Method, device and medium for simultaneously reconstructing three-dimensional temperature and spectral emissivity field of curved surface

Abstract

This invention discloses a method, device, and medium for simultaneous reconstruction of the three-dimensional temperature and spectral emissivity fields of a curved surface. The reconstruction method includes the following steps: 1) Based on the spectral light field image of the target surface, a series of refocused images focused at different depths and full-focused images at various wavelengths are generated band by band; 2) Based on the brightness distribution of the full-focused images, feature points are identified layer by layer along the target axis from the full-focused images, and the radial position and focal plane depth corresponding to the feature points are recorded; 3) The radial contour curve of the target is fitted layer by layer; 4) The three-dimensional model of the target is obtained based on the contour curve; 5) Based on the full-focused images of each band, the temperature and emissivity of each pixel are calculated and matched with the three-dimensional model of the target, thereby forming the three-dimensional temperature field and spectral emissivity field of the target surface. This invention overcomes the accuracy limitations imposed by the preset emissivity in traditional temperature reconstruction methods and expands the application field of spectral light field technology to the reconstruction of the physical properties of engineered three-dimensional targets.

Description

Technical Field

[0001] This invention relates to a method for simultaneously reconstructing the three-dimensional temperature and spectral emissivity field of a curved surface based on spectral light field images, belonging to the field of multispectral measurement. Background Technology

[0002] Curved surface components are widely used in various machines and equipment in the energy sector, such as the furnace and slag hopper of high-temperature boilers, and the turbine blades and combustion chamber of gas turbines. These curved surface components operate in high-temperature environments for extended periods, making them susceptible to thermal fatigue, thermal erosion, fracture, and other thermal stress damage. Therefore, achieving continuous monitoring of the surface temperature of curved surface components to ensure their safe and stable operation and to provide early warning of potential accidents is an ongoing research topic in the engineering field.

[0003] Radiation thermometry is a non-contact temperature measurement technology that inverts the target temperature based on the thermal radiation intensity of the target surface. Due to its wide measurement range, fast response, and lack of interference with the measured surface, it is widely used in high-temperature monitoring in engineering fields. In principle, the thermal radiation intensity of a target is positively correlated with both its spectral emissivity and temperature. However, a challenge in radiation thermometry lies in the fact that the target's spectral emissivity is unknown. In engineering applications, spectral emissivity is usually estimated empirically, which severely limits the accuracy of radiation thermometry. Currently, multispectral thermometry is commonly used to solve this problem. Multispectral imaging technology acquires two-dimensional projections of the target's spectral radiation intensity at multiple wavelengths pixel-by-pixel or wavelength-by-wavelength, thereby simultaneously reconstructing the target's two-dimensional projected temperature and spectral emissivity field. However, for the challenging three-dimensional temperature measurement of curved surfaces in engineering, traditional two-dimensional spectral imaging acquisition techniques are not suitable due to hardware limitations. Various three-dimensional multispectral imaging devices are still under rapid development.

[0004] Spectral light field imaging technology is a novel instantaneous three-dimensional multispectral imaging device. A single exposure (i.e., a single image) can record five-dimensional information about the target, including light intensity, direction, and spectral information across the entire frame. Therefore, this technology can acquire complete spectral information of a target on a three-dimensional surface, showing great potential for applications in measuring three-dimensional physical properties such as temperature and spectral emissivity of curved surfaces. However, the spectral light field images acquired by this technology are highly complex, and data decoding is difficult. Direct extraction of spectral data often suffers from spectral aliasing and depth aliasing, failing to fully utilize the multidimensional spectral light field information for accurate reconstruction of three-dimensional surface properties. Therefore, this technology is typically only used to measure two-dimensional planar properties, wasting the directional information of the light rays. In summary, three-dimensional surface measurement technology based on spectral light field imaging still requires further development and improvement. Summary of the Invention

[0005] The technical problem this invention aims to solve is to simultaneously reconstruct the three-dimensional temperature and emissivity fields of a surface based on multidimensional information from spectral light field images, in order to meet the urgent engineering need for monitoring the temperature and emissivity fields of three-dimensional curved surfaces. This invention proposes a method for simultaneously reconstructing the three-dimensional temperature and spectral emissivity fields of a surface, integrating spectral radiation data extraction, surface model reconstruction, and surface property reconstruction.

[0006] A method for simultaneously reconstructing the three-dimensional temperature and spectral emissivity fields of a curved surface based on spectral light field images includes the following steps:

[0007] Step 1: Based on the spectral light field image of the target surface, generate a series of refocusing images focused at different depths for each band; generate a refocusing depth distribution map and a full-focus image at each wavelength based on the refocusing images.

[0008] Step 2: Calculate the average pixel brightness of the full-focus image at each wavelength; based on the full-focus image with the highest average pixel brightness, find the feature points layer by layer along the target axis in the full-focus image, and record the radial position and focal plane depth corresponding to the feature points.

[0009] Step 3: Based on the radial position and focal plane depth corresponding to the feature points, fit the radial surface contour curve of the target layer by layer;

[0010] Step 4: Based on the fitted surface contour curve, calculate the fitting space depth of each pixel to obtain the complete 3D model of the target.

[0011] Step 5: Based on the full-focus images of each band, calculate the temperature and emissivity of each pixel and match them with the target 3D model to form the 3D temperature field and spectral emissivity field of the target surface.

[0012] In step one, the refocused images, full-focused images, and refocusing depth distribution maps at various wavelengths are obtained according to the imaging system and image processing method described in Chinese Invention Patent Publication No. CN115307732A.

[0013] Spectral light field images contain Radiation information at each wavelength, denoted as . Get the total Zhang Focus Refocusing images at different depths One full-focus image and one refocusing depth map. The refocusing image, full-focus image, and refocusing depth map have the same resolution. Each pixel corresponds one-to-one. The refocusing depth distribution map records the position and focal plane depth of each pixel in the fully focused image. The corresponding radial position, axial position, and focal depth are denoted as follows: , and .

[0014] In step two, the target surface is divided into several layers along the axis, with each layer having a relative height of [missing information]. The pixel; In the layer, if pixels eigenvalues Greater than the specified threshold If it is, then it is considered a feature point. ; Eigenvalues Calculate using the following function:

[0015] (1)

[0016] (2)

[0017] (3)

[0018] in, as well as pixels Radial and vertical cross-differential operators. Threshold. The number of feature points in each layer must be guaranteed. satisfy The constraints.

[0019] No. In the layer, a total of ( ) feature points, denoted as . No. Feature points The radial position and focal depth are read from the depth distribution map in step one, and denoted as... and .

[0020] In step three, an orthogonal triangular decomposition algorithm is used, based on the radial position of the feature points. and focal depth , Fitting the first The pixel depth in a layer is a function of the pixel's radial position, i.e., the contour curve. The contour curve is shown below.

[0021] (4)

[0022] In the formula, All are fitting coefficients; Representing pixels The depth of the fitting space, It is a pixel radial position

[0023] In step four, the radial position of each pixel is substituted into the contour curve of the layer (i.e., Equation 4) to calculate the fitting space depth of each pixel in order to obtain the three-dimensional model of the surface.

[0024] The three-dimensional model of the target surface is represented as a composite 3D cell array containing the radial position, axial position, and depth of the fitted space of each pixel. .

[0025] (5)

[0026] in, Cell array The first in Each element represents a pixel. The radial position, axial position, and fitting space depth.

[0027] In step five, the non-negative linear-square algorithm is used to solve the following system of equations to obtain the pixels. Temperature and emissivity:

[0028] (6)

[0029] in, Represents the first radiation constant; Represents the second radiation constant; Represents pixels At wavelength The radiation intensity at a given wavelength can be determined by the wavelength. The full-focus image is acquired directly. Pixels At wavelength The desired emission rate is as follows; Represents pixels The corresponding temperature value to be determined; The coefficients to be determined in the system of equations are denoted as . The system of equations is solved pixel by pixel to obtain the spectral emissivity field of the surface at three-dimensional temperature.

[0030] In step five, the three-dimensional temperature field of the target surface can be represented as a composite 4D cell array containing the radial position, axial position, fitting space depth, and temperature of each pixel. .

[0031] (7)

[0032] in, Cell array The first in Each element represents a pixel. The radial position, axial position, fitting space depth, and temperature.

[0033] The three-dimensional spectral emissivity field of the target surface can be represented as a composite field containing the radial position, axial position, fitting space depth, and spectral emissivity at each wavelength of each pixel. dimensional cell array .

[0034] (8)

[0035] in, Cell array The first in Each element represents a pixel. The radial position, axial position, fitting space depth, and spectral emissivity at each wavelength.

[0036] The present invention also provides an electronic device, comprising:

[0037] One or more processors;

[0038] Memory, used to store one or more programs;

[0039] When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the above-described method for simultaneous reconstruction of the three-dimensional temperature and spectral emissivity fields of curved surfaces.

[0040] The present invention also provides a storage medium storing a computer program that, when executed by a processor, implements the steps of the method for simultaneous reconstruction of the three-dimensional temperature and spectral emissivity field of a curved surface as described above.

[0041] Beneficial effects: Compared with existing methods for reconstructing surface temperature and emissivity fields, this invention has the following advantages:

[0042] 1) The method for reconstructing the three-dimensional temperature and emissivity field of curved surfaces in this invention, based on digital image processing technology, fully utilizes the light intensity, light direction, and spectral information in multi-dimensional spectral light field images. This invention, based on the extraction of spectral radiation data (step one of the method), couples three-dimensional model reconstruction (steps three to five of the method) and three-dimensional property reconstruction (step six of the method), expanding the application of spectral light field technology from the instantaneous measurement of the spectral distribution of two-dimensional targets to the property reconstruction of engineered three-dimensional targets.

[0043] 2) This invention balances the efficiency and effectiveness of spectral data processing. By evaluating the brightness of fully focused images and fitting the contours of target layers, it not only effectively avoids the errors and spectral aliasing caused by repeated calculations of spectral information, but also greatly improves the processing efficiency of spectral information.

[0044] 3) The three-dimensional temperature and emissivity field reconstruction method of the present invention achieves synchronous reconstruction of temperature and emissivity fields, greatly improving the accuracy of temperature field reconstruction and breaking through the accuracy limitation caused by the preset emissivity in traditional temperature reconstruction methods. Moreover, the synchronous reconstruction of temperature and emissivity fields is based on only a single spectral light field image, which eliminates the cumbersome process of acquiring data multiple times in advance compared to traditional scanning spectral methods. Attached Figure Description

[0045] Figure 1 This is a flowchart of a method for simultaneously reconstructing the three-dimensional temperature and spectral emissivity fields of a curved surface based on spectral light field images.

[0046] Figure 2 This is an overview of the experimental system used to acquire spectral light field images of target surfaces.

[0047] Figure 3 Schematic diagram of the target curved surface structure and thermocouple pre-embedded points.

[0048] Figure 4 The image shows the spectral field of the target surface and the corresponding spectral structure diagram.

[0049] Figure 5 Example of a reconstructed three-dimensional temperature field for the target surface.

[0050] Figure 6 An example of the three-dimensional emissivity field at a wavelength of 1065 nm for the reconstructed target surface. Detailed Implementation

[0051] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments:

[0052] This invention discloses a method for simultaneously reconstructing the three-dimensional temperature and spectral emissivity field of a target surface based on spectral light field imaging, comprising the following steps:

[0053] Step 1: Based on the spectral light field image of the target surface, generate a series of refocusing images focused at different depths for each band; generate full-focus images and refocusing depth distribution maps at each wavelength based on the refocusing images.

[0054] Figure 2 An experimental system for acquiring the spectral light field of a target curved surface. The target curved surface is a cylindrical metal surface, 100 mm high and 100 mm in diameter. A ceramic heating element array is attached to the back of the surface, serving as the heating source to simulate a high-temperature surface under operating conditions. Thermocouple pre-embedded points are provided within the surface, such as... Figure 3As shown. The temperature measurement value of the pre-embedded thermocouple is used as the true temperature value of the curved surface. The spectral light field image of the curved surface is acquired using the near-infrared spectral light field imaging system described in Chinese Invention Patent Publication No. CN115307732A, and used as material for reconstructing the three-dimensional temperature and emissivity field of the curved surface.

[0055] Figure 4 This image shows the spectral field image of the target surface and a schematic diagram of its corresponding spectral structure. The imaging system can simultaneously acquire spectral field images at four wavelengths (1064 nm, 1250 nm, 1535 nm, 1652 nm).

[0056] In this embodiment, the spectral light field image includes Radiation information at each wavelength, denoted as . Get the total Zhang focused on different wavelengths. Refocusing images at different depths. Based on the refocusing images, a [database / image] can be generated. One full-focus image in different bands, and one refocus depth distribution map.

[0057] The refocus depth map records the position and focal plane depth of each pixel in a fully focused image. The corresponding radial position, axial position, and focal depth are denoted as follows: , and The refocused image, the full-focused image, and the refocused depth map have the same resolution and are denoted as... Each pixel corresponds to a specific pixel.

[0058] Step 2: Calculate the average pixel brightness of the full-focus image at each wavelength. Based on the full-focus image with the highest average pixel brightness, identify feature points layer by layer along the target axis in the full-focus image, and record the radial position information and focal plane depth information corresponding to the feature points.

[0059] In this step, the target surface is divided into several layers along the axis, with each layer having a relative height of [missing information]. In this case, [number] pixels. ;

[0060] No. In the layer, if pixels eigenvalues Greater than the specified threshold , If , then it is considered a feature point, and its eigenvalue is... Calculate using the following function:

[0061] (1)

[0062] (2)

[0063] (3)

[0064] in, as well as pixels Radial and axial cross-differential operators. Threshold. The number of feature points in each layer must be guaranteed. satisfy The constraints. In this embodiment... (Pixel brightness value).

[0065] No. In the layer, a total of ( ) feature points, denoted as . No. Feature points The radial position and focal depth are read from the depth distribution map in step one, and denoted as... and In this embodiment, taking the first layer as an example, the first layer obtains a total of 7 feature points, denoted as... .

[0066] Step 3: Based on the radial position and focal plane depth corresponding to the feature points, fit the radial surface contour curve of the target layer by layer.

[0067] In this step, an orthogonal triangular decomposition algorithm is used, based on the radial position of the feature points. and focal depth , Fitting the first The pixel depth in a layer is a function of the pixel's radial position, i.e., the surface profile curve. The surface profile curve is shown below:

[0068] (4)

[0069] In the formula, All are fitting coefficients; Representing pixels The depth of the fitting space, It is a pixel The radial position.

[0070] Step 4: Based on the fitted surface contour curve, calculate the fitting space depth of each pixel to obtain the complete 3D model of the target.

[0071] In this step, the radial position information of each pixel is input into the contour curve of the layer (i.e., Equation 4) to calculate the fitting space depth of each pixel in order to obtain the three-dimensional model of the surface.

[0072] The three-dimensional model of the target surface is represented as a composite 3D cell array containing the radial position, axial position, and depth of the fitted space of each pixel. .

[0073] (5)

[0074] in, Cell array The first in One element; Representing pixels The radial position, axial position, and fitting space depth.

[0075] Step 5: Based on the full-focus images of each band, calculate the temperature and emissivity of each pixel and match them with the target 3D model to form the 3D temperature field and spectral emissivity field of the target surface.

[0076] In this step, the non-negative linear-square algorithm is used to solve the following system of equations to obtain the pixels. Temperature and emissivity:

[0077] (6)

[0078] in, This represents the first radiation constant, which is 3.7418 × 10⁻⁶. -16 W∙m 2 ; This represents the second radiation constant, which is 1.4388 × 10⁻⁶. -2 m∙K; Represents pixels At wavelength The radiation intensity at a given wavelength can be determined by the wavelength. The full-focus image is acquired directly. Pixels At wavelength The desired emission rate is as follows; Represents pixels The corresponding temperature value to be determined; The coefficients to be determined in the system of equations are denoted as . The system of equations is solved pixel by pixel to obtain the spectral emissivity field of the surface at three-dimensional temperature.

[0079] The three-dimensional temperature field of the target surface can be represented as a composite 4-dimensional cell array containing the radial position, axial position, fitting space depth, and temperature of each pixel. .

[0080] (7)

[0081] in, Cell array The first in Each element represents a pixel. The radial position, axial position, fitting space depth, and temperature of the reconstructed three-dimensional temperature field of the curved surface are as follows: Figure 5 As shown.

[0082] The three-dimensional spectral emissivity field of the target surface can be represented as a composite field containing the radial position, axial position, fitting space depth, and spectral emissivity at each wavelength of each pixel. dimensional cell array .

[0083] (8)

[0084] in, Cell array The first in Each element represents a pixel. The radial position, axial position, fitting space depth, and spectral emissivity at various wavelengths are determined. The reconstructed surface emissivity field at 1064 nm wavelength is shown below. Figure 6 As shown.

Claims

1. A method for simultaneously reconstructing the three-dimensional temperature and spectral emissivity field of a curved surface, characterized in that, Includes the following steps: Step 1: Based on the spectral light field image of the target surface, generate a series of refocusing images focused at different depths for each band; generate a refocusing depth distribution map and a full-focus image at each wavelength based on the refocusing images. Step 2: Calculate the average pixel brightness of the full-focus image at each wavelength; based on the full-focus image with the highest average pixel brightness, find the feature points layer by layer along the target axis in the full-focus image, and record the radial position and focal plane depth corresponding to the feature points. Step 3: Based on the radial position and focal plane depth corresponding to the feature points, fit the radial surface contour curve of the target layer by layer; Step 4: Based on the fitted surface contour curve, calculate the fitting space depth of each pixel to obtain the complete 3D model of the target. Step 5: Based on the full-focus images of each band, calculate the temperature and emissivity of each pixel and match them with the target 3D model to form the 3D temperature field and spectral emissivity field of the target surface.

2. The method for simultaneous reconstruction of three-dimensional temperature and spectral emissivity fields of curved surfaces according to claim 1, characterized in that, In step two, the target is divided into several layers along the axis, with each layer having a relative height of [missing information]. 1 pixel, ;No. In the layer, if pixels eigenvalues Greater than the specified threshold If it is, then it is considered a feature point. ; Eigenvalues Calculate using the following function: in, as well as pixels Radial and axial cross differential operators; No. In the layer, a total of Let there be 1 feature point, and denote it as _ ... ; No. Feature points The radial position and focal depth are read from the depth distribution map in step one, and denoted as... and .

3. The method for simultaneous reconstruction of three-dimensional temperature and spectral emissivity fields of curved surfaces according to claim 1, characterized in that, In step three, an orthogonal trigonometric decomposition algorithm is used to fit the first trigonometric decomposition based on the radial position of the feature points and the focal plane depth. The pixel depth in a layer is a function of the pixel's radial position, i.e., the contour curve. ; The contour curve is shown below: In the formula, All are fitting coefficients; Representing pixels The depth of the fitting space, It is a pixel The radial position.

4. The method for simultaneous reconstruction of three-dimensional temperature and spectral emissivity fields of curved surfaces according to claim 3, characterized in that, In step four, the radial position information of each pixel is input into the contour curve of the layer to calculate the fitting space depth of each pixel.

5. The method for simultaneous reconstruction of three-dimensional temperature and spectral emissivity fields of curved surfaces according to claim 4, characterized in that, In step four, the obtained 3D model of the surface is a composite 3D cell array containing the radial position, axial position, and depth of the fitted space of each pixel. : in, Cell array The first in Each element.

6. The method for simultaneous reconstruction of three-dimensional temperature and spectral emissivity fields of a curved surface according to claim 5, characterized in that, In step five, the non-negative least squares algorithm is used to solve the following system of equations to obtain the pixels. Temperature and emissivity: in, Represents the first radiation constant; Represents the second radiation constant; Represents pixels At wavelength The radiation intensity at a given wavelength can be determined by the wavelength. The full-focus image is acquired directly. Pixels At wavelength The desired emission rate is as follows; Represents pixels The corresponding temperature value to be determined; The coefficients to be determined for the system of equations are... , This refers to the number of wavelengths.

7. The method for simultaneous reconstruction of three-dimensional temperature and spectral emissivity fields of curved surfaces according to claim 6, characterized in that, In step five, the three-dimensional temperature field of the target surface is represented as a composite 4D cell array containing the radial position, axial position, fitting space depth, and temperature of each pixel. : in, Cell array The first in One element; The three-dimensional temperature field of the target surface is represented as a composite field containing the radial position, axial position, fitting space depth, and spectral emissivity of each pixel at various wavelengths. dimensional cell array : in, Cell array The first in Each element.

8. The method for simultaneous reconstruction of the three-dimensional temperature and spectral emissivity field of a curved surface according to any one of claims 1-7, characterized in that, Spectral light field images contain Radiation information at each wavelength, denoted as . ,total Zhang Focus Refocusing images at different depths; based on the refocusing images, generate One full-focus image and one refocusing depth map; the refocusing image, full-focus image, and refocusing depth map have the same resolution. Each pixel corresponds one-to-one; the refocusing depth map records the position and focal plane depth of each pixel in the fully focused image. The corresponding radial position, axial position, and focal depth are denoted as follows: , and .

9. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the method for simultaneous reconstruction of the three-dimensional temperature and spectral emissivity field of a surface as described in any one of claims 1-8.

10. A storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method for simultaneous reconstruction of the three-dimensional temperature and spectral emissivity field of a curved surface as described in any one of claims 1 to 8.

Images