channels respectively, thereby improving the visible light imaging quality and compensating for the infrared results, ultimately achieving high-fidelity construction of the three-dimensional temperature field.

[0005] The invention disclosed in CN 102706459B is a device and method for detecting a three-dimensional temperature field inside a furnace using a single CCD imaging system. This invention uses an endoscopic industrial area array color CCD camera to image the visible light radiation inside the furnace, acquiring radiation images at different focal planes. The three-dimensional temperature field is calculated using an optical layered imaging method combined with colorimetric thermometry. However, the accuracy of the three-dimensional topography constructed using the single-channel camera defocusing method is limited, and the layered images relied upon by the colorimetric thermometry method are easily affected by dust inside the furnace, resulting in low accuracy of the calculated temperature.

[0006] The invention patent CN 110400336A discloses a method for reconstructing the three-dimensional temperature field of a flame using a dual-light-field camera. This patent preprocesses and connects the flame region in the light-field image, establishes a cylindrical medium model covering the connected flame region, and divides it into a mesh. Then, it extracts the three-dimensional contour of the flame. Finally, based on the three-dimensional flame contour and the radiation intensity values ​​corresponding to each pixel in the light-field image, it establishes a radiative transfer equation and calculates the mesh within the three-dimensional contour using an inversion algorithm, thereby obtaining the three-dimensional temperature field. However, this method uses morphological closing operations when connecting the flame region, resulting in insufficient fidelity in restoring the original flame morphology. Furthermore, the fineness of the flame's three-dimensional morphology depends on the fineness of the mesh division, limiting the description of the flame's periphery.

[0007] The invention patent with publication number CN 111967206A discloses a method, system, and application for constructing a three-dimensional temperature field for a waste heat boiler. This patent obtains a preliminary temperature field model by performing CFD simulation on the waste heat boiler; it then measures the temperature of the boiler's cross-section to obtain two-dimensional temperature field data; finally, it corrects the preliminary temperature field model based on the two-dimensional temperature field data to obtain a three-dimensional temperature field. This method uses ultrasonic temperature measurement technology to measure the temperature of the waste heat boiler's cross-section. However, because ultrasonic waves are easily interfered with in complex industrial environments, their measurement accuracy is limited, and CFD simulation is difficult to implement for large furnaces. Summary of the Invention

[0008] The present invention provides a method and system for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns, which solves the technical problem of low accuracy in the construction of three-dimensional temperature fields on the surface of materials in existing industrial furnaces and kilns.

[0009] To address the aforementioned technical problems, the present invention proposes a method for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns, comprising:

[0010] Simultaneously acquire visible light and infrared images of the industrial furnace.

[0011] Based on the principle of prior knowledge of the bright and dark channels, the dust transmittance of visible light images is estimated.

[0012] A clear visible light image is obtained based on the dust transmittance of the visible light image.

[0013] The dust transmittance of the visible light image is mapped to obtain the dust transmittance of the infrared image, and an infrared temperature measurement compensation model is established based on the dust transmittance of the infrared image.

[0014] Based on clear visible light images and infrared temperature compensation models, a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns is constructed.

[0015] Furthermore, based on the principle of prior knowledge of the bright and dark channels, the dust transmittance of the visible light image is estimated as follows:

[0016] A dust transmittance estimation model based on priors for bright and dark channels is established.

[0017] Calculate the atmospheric light values ​​in different brightness regions of a visible light image.

[0018] Based on the atmospheric light value and dust transmittance estimation model of different brightness regions in visible light images, the dust transmittance of bright and dark channels in different brightness regions of visible light images is calculated.

[0019] A fitness function is constructed based on the image energy gradient, a particle swarm parameter optimization model is established, and the optimal values ​​of dust transmittance weights in the bright and dark channels are calculated.

[0020] The dust transmittance of a visible light image is obtained based on the optimal weighting of dust transmittance in the bright and dark channels and the dust transmittance in the bright and dark channels of different brightness regions in the visible light image.

[0021] Furthermore, calculating the atmospheric light values ​​in different brightness regions of a visible light image includes:

[0022] The visible light image was divided into an extremely bright central flame region using dual threshold segmentation. B Extremely dark edge area J D and the middle region J G .

[0023] Based on the prior of the bright channel, the atmospheric light value of the bright channel is calculated in the central flame region and the middle region of the visible light image, and is used as the atmospheric light value of the central flame region. The specific calculation formula is as follows:

[0024] A B =g(m(J) B )),

[0025] Among them, A B The atmospheric light value of the central flame region, m(J) B ) represents set J B The set of pixels with the highest grayscale values ​​in the top 0.1%, g(m(J) B )) represents m(J B The average grayscale value of ).

[0026] Based on the dark channel prior, the atmospheric light value A in the dark channel is calculated in the edge and middle regions of the visible light image. D As the atmospheric light value of the edge region, the specific calculation formula is as follows:

[0027] A D =g(m(J) D )),

[0028] Among them, A D The atmospheric light value in the dark channel, m(J) D ) represents set JD The set of pixels with the highest grayscale values ​​in the top 0.1%, g(m(J) D )) represents m(J D The average grayscale value of ).

[0029] The atmospheric light values ​​of the bright channel and the dark channel are averaged to obtain the atmospheric light value A in the middle region. G The specific calculation formula is as follows:

[0030]

[0031] Furthermore, based on the atmospheric light value and dust transmittance estimation model for different brightness regions in a visible light image, the formula for calculating the dust transmittance of the bright and dark channels in different brightness regions of a visible light image is as follows:

[0032]

[0033]

[0034] Where, τ dark and τ bright J represents the dust transmittance in the dark and bright channels of a visible light image, respectively. dark (x) and J bright (x) represents the dark channel and bright channel of the visible light image, respectively, where x is a pixel in the visible light image and ω is an empirical value for preserving the image depth.

[0035] Furthermore, a fitness function is constructed based on the image energy gradient, and a particle swarm optimization model is established to calculate the optimal values ​​of dust transmittance weights in the bright and dark channels, including:

[0036] Construct the fitness function, where the fitness function is specifically:

[0037]

[0038] Where fit(J(x,a)) is the fitness function, I(x) is the original image affected by dust and fog, J(x,a) is the visible light image, and τ dark and τ bright The dust transmittance of the dark and bright channels of a visible light image, a and a', are respectively. best Here, J(x,a) represents the dust transmittance weighting parameter and the optimal weighting value for the bright and dark channels, respectively. G(J(x,a)) is the sum of the boundary intensity values ​​after edge detection of the visible light image using the Sobel operator. H(J(x,a)) is the entropy of the visible light image, and E(J(x,a)) is the energy gradient value of the visible light image.

[0039] Based on the fitness function, the parameters of the dust transmittance weights in the bright and dark channels of the visible light imaging path are optimized to obtain the optimal values ​​of the dust transmittance weights in the bright and dark channels of the visible light imaging path.

[0040] Furthermore, based on the dust transmittance of the visible light image, the specific formula for obtaining a clear visible light image is as follows:

[0041]

[0042] J 清晰 (x) is a clear image in visible light, τ VIS Let τ be the dust transmittance of the visible light image, and τ be the dust transmittance. VIS =a best ·τ bright +(1-a best )·τ dark I(x) is the original image affected by dust and fog.

[0043] Furthermore, based on the dust transmittance of the visible light image, the dust transmittance of the infrared image is obtained by mapping, and an infrared thermometry compensation model is established based on the dust transmittance of the infrared image, including:

[0044] By using coordinate mapping, the dust transmittance τ' of the infrared image is obtained by mapping the dust transmittance of the visible light image. IR (x), the specific calculation formula is:

[0045]

[0046] Where τ' IR (x) and τ VIS (x) represents the dust transmission in the infrared and visible light images, respectively. R1, t1 represents the relative position between the visible light camera and the world coordinate system, and R2, t2 represents the relative position between the infrared camera and the world coordinate system.

[0047] The dust transmittance of the infrared image is corrected based on the baseline distance between the visible light camera and the infrared camera, the distance between the baseline center and the target, and the size of the infrared image.

[0048] Based on the dust transmittance of the corrected infrared image, an infrared thermometry compensation model is established, specifically as follows:

[0049]

[0050] Where T is the target's true temperature, T0 is the infrared thermometry value under dust interference, ΔT represents the temperature difference in the infrared image under dust interference, a, b, c, and d are the fitting parameters, and τ is the temperature difference in the infrared image under dust interference. IR This represents the dust transmittance of the corrected infrared image.

[0051] Furthermore, based on the baseline distance between the visible light camera and the infrared camera, the distance between the baseline center and the target, and the size of the infrared image, the dust transmittance of the infrared image is corrected, including:

[0052] The correction coefficients are obtained based on the baseline distance between the visible light camera and the infrared camera, the distance between the baseline center and the target, and the size of the infrared image. Specifically:

[0053]

[0054] Wherein, δ(τ') IR ) is the correction factor, τ' IR The value is the dust transmittance of the infrared image before correction, l is the baseline distance between the visible light camera and the infrared camera, s is the distance between the baseline center and the target, and w and h are the width and height of the infrared image size.

[0055] The dust transmittance of the infrared image is corrected based on the correction coefficient. The specific formula is as follows:

[0056] τ IR (x)=τ' IR (x)+δ(τ' IR ),

[0057] Where, τ IR (x) represents the dust transmittance of the corrected infrared image.

[0058] Furthermore, based on clear visible light images and an infrared temperature compensation model, a three-dimensional temperature field is constructed on the surface of materials in industrial furnaces, including:

[0059] A parallax image is obtained by stereo matching of a clear visible light image and an infrared image.

[0060] The parallax image is converted into a depth image, and the coordinate transformation is performed on the depth image to obtain a three-dimensional point cloud of the material surface in the industrial furnace.

[0061] The target real temperature output by the infrared temperature measurement compensation model is mapped onto a three-dimensional point cloud through coordinate mapping to obtain the three-dimensional temperature field on the surface of materials in industrial furnaces and kilns.

[0062] The three-dimensional temperature field construction system for the surface of materials in industrial furnaces and kilns provided by this invention includes:

[0063] The present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method for constructing a three-dimensional temperature field on the surface of materials in an industrial furnace provided by the present invention.

[0064] The present invention provides a method and system for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns. This method and system simultaneously acquires visible light and infrared images of the industrial furnace and kiln. Based on the principle of light-dark channel prior, it estimates the dust transmittance of the visible light image. A clear visible light image is obtained based on the dust transmittance of the visible light image. The dust transmittance of the infrared image is then mapped from the visible light image. An infrared temperature compensation model is established based on the dust transmittance of the infrared image, and a three-dimensional temperature field on the surface of materials in the industrial furnace and kiln is constructed based on the clear visible light image and the infrared temperature compensation model. This method solves the technical problem of low accuracy in constructing three-dimensional temperature fields on the surface of materials in existing industrial furnaces and kilns. It enables high-fidelity detection and intuitive display of the three-dimensional temperature field on the surface of materials in industrial furnaces and kilns, thus providing important basis for judging the furnace condition and controlling reactions within the furnace.

[0065] The beneficial effects of this invention are:

[0066] (1) In view of the difficulty in obtaining surface information of materials in industrial furnaces and kilns, this invention applies infrared and visible light cross-spectral stereo vision detection system to the acquisition of three-dimensional temperature field of material surface in industrial furnaces and kilns for the first time. High-fidelity three-dimensional temperature field is constructed by using the acquired visible light and infrared temperature information.

[0067] (2) In view of the impact of high dust in industrial furnaces on temperature field detection, this invention proposes an atmospheric light value estimation method based on the difference in image brightness regions, combined with the prior of the bright and dark channels, to obtain the preliminary dust transmittance of the bright and dark channels. The dust transmittance weight of the bright and dark channels is optimized by combining the improved image energy gradient function and particle swarm optimization algorithm to obtain the accurate dust transmittance.

[0068] (3) Based on the calculation of dust transmittance in the visible light imaging path, this invention establishes a visible light image sharpening model to restore the clarity of the visible light image and ensure image quality.

[0069] (4) The present invention constructs a compensation model between dust transmittance and infrared temperature measurement error, realizes the compensation of infrared temperature measurement results under dust interference, and ensures the temperature accuracy of the three-dimensional temperature field.

[0070] (5) The present invention corrects the dust transmittance of infrared images based on the baseline distance between visible light cameras and infrared cameras, the distance between the baseline center and the target, and the size of infrared images, thereby obtaining accurate dust transmittance of infrared images and further improving the accuracy of constructing three-dimensional temperature field on the surface of industrial furnace materials.

[0071] (6) This invention deploys visible light and infrared cameras in a cross-spectral stereo vision detection system, incorporates cross-spectral visible light morphology information and temperature information into the same system, and then uses direct coordinate mapping to map the temperature data onto the three-dimensional point cloud, effectively ensuring the accuracy of the three-dimensional temperature field. Attached Figure Description

[0072] Figure 1 This is a schematic diagram of the method for constructing a three-dimensional temperature field on the surface of materials in an industrial furnace according to Embodiment 2 of the present invention;

[0073] Figure 2 This is a structural block diagram of the industrial furnace material surface three-dimensional temperature field construction system according to an embodiment of the present invention.

[0074] Figure label:

[0075] 10. Memory; 20. Processor. Detailed Implementation

[0076] To facilitate understanding of the present invention, the present invention will be described more fully and in detail below with reference to the accompanying drawings and preferred embodiments, but the scope of protection of the present invention is not limited to the following specific embodiments.

[0077] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings, but the present invention can be implemented in many different ways as defined and covered by the claims.

[0078] Example 1

[0079] The method for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns provided in Embodiment 1 of the present invention includes:

[0080] Step S101: Simultaneously acquire visible light and infrared images of the industrial furnace.

[0081] Step S102: Based on the prior principle of the bright and dark channels, estimate the dust transmittance of the visible light image.

[0082] Step S103: Obtain a clear visible light image based on the dust transmittance of the visible light image.

[0083] Step S104: Based on the dust transmittance of the visible light image, the dust transmittance of the infrared image is obtained by mapping, and an infrared temperature measurement compensation model is established based on the dust transmittance of the infrared image.

[0084] Step S105: Based on clear visible light images and infrared temperature compensation models, construct a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns.

[0085] The method for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces provided in this invention simultaneously acquires visible light and infrared images of the industrial furnace. Based on the principle of light-dark channel prior, it estimates the dust transmittance of the visible light image. Based on the dust transmittance of the visible light image, it obtains a clear visible light image. Based on the dust transmittance of the visible light image, it maps the dust transmittance of the infrared image to obtain the dust transmittance. Based on the dust transmittance of the infrared image, it establishes an infrared temperature measurement compensation model. Based on the clear visible light image and the infrared temperature measurement compensation model, it constructs a three-dimensional temperature field on the surface of materials in industrial furnaces. This method solves the technical problem of low accuracy in constructing the three-dimensional temperature field on the surface of materials in existing industrial furnaces. It can achieve high-fidelity detection and intuitive display of the three-dimensional temperature field on the surface of materials in industrial furnaces, thus providing important basis for judging the furnace condition and controlling the reaction inside the furnace.

[0086] Specifically, this invention focuses on the three-dimensional temperature field of materials on the surface of industrial furnaces and kilns, proposing a method and system for constructing a three-dimensional temperature field. By using visible light and infrared cameras, a cross-spectral stereo vision detection system for visible light and infrared is built. A dust estimation model combining priors for bright and dark channels is constructed, and an atmospheric light value estimation method based on image brightness differences is proposed. The dust transmittance of the visible light channel is initially estimated, and an image quality evaluation and sharpening model is established. The weights of dust transmittance in the bright and dark channels are optimized using the image energy gradient function and particle swarm optimization algorithm to obtain the dust transmittance weight coefficients of the visible light imaging path, recovering the visible light image under dust-free conditions and improving the accuracy of the three-dimensional morphology. Simultaneously, based on the relative pose relationship between the visible light and infrared cameras, the dust transmittance of the visible light imaging path is obtained from the dust transmittance of the infrared imaging path using coordinate mapping. Then, this invention establishes an infrared temperature compensation model, using experimental data to fit the relationship between dust transmittance and temperature error, compensating for the infrared channel temperature measurement results to ensure the temperature accuracy of the constructed three-dimensional temperature field. Finally, in this embodiment of the invention, the temperature information from the infrared image is directly mapped to a three-dimensional point cloud through coordinate transformation, ultimately obtaining a three-dimensional temperature field. The method and system for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces proposed in this embodiment of the invention can achieve high-fidelity construction of a three-dimensional temperature field within industrial furnaces, thereby providing important basis for judging furnace conditions and controlling reactions within the furnace.

[0087] Example 2

[0088] This invention provides a method and system for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns. First, a cross-spectral stereo vision detection system is constructed to simultaneously acquire visible light and infrared images of the interior of the industrial furnace or kiln. Then, a visible light image sharpening model based on dust transmittance estimation and an infrared temperature measurement compensation model are successively established to gradually achieve high-fidelity detection of the temperature distribution on the surface of materials in the industrial furnace or kiln. Finally, the temperature data is directly mapped to a three-dimensional point cloud to construct a three-dimensional temperature field. A schematic diagram of the method for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns according to an embodiment of this invention is shown below. Figure 1 As shown, it includes the following main steps:

[0089] (1) Design a cross-spectral stereo vision detection system based on visible light and infrared vision. Design a multispectral camera calibration board that is applicable to both visible and infrared spectra. Calibrate the visible light camera and infrared camera of the dual-channel three-dimensional thermal imager respectively and obtain the internal and external parameters of the two-channel cameras.

[0090] (2) Establish a dust transmittance estimation model based on the prior of the bright and dark channels, calculate the atmospheric light value in different regions of the visible light image, and then calculate the dust transmittance of the corresponding region in the bright and dark channels based on the prior of the bright and dark channels to obtain the preliminary dust transmittance distribution. Then, construct a fitness function based on the image energy gradient, establish a particle swarm parameter optimization model, calculate the dust transmittance weights of the bright and dark channels in the visible light imaging path, and obtain the accurate dust transmittance of the visible light imaging path.

[0091] (3) Establish an image sharpening model, take the calculated dust transmittance of the visible light imaging path as input, and restore the visible light image in the dust-free state.

[0092] (4) Combining the camera calibration parameters, the visible spectrum dust transmittance distribution is mapped to the infrared scene through the homography matrix. Then, an infrared temperature measurement compensation model is established to compensate for the temperature measurement error caused by dust, ensuring the temperature accuracy of the constructed three-dimensional temperature field.

[0093] (5) Use the cleared visible light image and infrared image to perform cross-spectral stereo matching to obtain a parallax image, which is then converted into a depth image and a three-dimensional point cloud. Based on the dual-channel camera internal and external parameters obtained in step (1), the temperature data is directly mapped onto the three-dimensional point cloud to form a three-dimensional temperature field.

[0094] The specific implementation plan is as follows:

[0095] (1) Construction of a transspectral stereo vision detection system and camera calibration

[0096] Zhang Zhengyou's calibration method is a mature and accurate method for calibrating camera parameters. For the calibration of visible light cameras, a checkerboard pattern made of printed paper is generally used as a calibration plate. However, since the infrared emissivity of the checkerboard pattern is basically the same, it is difficult to clearly describe the feature information such as the corner points of the checkerboard pattern during infrared imaging. Therefore, it is necessary to remake a checkerboard pattern made of materials with different emissivity to adapt to the calibration of multispectral cameras.

[0097] Specifically, this invention uses an aluminum plate with an emissivity of 0.11 to 0.19 at room temperature and a matte black paint with an emissivity of approximately 0.95 at room temperature. After the surface of the aluminum plate is brushed, the black parts of the checkerboard pattern are evenly filled with paint, as shown in formula (1). Zhang's checkerboard calibration method can be described as follows:

[0098]

[0099] Assume the camera calibration plate lies in the XOY plane of the world coordinate system, i.e., Z = 0, where [uv 1] T This represents the homogeneous coordinates of a point projected from the calibration plate plane onto the image plane, [XY 1]. T Let represent the homogeneous coordinates of a point on the calibration plane. K is the intrinsic parameter matrix of the camera, and R = [r1r2r3] and t are the rotation matrix and translation vector of the camera coordinate system relative to the world coordinate system, respectively. Let:

[0100]

[0101] Where H is the homography matrix, and according to the properties of rotation matrices, r1 T If r2 = 0 and ||r1|| = ||r2|| = 1, then each image has the following constraints on the camera intrinsic parameter matrix:

[0102]

[0103] By combining equations (2) and (3), and based on the constraints of the n images, the camera intrinsic parameters can be obtained. Furthermore, the extrinsic parameters r1, r2, r3, t can be obtained, where...

[0104]

[0105] For the visible light channel and the infrared channel, let R1, t1 and R2, t2 represent their respective extrinsic parameters, i.e., R1, t1 represents the relative position between the visible light camera and the world coordinate system, and R2, t2 represents the relative position between the infrared camera and the world coordinate system. For any point P in the world coordinate system, let its non-homogeneous coordinates in the world coordinate system, the visible light camera coordinate system, and the infrared camera coordinate system be x, x, and t2, respectively. w x1, x2, then by:

[0106]

[0107] Eliminate x w The geometric relationship between the visible light camera and the infrared camera is obtained:

[0108]

[0109] (2) Calculation of dust transmittance in visible light imaging path

[0110] The high dust environment inside industrial furnaces and kilns affects the imaging and temperature measurement accuracy of equipment. Analysis of visible light images inside the furnaces and kilns reveals that there are both excessively bright and excessively dark areas, such as the gas flame (excessively bright) and the edge of the material surface (excessively dark). Therefore, this invention integrates the prior knowledge of the bright and dark channels and first uses visible light images to estimate the dust transmittance in the imaging path.

[0111] Based on prior knowledge of the dark channel, for image J, when there is no interference such as dust in the imaging path, the dark channel J of the image... dark The value tends towards 0; when interference such as dust is present, the dark channel J of the image... dark It will not tend towards 0, meaning that in the visible light and infrared images obtained by this invention, the dark channel:

[0112]

[0113] J C Let Ω(x) represent the color channels of image J, and Ω(x) represent the window centered at pixel x. Based on the fog image formation model:

[0114] I(x)=J(x)τ(x)+A(1-τ(x)) (8)

[0115] Where I(x) is the original foggy image, J(x) is the fog-free image, and A is the global value of atmospheric light. Let τ(x) represent the transmittance of the imaging path, then:

[0116]

[0117] Then, by taking the minimum values ​​of both regions and color channels on both sides of the equation, the dark channel transmittance τ can be obtained. dark The estimation formula is:

[0118]

[0119] Where ω = 0.95 is an empirical value for preserving image depth of field.

[0120] The bright channel prior corresponds to the dark channel prior, meaning that in most local regions of natural scenes, at least one color channel has a relatively large pixel value. In the visible light and infrared images with dust obtained in this invention, the bright channel J... brightIt will not tend towards 255, but rather towards a local quantity A(x) of atmospheric light value:

[0121]

[0122] Similar to the dark channel, the dust transmittance τ in the bright channel... bright The estimate is:

[0123]

[0124] Then, the weighted sum of dust transmittance in the bright and dark channels is used as the dust transmittance distribution in the visible light imaging path:

[0125] τ VIS =a·τ bright +(1-a)·τ dark (13)

[0126] The value of a is in the range of 0 ≤ a ≤ 1.

[0127] According to the formula for calculating dust transmittance in both bright and dark channels, obtaining the atmospheric light value A is crucial for calculating the dust transmittance in both channels. Since the brightness distribution of visible light images on the surface of materials in industrial furnaces is uneven, extremely bright flame areas do not conform to the prior theory of the dark channel, and similarly, extremely dark areas at the image edges do not conform to the prior theory of the bright channel. Therefore, this invention proposes a method for calculating the atmospheric light value based on the differences in image brightness regions. The basic steps are as follows:

[0128] (1) The visible light image J is divided into an extremely bright central flame region J by dual threshold segmentation. B Extremely dark edge area J D and the middle region J G .

[0129] (2) Based on the prior of the bright channel, the atmospheric light value A of the bright channel is calculated in the flame region at the center of the image and the middle region. B Let A be the atmospheric light value of the central flame region. Define a function m(P) to represent the top 0.1% of the values ​​in set P, and let g(Q) represent the average gray value of the pixel set Q. B for:

[0130] A B =g(m(J) B (14)

[0131] (3) Based on the dark channel prior, the atmospheric light value A of the dark channel is calculated in the image edge region and the middle region. D As the atmospheric light value of the edge region:

[0132] A D =g(m(J) D (15)

[0133] (4) Average the atmospheric light values ​​of the bright channel and the dark channel to obtain the atmospheric light value A for the general brightness area. G :

[0134]

[0135] After obtaining the atmospheric light values ​​that differentiate the brightness regions of the image, for a pixel x in the visible light image, its dark channel transmittance is:

[0136]

[0137] The transmittance of the bright channel is calculated as follows:

[0138]

[0139] After obtaining the preliminary dust transmittance distribution in the bright and dark channels, it is necessary to further determine the weighting coefficients of the dust transmittance in the bright and dark channels. This embodiment of the invention uses an intelligent algorithm and employs effective evaluation criteria to measure the quality of the visible light image recovered based on the dust transmittance, combining the fitness function fit with image entropy, edge intensity, and image energy gradient:

[0140] fit(J(x,a))=log(log(G(J(x,a))))·H(J(x,a))·E(J(x,a)) (19)

[0141] Where J(x,a) is the visible light image to be evaluated, calculated from the weighted dust transmittance distribution:

[0142]

[0143] G(J(x,a)) is the sum of boundary intensity values ​​after edge detection of the visible light image using the Sobel operator, H(J(x,a)) is the entropy of the visible light image, and E(J(x,a)) is the energy gradient value of the visible light image, which are calculated as follows:

[0144]

[0145]

[0146]

[0147] Where p i This represents the proportion of pixels with a gray value of i (0≤i≤255) in the image.

[0148] From (20), the image quality is basically determined by the value of a. Using the above fitness function to continuously optimize the value of a, the value of a corresponding to the maximum value of the fitness function fit(J(x,a)) is taken as the optimal weight value, and τ is obtained. VIS The optimal result is the determined dust transmittance distribution in the visible light imaging path.

[0149] (3) Visible light image sharpening

[0150] Since the visible light imaging quality in this invention is mainly affected by dust interference, based on the accurate calculation of dust transmittance, this invention, according to the image degradation model under dust interference, optimizes the dust transmittance distribution τ obtained in the previous step. VIS Substitute:

[0151]

[0152] J 清晰 (x) is a clear image in visible light, τ VIS Let τ be the dust transmittance of the visible light image, and τ be the dust transmittance. VIS =a best ·τ bright +(1-a best )·τ dark I(x) is the original image affected by dust and fog, and the final visible light image without dust interference is obtained.

[0153] (4) Infrared temperature measurement compensation model

[0154] In transspectral stereo vision detection systems, the dust distribution in the visible light and infrared imaging paths is similar, but differences still exist. To estimate the transmittance distribution of the infrared imaging path using the dust transmittance distribution of the visible light imaging path, this invention, based on dual-channel stereo calibration, firstly uses coordinate mapping to obtain the dust transmittance τ' of the corresponding pixel in the infrared imaging channel by mapping the acquired visible light path dust transmittance. IR (x):

[0155] τ' IR (x)=τ VIS (x)·R+t (25)

[0156] Then, a correction term δ is defined, which is related to the baseline distance l between the two cameras in the transspectral stereo vision detection system, the distance s between the baseline center and the target, and the pixel transmittance τ'. IR The visible light and infrared image sizes h and w are related and are used to correct for transmittance differences caused by differences in the viewing angle of the imaging system. They are obtained through experimental calibration.

[0157]

[0158] The dust transmittance in the infrared imaging path can then be expressed as:

[0159] τ IR (x)=τ' IR (x)+δ(τ' IR (27)

[0160] Infrared images reflect the heat distribution on the surface of the object being measured. However, due to the nonlinear relationship between the infrared radiation received by the thermal imaging device and the temperature of the object, coupled with the influence of environmental factors such as the surface emissivity and atmospheric attenuation, infrared images can only provide a qualitative description of the surface radiation temperature. Therefore, it is necessary to calibrate the absolute temperature value of the object by comparing it with a reference object. To study the actual impact of dust on infrared thermometry, this embodiment of the invention uses a high-precision blackbody furnace as a reference object, measures the furnace temperature with an infrared camera, and constructs an infrared imaging path with a dust transmittance τ from a data modeling perspective. IR The function relating infrared measurement data T0 and the target's actual temperature data T:

[0161]

[0162] Here, ΔT represents the temperature difference of corresponding pixels in the infrared image under dust interference. For ΔT, this invention collects blackbody furnace temperature data with and without dust interference under constant ambient temperature and fixed positions of the thermal imager and blackbody furnace. The measurement data are fitted using the least squares method to obtain the closest fitting curve between the temperature difference and dust transmittance, where a, b, c, and d are the fitting parameters. Due to the nonlinearity of the fitting curve, this model has high accuracy.

[0163] (5) Construction of three-dimensional temperature field based on direct coordinate mapping

[0164] After obtaining a clear visible light image of the material surface in the industrial furnace and the compensated temperature data, the visible light image and the infrared image are stereo matched according to the general method of binocular stereo vision 3D reconstruction to obtain a visible light and infrared parallax image, which is then converted to obtain a depth map. Then, according to the coordinate system transformation relationship, the depth image is converted into a 3D point cloud.

[0165] If Z is denoted as the depth value of the corresponding point in the depth image of the material surface of an industrial furnace, then p(x) IR ,y IR ,T IR Let f be a point in an infrared image containing real temperature information, f be the focal length of the infrared camera, and p be a point in an infrared image containing real temperature information. W (X W ,Y W Z W T WLet p(x) be a point on the surface of the material in an industrial furnace that contains temperature information. Then, based on the coordinate system mapping relationship obtained during the system calibration process, p(x) can be... IR ,y IR ,T IR Mapping this onto a 3D point cloud allows us to expand the 3D point cloud into a 3D temperature field.

[0166]

[0167] This completes the construction of a high-fidelity three-dimensional temperature field.

[0168] This invention focuses on the three-dimensional temperature field of materials on the surface of industrial furnaces and kilns, proposing a method and system for constructing a three-dimensional temperature field. By using visible light and infrared cameras, a cross-spectral stereo vision detection system for visible light and infrared is built. A dust estimation model combining priors for bright and dark channels is constructed, and an atmospheric light value estimation method based on image brightness differences is proposed. This initially estimates the dust transmittance of the visible light channel and establishes an image quality evaluation and sharpening model. The weights of dust transmittance in bright and dark channels are optimized using the image energy gradient function and particle swarm optimization algorithm to obtain the dust transmittance weight coefficients for the visible light imaging path, thus restoring the visible light image in a dust-free state and improving the accuracy of the three-dimensional morphology. Simultaneously, based on the relative pose relationship between the visible light and infrared cameras, the dust transmittance of the visible light imaging path is obtained by coordinate mapping to the dust transmittance of the infrared imaging path. Then, this invention establishes an infrared temperature compensation model, using experimental data to fit the relationship between dust transmittance and temperature error, compensating for the infrared channel temperature measurement results to ensure the temperature accuracy of the constructed three-dimensional temperature field. Finally, this invention directly maps the temperature information of the infrared image to a three-dimensional point cloud through coordinate transformation, ultimately obtaining the three-dimensional temperature field. The method and system for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns proposed in this invention can achieve high-fidelity construction of a three-dimensional temperature field inside industrial furnaces and kilns, thereby providing important basis and conditions for judging the furnace condition and controlling the reaction inside the furnace.

[0169] Example 3

[0170] This embodiment uses a laboratory blast furnace simulation model as the experimental platform. The method and system for constructing a three-dimensional temperature field on the surface of materials in an industrial furnace, as proposed in this invention, are deployed on this platform. The model and method proposed in this invention are used to accurately calculate the dust emissivity in the visible light imaging path, and to achieve clarity of the visible light image and infrared thermometry compensation, thereby obtaining a high-fidelity three-dimensional temperature field inside the furnace. The specific steps are as follows:

[0171] 1. Build a visible light and infrared cross-spectral stereo vision detection system, and use a multispectral camera calibration board to calibrate the intrinsic and extrinsic parameters of the binocular camera to obtain the intrinsic and extrinsic parameters of the visible light and infrared camera.

[0172] 2. A dust environment was set up inside a simulated blast furnace. Visible light and infrared image sequences inside the furnace were acquired using a binocular camera. The atmospheric light value calculation method based on the difference in image brightness regions was used to preliminarily calculate the dust transmittance distribution in the bright and dark channels of the visible light imaging path. Then, an image quality evaluation mechanism based on the image energy gradient function was established. The particle swarm optimization algorithm was used to optimize the weight coefficients of dust transmittance in the bright and dark channels to obtain an accurate dust transmittance distribution in the visible light imaging path.

[0173] 3. Based on the accurate calculation of dust transmittance in the visible light imaging path, an image sharpening model is established to obtain a clear visible light image.

[0174] 4. Combining camera calibration parameters, the dust transmittance of the infrared imaging path is obtained through coordinate mapping. A compensation model between dust transmittance and infrared temperature measurement error is established to correct the infrared temperature measurement results.

[0175] 5. The image sequence is subjected to distortion correction and epipolar correction. A stereo matching algorithm is used to perform cross-spectral stereo matching on the corrected visible light and infrared image pairs to obtain a parallax image, which is then converted into a depth map and a 3D point cloud.

[0176] 6. Based on the camera's external parameters, the corrected infrared temperature is directly mapped onto the 3D point cloud to obtain the 3D temperature field.

[0177] Reference Figure 2 The three-dimensional temperature field construction system for the surface of industrial furnace materials proposed in this embodiment includes a memory 10, a processor 20, and a computer program stored on the memory 10 and run on the processor 20. When the processor 20 executes the computer program, it implements the steps of the three-dimensional temperature field construction method for the surface of industrial furnace materials proposed in this embodiment.

[0178] The specific working process and working principle of the industrial furnace material surface three-dimensional temperature field construction system in this embodiment can be referred to the working process and working principle of the industrial furnace material surface three-dimensional temperature field construction method in this embodiment.

[0179] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for constructing a three-dimensional temperature field on the surface of materials in an industrial furnace, characterized in that, The method includes: Simultaneously acquire visible light and infrared images of the industrial furnace; Based on the principle of brightness-dark channel prior, the dust transmittance of a visible light image is estimated. This estimation of dust transmittance in a visible light image, based on the principle of brightness-dark channel prior, includes: A dust transmittance estimation model based on priors of the bright and dark channels was established. Calculate the atmospheric light values ​​of different brightness regions in a visible light image; Based on the atmospheric light value and dust transmittance estimation model of different brightness regions in visible light images, the dust transmittance of bright and dark channels in different brightness regions of visible light images is calculated. A fitness function is constructed based on the image energy gradient, a particle swarm parameter optimization model is established, and the optimal value of the dust transmittance weight in the bright and dark channels is calculated. Based on the optimal weight of dust transmittance in the bright and dark channels, and the dust transmittance in the bright and dark channels of different brightness regions in the visible light image, the dust transmittance of the visible light image is obtained. A clear visible light image is obtained based on the dust transmittance of the visible light image; Based on the dust transmittance of the visible light image, the dust transmittance of the infrared image is obtained by mapping, and an infrared temperature measurement compensation model is established based on the dust transmittance of the infrared image. Based on clear visible light images and infrared temperature compensation models, a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns is constructed.

2. The method for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns according to claim 1, characterized in that, Calculating atmospheric light values ​​in different brightness regions of a visible light image includes: The visible light image was divided into an extremely bright central flame region using dual threshold segmentation. Extremely dark edge areas and the middle area ; Based on the prior of the bright channel, the atmospheric light value of the bright channel is calculated in the central flame region and the middle region of the visible light image, and is used as the atmospheric light value of the central flame region. The specific calculation formula is as follows: ， in, The atmospheric light value of the central flame region. Represents a set The set of pixels with the highest grayscale values, representing the top 0.1% of all pixels. express The average grayscale value; Based on the dark channel prior, the atmospheric light values ​​of the dark channel are calculated in the edge and middle regions of the visible light image. As the atmospheric light value of the edge region, the specific calculation formula is as follows: ， in, This represents the atmospheric light value in the dark channel. Represents a set The set of pixels with the highest grayscale values, representing the top 0.1% of all pixels. express The average grayscale value; The atmospheric light values ​​of the bright channel and the dark channel are averaged to obtain the atmospheric light value of the intermediate region. The specific calculation formula is as follows: 。 3. The method for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns according to claim 2, characterized in that, Based on the atmospheric light value and dust transmittance estimation model for different brightness regions in visible light images, the formula for calculating the dust transmittance of bright and dark channels in different brightness regions of visible light images is as follows: ， ， in, and These represent the dust transmittance in the dark and bright channels of a visible light image, respectively. and These are the dark and bright channels of a visible light image, respectively. For pixels in a visible light image, To preserve empirical values ​​for image depth of field.

4. The method for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns according to claim 3, characterized in that, Based on the image energy gradient, a fitness function is constructed, a particle swarm optimization model is established, and the optimal values ​​of dust transmittance weights in the bright and dark channels are calculated, including: Construct the fitness function, where the fitness function is specifically: ， in, For the fitness function, The original image was affected by dust and fog. Visible light image, and These represent the dust transmittance in the dark and bright channels of a visible light image, respectively. and These are the weighted parameters and optimal values ​​for dust transmittance in the bright and dark channels, respectively. This is the sum of boundary intensity values ​​after edge detection of a visible light image using the Sobel operator. The entropy of a visible light image. This represents the energy gradient value of the visible light image. Based on the fitness function, the parameters of the dust transmittance weights in the bright and dark channels of the visible light imaging path are optimized to obtain the optimal values ​​of the dust transmittance weights in the bright and dark channels of the visible light imaging path.

5. The method for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns according to claim 4, characterized in that, The specific formula for obtaining a clear visible light image based on the dust transmittance of a visible light image is as follows: ， in For clear images in visible light, The dust transmittance of the visible light image, and , The original image is affected by dust and fog.

6. The method for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns according to claim 1 or 5, characterized in that, Based on the dust transmittance of the visible light image, the dust transmittance of the infrared image is obtained by mapping, and an infrared thermometry compensation model is established based on the dust transmittance of the infrared image, including: By using coordinate mapping, the dust transmittance of an infrared image is obtained by mapping the dust transmittance of a visible light image. The specific calculation formula is as follows: ， in and Dust transmission in infrared and visible light images, respectively. , This indicates the relative position of the visible light camera to the world coordinate system. , Indicates the relative position of the infrared camera to the world coordinate system; The dust transmittance of the infrared image is corrected based on the baseline distance between the visible light camera and the infrared camera, the distance between the baseline center and the target, and the size of the infrared image. Based on the dust transmittance of the corrected infrared image, an infrared thermometry compensation model is established, specifically as follows: ； in, To achieve the target true temperature, Infrared temperature measurement value under dust interference. This indicates the temperature difference in infrared images under dust interference. These are the fitting parameters, This represents the dust transmittance of the corrected infrared image.

7. The method for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns according to claim 6, characterized in that, The dust transmittance correction of the infrared image is performed based on the baseline distance between the visible light camera and the infrared camera, the distance between the baseline center and the target, and the size of the infrared image. The correction coefficients are obtained based on the baseline distance between the visible light camera and the infrared camera, the distance between the baseline center and the target, and the size of the infrared image. Specifically: ， in, For correction factors, The dust transmittance of the infrared image before correction. The baseline distance between the visible light camera and the infrared camera. The distance between the baseline center and the target. and The width and height of the infrared image; The dust transmittance of the infrared image is corrected based on the correction coefficient. The specific formula is as follows: ， in, This represents the dust transmittance of the corrected infrared image.

8. The method for constructing a three-dimensional temperature field on the surface of materials in industrial furnaces and kilns according to claim 7, characterized in that, Based on clear visible light images and an infrared temperature compensation model, a three-dimensional temperature field is constructed on the surface of materials in industrial furnaces, including: Stereo matching is performed between clear visible light images and infrared images to obtain parallax images; The parallax image is converted into a depth image, and the coordinate transformation is performed on the depth image to obtain a three-dimensional point cloud of the material surface in the industrial furnace. The target real temperature output by the infrared temperature measurement compensation model is mapped onto a three-dimensional point cloud through coordinate mapping to obtain the three-dimensional temperature field on the surface of materials in industrial furnaces and kilns.

9. A system for constructing a three-dimensional temperature field on the surface of materials in an industrial furnace, the system comprising: The memory (10), the processor (20), and the computer program stored in the memory (10) and executable on the processor (20) are characterized in that the processor (20) implements the steps of the method according to any one of claims 1 to 8 when executing the computer program.

Images