A cast cooling temperature field reconstruction method and system based on an infrared thermal image
By performing grayscale truncation and geometric correction on infrared thermal images, and utilizing the geometric convergence index and mutual radiation gain coefficient, the problem of mutual radiation interference during the cooling process of complex castings was solved, enabling accurate reconstruction and monitoring of the casting temperature field.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- XIAN ISE MACHINERY CO LTD
- Filing Date
- 2026-03-12
- Publication Date
- 2026-06-09
AI Technical Summary
Existing infrared temperature measurement technology suffers from mutual radiation interference during the cooling process of castings with complex geometries, leading to temperature measurement distortion and affecting the adjustment of cooling process parameters and casting quality.
By acquiring infrared thermal images of the casting during the cooling process, and using methods such as grayscale histogram truncation, geometric convergence index, local energy concentration degree, and mutual radiation gain coefficient, the infrared thermal images are corrected to restore the true surface temperature of the casting.
This enables a more accurate reconstruction of the temperature field during the cooling process of castings, improves the accuracy of temperature monitoring, and reduces the risk of casting scrap rate and substandard performance.
Smart Images
Figure CN121860904B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of image processing technology, and in particular to a method and system for reconstructing the cooling temperature field of castings based on infrared thermography. Background Technology
[0002] Infrared thermal imaging technology, with its advantages of being non-contact, having a fast response speed, and being able to measure temperature across the entire field, has been applied to the monitoring of the cooling process in the metal casting industry. During the cooling process of castings, real-time and accurate temperature field distribution data plays a crucial role in controlling the cooling rate, preventing the generation of hot cracks, and optimizing the distribution of residual stress.
[0003] However, when performing infrared thermometry on castings with complex geometries, there is often a serious problem of mutual radiation interference. Since the surface of the casting usually has a high temperature and the metal material has certain reflective properties in the liquid or semi-solid state, the concave areas, internal cavities, or adjacent high-temperature components of the casting itself will form a blackbody-like cavity effect. The target pixel in this area not only radiates its own infrared energy, but also reflects the radiation energy from the neighboring high-temperature surface. This mutual radiation effect caused by the geometry will make the radiation energy received by the thermal imager higher than the radiation energy corresponding to the actual temperature of the target pixel, resulting in an artificially high measured temperature value.
[0004] Infrared thermometry methods typically only correct the entire image using a uniform emissivity or perform simple background temperature subtraction. They cannot perceive the complex influence of the local micro-geometry of the casting on the radiation transmission path, and it is difficult to eliminate the mutual radiation noise caused by the geometric structure and local energy accumulation.
[0005] If this mutual radiation interference cannot be effectively removed, the acquired temperature field data may be distorted, causing the control system to misjudge the cooling state, which in turn leads to incorrect adjustment of cooling process parameters, resulting in an increase in casting scrap rate or failure to meet performance standards. Therefore, it is necessary to reconstruct the temperature field of the casting more accurately during the cooling process. Summary of the Invention
[0006] To more accurately reconstruct the temperature field of castings during the cooling process, this application provides a method and system for reconstructing the cooling temperature field of castings based on infrared thermography.
[0007] According to a first aspect of the embodiments of this application, a method for reconstructing the cooling temperature field of a casting based on an infrared thermal image is provided, comprising: acquiring a first infrared thermal image of the casting during the cooling process; determining the cumulative probability quantile of the grayscale histogram of the first infrared thermal image; truncating the first infrared thermal image according to the cumulative probability quantile to obtain a second infrared thermal image; constructing the grayscale gradient vector of the neighboring pixels of a target pixel in the second infrared thermal image; determining the position vector pointing from the neighboring pixels to the target pixel; determining the geometric convergence index of the target pixel using the cosine of the angle between the grayscale gradient vector and the position vector, and the logarithmic weighting term of the brightness of the neighboring pixels relative to the average brightness of the second infrared thermal image; determining the local energy concentration of the target pixel based on the grayscale extreme values in the neighborhood; determining the cross-radiation gain coefficient of the target pixel based on the geometric convergence index, the local energy concentration, and the reflection coupling coefficient of the casting surface; determining the equivalent intrinsic grayscale of the target pixel based on the ratio of the target pixel to the average grayscale of the neighborhood and the cross-radiation gain coefficient; and converting the equivalent intrinsic grayscale to obtain the surface temperature information of the casting.
[0008] This allows for a more accurate reconstruction of the temperature field of the casting during the cooling process based on infrared images, facilitating subsequent processing of the casting by combining the obtained temperature information.
[0009] Optionally, the second infrared thermal image is obtained by truncating the first infrared thermal image based on the cumulative probability quantiles, including: using the cumulative probability quantiles of the first proportion of the grayscale histogram as a lower threshold and the cumulative probability quantiles of the second proportion as an upper threshold; the first proportion is less than the second proportion; assigning the upper threshold to pixels in the first infrared thermal image that are above the upper threshold, assigning the lower threshold to pixels in the first infrared thermal image that are below the lower threshold, and retaining the pixel values of pixels in the first infrared thermal image that are between the upper and lower thresholds to obtain the second infrared thermal image.
[0010] In this way, by using dual threshold truncation, the high-brightness noise caused by specular reflection and the low-temperature noise in the background environment can be effectively filtered out, so that the subsequent processing can focus on the effective temperature grayscale range of the casting and improve the robustness of the algorithm.
[0011] Optionally, before determining the geometric convergence index, the method further includes: calculating the absolute difference between the target pixel and its four neighboring pixels, constructing an adaptive weighting function based on the absolute difference; the adaptive weighting function is negatively correlated with the square of the absolute difference; using the adaptive weighting function to smooth the second infrared thermogram, and using the filtered image as the second infrared thermogram again.
[0012] This allows for the smoothing of thermal noise in the image while preserving the edge features and texture details of the casting surface to the maximum extent, preventing the loss of geometric features due to over-smoothing.
[0013] Optionally, the geometric convergence index of the target pixel is determined by the following formula: ,in, Geometric convergence index; For the neighborhood Total number of pixels within; For neighboring pixels The gray-level gradient vector; To obtain from neighboring pixels The position vector pointing to the target pixel; Represents the magnitude of a vector; Represents the dot product of vectors; To correct the linear unit function; For neighboring pixels grayscale value; This represents the average grayscale value of the entire infrared thermal image. It is the natural logarithm function. It is a natural constant.
[0014] In this way, the vector dot product represents whether the heat flow direction of the neighboring pixels points to the target pixel. The ReLU function is used to filter out the components that have a positive radiation contribution to the target pixel. The relative intensity of the neighborhood energy is considered by combining the logarithmic weighting term. The constructed geometric convergence index can represent the degree of depression or convexity of the local micro geometry.
[0015] Optionally, the modified linear unit function is used to retain the component with a positive cosine value of the angle between the gray-level gradient vector and the position vector, and to set the component with a negative cosine value of the angle to zero.
[0016] Optionally, the local energy concentration of the target pixel is determined by: using the statistical maximum and minimum gray values of the second infrared thermal image, mapping the gray values of each pixel in the neighborhood to a normalized interval; calculating the square of the normalized gray values of each pixel, and using the sum of the square values of all pixels in the neighborhood as the local energy concentration.
[0017] In this way, by normalizing and summing the squares, the contribution weight of high-energy pixels to the local radiation field is highlighted, which can sensitively capture the energy accumulation effect in hot spots.
[0018] Optionally, the mutual radiation gain coefficient is determined by the following formula: ,in, The mutual radiation gain coefficient; The reflection coupling coefficient of the material surface is denoted as . Geometric convergence index, The preset geometric weight adjustment index, For local energy concentration, The preset sensitivity threshold, It is a natural exponential function.
[0019] In this way, by comprehensively considering material properties, geometric structure factors, and energy density factors, and utilizing the saturation characteristics of exponential functions to simulate the nonlinear laws of radiative heat exchange, the calculated gain coefficient is closer to the real physical process.
[0020] Optionally, the equivalent intrinsic gray level is determined by the following formula: ,in, The equivalent intrinsic gray level of the target pixel; The grayscale value of the target pixel; The mutual radiation gain coefficient; The average gray value of the neighborhood of the target pixel. It is a preset positive number.
[0021] In this way, by reverse derivation, the increment caused by mutual radiation can be separated from the total received radiation of the observed gray level, thereby recovering the intrinsic gray level value determined only by the temperature of the target pixel itself.
[0022] Optionally, the equivalent intrinsic grayscale is converted to obtain the surface temperature information of the casting, including: taking the difference between the equivalent intrinsic grayscale and the dark current bias value of the infrared thermal imager system as the net radiation response value; taking the product of the photoelectric response parameters of the infrared thermal imager system, the emissivity of the casting surface, and the transmittance of the environment as the system response coefficient; and using the ratio of the net radiation response value to the system response coefficient to determine the temperature information of the target pixel to obtain the surface temperature information of all surface pixels in the casting.
[0023] According to a second aspect of the present application, a casting cooling temperature field reconstruction system based on infrared thermal imaging is provided, comprising: a processor and a memory, wherein the memory stores computer program instructions, and the computer program instructions, when executed by the processor, implement the steps of the casting cooling temperature field reconstruction method based on infrared thermal imaging provided in the first aspect of the present application.
[0024] The technical solutions provided by the embodiments of this application may include the following beneficial effects: acquiring a first infrared thermal image of the casting during the cooling process, and truncating the first infrared thermal image to obtain a second infrared thermal image; by constructing the geometric convergence index of the pixels in the second infrared thermal image, the geometric perspective of radiation transmission can be simulated using the angle relationship between the gradient vector and the position vector; combined with the local energy concentration, the thermal radiation interference intensity of the high-temperature energy concentration area on the measurement point can be adaptively identified; by decoupling and correcting the original grayscale through the mutual radiation gain coefficient, the equivalent intrinsic grayscale representing the true temperature of the casting is restored, eliminating the mutual radiation influence between complex surfaces, and obtaining a more accurate temperature field on the surface of the casting, which helps to improve the accuracy of temperature field monitoring during the casting cooling process.
[0025] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description
[0026] Figure 1 This is a flowchart illustrating a method for reconstructing the cooling temperature field of a casting based on an infrared thermogram, according to an exemplary embodiment.
[0027] Figure 2 This is the original infrared thermogram of the casting before correction during the cooling process;
[0028] Figure 3 This is a schematic diagram of the surface temperature field after correction of the original infrared thermal image;
[0029] Figure 4 It is a comparison chart of the temperature data of key sections of the original infrared thermal image and the reconstructed temperature field.
[0030] Figure 5 This is a schematic diagram illustrating the structure of a casting cooling temperature field reconstruction system based on infrared thermograms, according to an exemplary embodiment. Detailed Implementation
[0031] First, a brief introduction to the application scenario of the embodiments of this application will be given. The application scenario of the embodiments of this application can be the monitoring of the cooling process of large, complex, thin-walled castings on the casting production line.
[0032] When a casting is first demolded or is in the cooling stage after mold opening, its surface temperature is usually between 300 and 800 degrees Celsius. Due to the complex structure of the casting, there may be a large number of reinforcing ribs, grooves and internal cavities on the surface. These structures will form a strong cavity mutual radiation effect at high temperatures, which will cause the temperature reading at the concave corner in the image acquired by the infrared thermal imager to be tens of degrees Celsius higher than the actual temperature, interfering with the judgment of the cooling rate of the stress concentration area.
[0033] To address the aforementioned technical problems, embodiments of this application provide a method and system for reconstructing the cooling temperature field of castings based on infrared thermography. Figure 1 This is a flowchart illustrating a method for reconstructing the cooling temperature field of a casting based on an infrared thermogram, according to an exemplary embodiment. Figure 1 As shown, the method includes the following steps.
[0034] In step S101, a first infrared thermal image of the casting during the cooling process is obtained, and the cumulative probability quantile of the grayscale histogram of the first infrared thermal image is determined. The first infrared thermal image is then truncated based on the cumulative probability quantile to obtain a second infrared thermal image.
[0035] The first infrared thermal image can be acquired using an industrial-grade uncooled or cooled infrared focal plane array detector. The acquired first infrared thermal image has a resolution of 640 x 480 pixels and a bit depth of 14 or 16 bits to ensure sufficient dynamic range to capture subtle temperature changes on the surface of the casting.
[0036] After acquiring the first infrared thermal image, the processor can first calculate the grayscale histogram of the entire image; the horizontal axis of the grayscale histogram represents the grayscale value, and the vertical axis represents the frequency of that grayscale value in the image.
[0037] The second infrared thermal image is obtained by truncating the first infrared thermal image based on the cumulative probability quantiles, including: using the cumulative probability quantiles of the first proportion of the grayscale histogram as the lower threshold and the cumulative probability quantiles of the second proportion as the upper threshold; the first proportion is less than the second proportion; assigning the upper threshold value to the pixels in the first infrared thermal image that are above the upper threshold value, assigning the lower threshold value to the pixels in the first infrared thermal image that are below the lower threshold value, and retaining the pixel values of the pixels in the first infrared thermal image that are between the upper threshold value and the lower threshold value, so as to obtain the second infrared thermal image.
[0038] The selection of the first and second percentages, for example, setting the first percentage to 0.02 and the second percentage to 0.98, can automatically remove the influence of the darkest 2% and the brightest 2% of pixels in the image.
[0039] In actual casting environments, the brightest pixels are often caused by sparks from splashing molten metal, specular reflections from lighting equipment, or faulty pixels on detectors; while the darkest pixels usually correspond to a deep background or cold air mass far from the casting. By calculating the cumulative distribution function of the histogram, the gray values corresponding to these two quantiles can be locked. For example, the original infrared thermal image data often contains non-normally distributed impulse noise. The existence of these extreme values will greatly stretch the range of subsequent normalization processing, causing the effective gray value changes of the casting body to be compressed into a very small numerical range, thereby reducing the sensitivity of feature extraction.
[0040] By using statistical histogram truncation, the dynamic range of the image can be adaptively focused within the temperature distribution range of the casting body, which enhances the image contrast and provides a cleaner and more stable data foundation for subsequent gradient-based geometric feature extraction, avoiding the misjudgment of bright noise as a source of high-temperature mutual radiation.
[0041] In one embodiment, before determining the geometric convergence index, the second infrared thermal image can be subjected to anisotropic suppression of high-frequency noise. Specifically, this includes: calculating the absolute difference between the target pixel and its four neighboring pixels; constructing an adaptive weighting function based on the absolute difference; the adaptive weighting function being negatively correlated with the square of the absolute difference; and using the adaptive weighting function to smooth and filter the infrared thermal image to obtain an infrared thermal image for determining the geometric convergence index.
[0042] For any target pixel in an image, its four adjacent pixels above, below, left, and right can be selected to form a four-neighborhood.
[0043] Calculate the absolute value of the difference between the grayscale value of the target pixel and its four neighboring pixels. Let the grayscale value of the target pixel be... The grayscale of a certain neighboring pixel is The absolute value of the difference is .
[0044] The constructed adaptive weight function is, for example, a Gaussian decay function, and the form of the adaptive weight function can be expressed as: ,in This is a constant used to control the decay rate; the smaller the grayscale difference between adjacent pixels and the target pixel, the greater the adaptive weight obtained; the greater the grayscale difference between adjacent pixels and the target pixel, the smaller the adaptive weight obtained.
[0045] Linear filtering, such as mean filtering or Gaussian filtering, can blur the edges of an image while removing thermal noise. However, in the solution of this application, the geometric normal information of the casting surface can be inferred through the gray-level gradient vector. If the edges are blurred, the direction of the gradient will shift, resulting in distortion of the geometric reconstruction.
[0046] Anisotropic diffusion filtering can smooth flat areas while blocking smoothing operations across strong edges. The processed infrared thermogram retains clear casting contours and structural lines, allowing the subsequently calculated gradient vector to accurately reflect the physical undulations of the casting surface, rather than random fluctuations caused by noise.
[0047] In step S102, the gray-level gradient vectors of the neighboring pixels of the target pixel in the second infrared thermal image are constructed, and the position vectors pointing from the neighboring pixels to the target pixel are determined. The geometric convergence index of the target pixel is determined by using the cosine of the angle between the gray-level gradient vector and the position vector, and the logarithmic weighted term of the brightness of the neighboring pixels relative to the average brightness of the second infrared thermal image.
[0048] When constructing grayscale gradient vectors, the Sobel operator, Prewitt operator, or Scharr operator can be used to perform convolution operations on the processed infrared thermal image.
[0049] For the target pixel Its gradient component in the horizontal direction and the gradient component in the vertical direction Together they form the gray-level gradient vector .
[0050] In thermal infrared images, the grayscale gradient direction typically points towards the direction of the greatest rate of temperature change, and on the surface of a casting, it is usually related to the projection direction of the surface normal; for example, at the edge of a groove, the gradient direction often points towards the depth of the groove or a high-temperature region; the position vector is a geometric vector determined based on the image coordinate system; if the target pixel coordinates are... A pixel in the neighborhood The coordinates are Then the position vector .
[0051] The geometric convergence index of the target pixel is determined by the following formula: ,in, Geometric convergence index; For the neighborhood Total number of pixels within; For neighboring pixels The gray-level gradient vector; To obtain from neighboring pixels The position vector pointing to the target pixel; Represents the magnitude of a vector; Represents the dot product of vectors; To correct the linear unit function; For neighboring pixels grayscale value; This represents the average grayscale value of the entire infrared thermal image. It is the natural logarithm function. It is a natural constant.
[0052] The modified linear unit function is used to retain the component with a positive cosine value of the angle between the gray-level gradient vector and the position vector, and to set the component with a negative cosine value of the angle to zero.
[0053] equal to gradient vector With position vector The cosine of the angle between them; when this cosine is positive, the neighboring pixels... The gradient direction or the normal trend of heat flow generally points towards the target pixel. If the cosine value is negative, it means that the gradient of the neighboring pixel is away from the target pixel, that is, there may be a bulge or no direct radiation path between them. Therefore, the contribution of the neighboring pixel is set to zero by the ReLU function.
[0054] The relationship between radiant energy and temperature or grayscale values in thermal infrared images is non-linear. Capable of measuring neighboring pixels The relative heat of the image relative to the background; if Much larger ,but A larger value for this term indicates that the neighboring pixel is a strong radiation source, hence the introduction of a constant. This is to ensure that the input to the logarithmic function is always greater than This ensures that the calculation result is always greater than 1, avoiding negative or zero values that could cause the weights to become invalid.
[0055] This application's embodiments capture the hidden concave features in the image through the geometric relationship between the gradient direction and the position vector; and capture the intensity features of the radiation source through gray-level logarithmic weighting.
[0056] A higher value for the geometric convergence index indicates that the target pixel is located in the center of a concave region surrounded by high-temperature neighbors, and is more likely to be subject to mutual radiation interference. This allows for the recovery of key geometric thermal radiation information using a single 2D thermal image without a 3D scanning model, thus reducing the cost of temperature information reconstruction.
[0057] In step S103, the local energy concentration of the target pixel is determined based on the gray-level extreme values in the neighborhood, and the cross-radiation gain coefficient of the target pixel is determined based on the geometric convergence index, the local energy concentration, and the reflection coupling coefficient of the casting surface.
[0058] In one embodiment, the local energy concentration of the target pixel is determined by: using the statistical maximum and minimum gray values of the second infrared thermal image, mapping the gray values of each pixel in the neighborhood to a normalized interval; calculating the square of the normalized gray values of each pixel, and using the sum of the squares of all pixels in the neighborhood as the local energy concentration.
[0059] We can first count the maximum grayscale value in the entire second infrared thermal image. and minimum gray value For the neighborhood Each pixel grayscale within It can be normalized to , making It is between 0 and 1.
[0060] calculate The size of the neighborhood can be set according to the actual wall thickness characteristics of the casting, for example, by selecting... or The window.
[0061] Radiant energy is proportional to the fourth power of temperature. In the grayscale domain, high grayscale values represent extremely high energy density. The square operation can non-linearly amplify the weight of bright pixels and suppress the influence of low grayscale pixels; local energy concentration. It reflects the total thermal energy density in a tiny area surrounding the target pixel.
[0062] if A higher value indicates that the region not only has a geometrically possible convergence point, but also a high probability of having energy reserves, making it a potential source of strong radiation interference.
[0063] In one embodiment, the cross-radiation gain coefficient is determined by the following formula: ,in, The mutual radiation gain coefficient; The reflection coupling coefficient of the material surface is denoted as . Geometric convergence index, The preset geometric weight adjustment index, For local energy concentration, The preset sensitivity threshold, It is a natural exponential function.
[0064] It characterizes the reflective properties of the casting material itself; for high-reflectivity materials such as aluminum alloys, The value should be relatively large, for example, 0.3 to 0.5; for sand-cast iron surfaces, The value is relatively small, for example, 0.1 to 0.2; The value used to adjust the nonlinearity of the geometric effect is usually between 1 and 2.
[0065] If the surface geometry of the casting is sharp and deeply recessed, it can increase... The value is used to reinforce the influence of the geometric index. It is a saturation function, taking values between 0 and 1; when the local energy concentration... When smaller, The value of this term increases approximately linearly.
[0066] when Exceeding the sensitivity threshold After that, when the term approaches saturation, for example, when it is close to 1, once the surrounding energy density reaches a certain level, the influence of mutual radiation will not increase indefinitely, but will be limited by the field of view and the surface absorptivity.
[0067] The value is dynamically generated for each pixel in a flat and uniformly temperatured region. The value approaches 0, leading to When the value approaches 0, no correction is performed; at the bottom of the high-temperature groove, Large value and Large values lead to The value increased significantly, and the measured value included a large number of interradiative components, achieving adaptive compensation for emissivity.
[0068] In step S104, the equivalent intrinsic gray level of the target pixel is determined based on the ratio of the target pixel to the average gray level of its neighborhood and the cross-radiation gain coefficient, so as to convert the equivalent intrinsic gray level to obtain the surface temperature information of the casting.
[0069] In one embodiment, the equivalent intrinsic gray level is determined by the following calculation formula: ,in, The equivalent intrinsic gray level of the target pixel; The grayscale value of the target pixel; The mutual radiation gain coefficient; The average gray value of the neighborhood of the target pixel. It is a preset positive number.
[0070] In the formula for calculating equivalent intrinsic gray level, The surrounding neighborhood of the target pixel, for example The average grayscale value of the window It is a very small positive number, such as 0.0001, used to prevent the denominator from being zero, ensuring the stability of numerical calculations, and also representing the algorithm's tolerance to small background noise.
[0071] The formula for calculating equivalent intrinsic gray level A correction factor based on local contrast was constructed if the gray level of the target pixel is much larger than the neighborhood average. This indicates that the target pixel itself is the center of the heat source. At this time, the ratio is relatively large, and the correction intensity will be adjusted accordingly. The observed total radiation signal is equal to the superposition of the intrinsic radiation and the environmental mutual radiation term. The mutual radiation gain can be removed by division.
[0072] The cross-radiation effect manifests as a mixture of multiplicative and additive noise in image grayscale. In regions dominated by high-temperature radiation, through... Normalization, used as the denominator, can bring potentially inflated grayscale values back to their true levels.
[0073] In one embodiment, converting the equivalent intrinsic grayscale to obtain the surface temperature information of the casting includes: using the difference between the equivalent intrinsic grayscale and the dark current bias value of the infrared thermal imager system as the net radiation response value; using the product of the photoelectric response parameters of the infrared thermal imager system, the emissivity of the casting surface, and the transmittance of the environment as the system response coefficient; and using the ratio of the net radiation response value to the system response coefficient to determine the temperature information of the target pixel to obtain the surface temperature information of all surface pixels in the casting.
[0074] Output grayscale value of infrared detector The relationship between the incident radiation flux and the response is usually linear, and the physical equation can be expressed as: Where P is the dark current bias value, i.e. the noise floor reading of the detector when there is no radiation input; These are photoelectric response parameters; It is the emissivity; It refers to atmospheric or lens transmittance; It is a temperature-radiation function determined according to Planck's formula or Stefan-Boltzmann's law.
[0075] First, calculate the net radiation response value. Construct system response coefficients The radiance of the target pixel can be obtained through The radiance is obtained by converting it into Celsius or Kelvin temperature based on the temperature-brightness lookup table or inverse function determined during the infrared thermal imager calibration process.
[0076] After correcting the equivalent intrinsic grayscale in the preceding steps, the calculated temperature field can truly reflect the cooling state of the casting surface, eliminating the illusion of local hot spots caused by concave corner reflection, which helps to determine whether the casting meets the unpacking conditions and predict the distribution of residual stress.
[0077] Figure 2 This is the original infrared thermal image of the casting before correction during the cooling process; the original infrared thermal image is the first infrared thermal image obtained. Figure 2 As shown, the deep cavity area at the top of the casting and the three circular groove structures on the side have higher temperature values. These areas form a geometrically semi-enclosed space, causing the infrared radiation emitted from each surface to be reflected and superimposed multiple times in the cavity. This mutual radiation effect makes the radiation energy received by the infrared thermal imager higher than the radiation energy corresponding to the actual temperature at that point, resulting in an artificially high measurement result that cannot reflect the true temperature state of the casting.
[0078] Figure 3 This is a schematic diagram of the surface temperature field after correction of the original infrared thermal image, such as... Figure 3As shown, after applying the casting cooling temperature field reconstruction method based on infrared thermograms according to the embodiments of this application, the brightness of the deep cavity at the top of the casting and the circular groove on the side decreases, and the temperature distribution of the flat outer surface of the casting changes little before and after correction. The method of this application has good geometric adaptability, can identify and correct those high-temperature depression areas that are severely affected by mutual radiation interference, while maintaining the authenticity of the original data for flat areas without mutual radiation interference, and restoring the temperature details of the casting surface.
[0079] Figure 4 This is a comparison chart of the original infrared thermal image and the key profile temperature data of the reconstructed temperature field, such as... Figure 4 As shown, this enables a more intuitive comparison between the original infrared thermal image and the reconstructed temperature field. Figure 4 The paper presents a comparison of the temperatures of pixels at different vertical positions in a row of pixel data spanning a key concave structure. The geometric convergence index in this application can effectively distinguish between concave and flat areas, and only performs gain correction when there is a risk of geometric interradiation, thus avoiding the errors caused by blindly homogenizing the entire image.
[0080] Figure 5 This is a schematic diagram illustrating the structure of a casting cooling temperature field reconstruction system 1000 based on infrared thermography, according to an exemplary embodiment. (Refer to...) Figure 5 The casting cooling temperature field reconstruction system 1000 based on infrared thermal imaging includes a processor 1100 and a memory 1200. The memory 1200 stores computer program instructions, which, when executed by the processor 1100, implement all or part of the steps of the casting cooling temperature field reconstruction method based on infrared thermal imaging in this application.
[0081] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only.
[0082] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope.
Claims
1. A method for reconstructing the cooling temperature field of castings based on infrared thermal imaging, characterized in that, include: A first infrared thermal image of the casting during the cooling process is obtained, and the cumulative probability quantile of the grayscale histogram of the first infrared thermal image is determined. The first infrared thermal image is then truncated based on the cumulative probability quantile to obtain a second infrared thermal image. Construct the gray-level gradient vector of the neighboring pixels of the target pixel in the second infrared thermal image, and determine the position vector from the neighboring pixels to the target pixel. Use the cosine of the angle between the gray-level gradient vector and the position vector, and the logarithmic weighting term of the brightness of the neighboring pixels relative to the average brightness of the second infrared thermal image, to determine the geometric convergence index of the target pixel. The local energy concentration of the target pixel is determined based on the gray-level extreme values in the neighborhood. The cross-radiation gain coefficient of the target pixel is determined based on the geometric convergence index, the local energy concentration, and the reflection coupling coefficient of the casting surface. Based on the ratio of the target pixel to the average gray level of its neighborhood and the cross-radiation gain coefficient, the equivalent intrinsic gray level of the target pixel is determined, and the equivalent intrinsic gray level is converted to obtain the surface temperature information of the casting. The geometric convergence index of the target pixel is determined by the following formula: ,in, Geometric convergence index; For the neighborhood Total number of pixels within; For neighboring pixels The gray-level gradient vector; To obtain from neighboring pixels The position vector pointing to the target pixel; Represents the magnitude of a vector; Represents the dot product of vectors; To correct the linear unit function; For neighboring pixels grayscale value; Let be the average grayscale value of the entire infrared thermal image, and ln be the natural logarithm function. It is a natural constant; The local energy concentration of the target pixel is determined as follows: using the statistical maximum and minimum gray values of the second infrared thermal image, the gray values of each pixel in the neighborhood are mapped to a normalized interval; the square of the normalized gray value of each pixel is calculated, and the sum of the square values of all pixels in the neighborhood is taken as the local energy concentration. The mutual radiation gain coefficient is determined by the following formula: ,in, The mutual radiation gain coefficient; The reflection coupling coefficient of the material surface is denoted as . Geometric convergence index, The preset geometric weight adjustment index, For local energy concentration, The preset sensitivity threshold, It is a natural exponential function; The equivalent intrinsic gray level is determined by the following formula: ,in, The equivalent intrinsic gray level of the target pixel; The grayscale value of the target pixel; The mutual radiation gain coefficient; The average gray value of the neighborhood of the target pixel. It is a preset positive number.
2. The method for reconstructing the cooling temperature field of castings based on infrared thermography according to claim 1, characterized in that, The second infrared thermal image is obtained by truncating the first infrared thermal image based on the cumulative probability quantiles, including: The cumulative probability quantile of the first proportion of the grayscale histogram is used as the lower threshold, and the cumulative probability quantile of the second proportion is used as the upper threshold; the first proportion is less than the second proportion; In the first infrared thermal image, pixels above the upper limit threshold are assigned the upper limit threshold value, pixels below the lower limit threshold are assigned the lower limit threshold value, and the pixel values of pixels in the first infrared thermal image that are between the upper and lower limit thresholds are retained to obtain the second infrared thermal image.
3. The method for reconstructing the cooling temperature field of castings based on infrared thermography according to claim 1, characterized in that, Before determining the geometric convergence index, the method further includes: Calculate the absolute difference between the target pixel and its four neighboring pixels, and construct an adaptive weight function based on the absolute difference. The adaptive weight function is negatively correlated with the square of the absolute difference. An adaptive weighting function is used to smooth the second infrared thermogram, and the filtered image is then used as the second infrared thermogram again.
4. The method for reconstructing the cooling temperature field of castings based on infrared thermography according to claim 1, characterized in that, The modified linear unit function is used to retain the component with a positive cosine value of the angle between the gray-level gradient vector and the position vector, and to set the component with a negative cosine value of the angle to zero.
5. The method for reconstructing the cooling temperature field of castings based on infrared thermography according to claim 1, characterized in that, The surface temperature information of the casting is obtained by converting the equivalent intrinsic grayscale, including: The difference between the equivalent intrinsic gray level and the dark current bias value of the infrared thermal imager system is taken as the net radiation response value, and the product of the photoelectric response parameters of the infrared thermal imager system, the emissivity of the casting surface, and the transmittance of the environment is taken as the system response coefficient. The temperature information of the target pixel is determined by the ratio of the net radiation response value to the system response coefficient, so as to obtain the surface temperature information of all surface pixels in the casting.
6. A casting cooling temperature field reconstruction system based on infrared thermal imaging, characterized in that, include: The processor and memory, wherein the memory stores computer program instructions that, when executed by the processor, implement the casting cooling temperature field reconstruction method based on infrared thermograms according to any one of claims 1-5.