Thermal radiation curved surface reconstruction and correction method based on curvature flow and control point screening

Abstract

The invention discloses a thermal radiation curved surface reconstruction and correction method based on curvature flow and control point screening. The method comprises the steps that a thermal radiation degradation image is acquired and preprocessed; calculating the local average curvature and gradient magnitude of each point in the graph, iteratively updating the height of each point of the curved surface through curvature flow operation, and adaptively terminating iteration in combination with the maximum number of iterations and a convergence criterion; morphological top-hat transformation is carried out on the image after curvature flow processing, and downsampling is carried out to obtain a thermal radiation curved surface control point; calculating the difference between a local plane normal vector of each control point and a neighborhood control point to obtain a Laplace curvature, screening out the control points smaller than a curvature threshold, and forming a final control point by the control points and the boundary points; reconstructing a heat radiation curved surface through Delaunay triangulation interpolation, and performing Lowess smoothing processing to obtain a continuous and smooth heat radiation curved surface; and subtracting the image from the thermal radiation degraded image to obtain a corrected image. According to the invention, the thermal radiation degraded image with a highlight target or an abnormal extreme value can be corrected.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of aerodynamic optical thermal radiation effect correction, and particularly relates to a thermal radiation curved surface reconstruction and correction method based on curvature flow and control point screening. BACKGROUND

[0002] In a high-temperature and strong-airflow environment, the high temperature of the window surface of an aircraft infrared imaging device will cause radiation distortion, making it difficult for an optical detector to accurately obtain target information. The method of using deep learning to establish a thermal radiation image degradation model and a correction parameter database to correct the thermal radiation degradation image has great limitations. The physical means such as imaging window coating and cooling adopted not only increase the design difficulty and cost, but also cannot significantly improve the imaging clarity.

[0003] Meanwhile, due to the complexity and diversity of the imaging conditions, imaging equipment and imaging targets of high-speed aircraft, the applicability of various correction algorithms is relatively narrow, and they are generally used for 8-bit low dynamic range degradation images, but when applied to 12-bit, 16-bit high dynamic range degradation images with saturated bright targets or abnormal values, overfitting or overcompensation may occur, so that high-quality clear images cannot be restored.

[0004] The existing traditional correction algorithm uses control points for curved surface fitting. Once the control points are not selected properly, the curved surface calculation result will be affected by the control points located at the target or noise position. Meanwhile, no matter how high the order of the curved surface is, it will still be strictly constrained by the mathematical model, and problems such as excessive calculation and long time consumption may also occur. SUMMARY

[0005] The purpose of the present application is to provide a thermal radiation curved surface reconstruction and correction method based on curvature flow and control point screening, which can be used for the correction of high dynamic range aerodynamic optical thermal radiation effect degradation images with saturated bright targets.

[0006] The technical solutions adopted by the present application are as follows: The first aspect of the present application provides a thermal radiation curved surface reconstruction and correction method based on curvature flow and control point screening, which comprises the following steps: S1, obtaining a thermal radiation degradation image and pre-processing the thermal radiation degradation image; S2, based on the pre-processed image, calculating the local average curvature and gradient amplitude of each point in the image, iteratively updating the height of each point of the curved surface through the curvature flow operation, and simultaneously combining the maximum iteration number and the convergence criterion to adaptively terminate the iteration to obtain the image after curvature flow processing; S3, performing morphological top-hat transformation on the image processed by the curvature flow to suppress local abnormal extreme values in the image, extracting a thermal radiation feature image, and performing down-sampling on the thermal radiation feature image to obtain thermal radiation curved surface control points; S4, determining local plane normal vectors of all control points of the thermal radiation curved surface; S5, calculating differences between the local plane normal vector of each control point and local plane normal vectors of its neighborhood control points to obtain Laplace curvatures of the control points, screening out control points with a Laplace curvature less than a curvature threshold value to form final control points of the thermal radiation curved surface together with the boundary points; S6, reconstructing the thermal radiation curved surface by Delaunay triangulation interpolation according to the final control points of the thermal radiation curved surface to obtain a preliminary thermal radiation curved surface, and performing Lowess smoothing processing on the preliminary thermal radiation curved surface to obtain a continuous and smooth thermal radiation curved surface; S7, subtracting the continuous and smooth thermal radiation curved surface from the thermal radiation degeneration image to realize correction of the thermal radiation degeneration image, and obtaining a corrected image.

[0007] In the above scheme, the preprocessing includes gray scale conversion, normalization and boundary filling.

[0008] In the above scheme, step S2 specifically includes: based on the preprocessed image Z, calculating local average curvatures and gradient amplitudes H of each point in the image:

[0009]

[0010] In the formula, H and are the local average curvatures and gradient amplitudes of each point in the image Z; , , , , are the first-order partial derivatives and second-order partial derivatives of each point in the image Z; taking each point as a unit, combining the local average curvature H and the gradient amplitude of the point, and iteratively updating the curved surface height:

[0011] In the formula, is the curved surface after one iteration update; is the initial value of each point in the pre-processed image Z; is the time step; The maximum amplitude of the change of the surface between two adjacent iterations is calculated, and if it is less than a set threshold value, or the current iteration number reaches a preset maximum iteration number, the iteration is terminated to obtain the curvature flow processed image; otherwise, the local average curvature and gradient amplitude of each point in the image are recalculated based on the updated surface of the iteration, and the iteration is continued.

[0012] In the above scheme, after each iteration update, it is judged whether the value of each point is between 0 and 1, if yes, the original value is kept, if greater than 1, it is set to 1, and if less than 0, it is set to 0.

[0013] In the above scheme, step S5 specifically includes: The Euclidean difference average between the local plane normal vector of each control point and the local plane normal vector of its neighborhood control points is calculated as the local curvature index of the control point:

[0014] In the formula, represents the Laplace curvature of the control point; represents the neighborhood control point set of the control point, represents the number of control points in the neighborhood control point set; represents the local plane normal vector of the control point, represents the local plane normal vector of the control point in the neighborhood control point set; According to the set curvature threshold , the control points with Laplace curvature less than the curvature threshold are screened out to form the final control points of the thermal radiation surface together with the boundary points:

[0015] In the formula, is the final control point set of the thermal radiation surface, including all control points with Laplace curvature less than the curvature threshold and the boundary point set.

[0016] In the above scheme, according to the final control points of the thermal radiation surface, the thermal radiation surface is interpolated and reconstructed by Delaunay triangulation to obtain a preliminary thermal radiation surface, including: According to the final control points of the thermal radiation surface, a Delaunay triangular mesh is constructed with the final control points as vertices by using the Delaunay triangulation method, and the entire surface is divided into several non-overlapping triangles; ​​For all points in the thermal radiation degradation image, if a corresponding final control point exists, its value is taken; otherwise, it is used as an interpolation point. For each interpolation point, the triangle it belongs to is determined, and the gray value of the interpolation point is calculated by weighting the area using the coordinates and gray values ​​of the three vertices of the triangle. :

[0017] In the formula, Indicates the interpolation point The value; This represents the values ​​of the three vertices of the triangle to which the interpolation point belongs; For the interpolation weights corresponding to each vertex, Let the area be the area of ​​the sub-triangle formed by the interpolation point and the remaining two vertices of the triangle; After traversing all interpolation points, a continuous surface is formed, namely the preliminary thermal radiation surface.

[0018] In the above scheme, Lowess smoothing is performed as follows:

[0019]

[0020] In the formula, Midpoint of the preliminary thermal radiation surface Values ​​after Lowess smoothing; Point local neighborhood All points within , For local neighborhood radius; Representing neighborhood points The value; Representing neighborhood points The weight.

[0021] In the above scheme, step S7 further includes: The brightness of points in the corrected image that exceed the brightness threshold is set as the brightness threshold, and low-brightness areas in the image are enhanced by linear gray-scale stretching.

[0022] According to a second aspect of the present invention, a computer device is provided, comprising: a processor and a memory, the memory storing a program or instructions executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the thermal radiation surface reconstruction and correction method based on curvature flow and control point screening as described in any one of the first aspects.

[0023] According to a third aspect of the present invention, a computer-readable storage medium is provided having a program or instructions stored thereon, which, when executed by a processor, implement the steps of the thermal radiation surface reconstruction and correction method based on curvature flow and control point screening as described in any one of the first aspects.

[0024] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects: This invention provides a method for reconstructing and correcting thermal radiation surfaces based on curvature flow and control point selection. When processing high dynamic range degraded images, it avoids the influence of saturated bright targets and outliers on the thermal radiation surface calculation process. Furthermore, before surface reconstruction, the selected control points are screened based on their curvature characteristics, avoiding the use of unreasonable control points that could affect the correction results. Simultaneously, by using irregular control points and employing a weighted interpolation method for surface reconstruction, the interpolation points are only affected by neighboring control points, avoiding constraints from the mathematical model. This method can correct different types of aero-optical thermal radiation effects, especially for high dynamic range thermal radiation degraded images with abnormally bright targets or significant noise and targets, achieving better results, improving the image signal-to-noise ratio, and preparing for target recognition and detection operations. Attached Figure Description

[0025] Figure 1 A flowchart illustrating a method for reconstructing and correcting thermal radiation surfaces based on curvature flow and control point screening, provided for an embodiment of the present invention; Figure 2 This invention provides a 16-bit high dynamic range thermal radiation degradation map of a saturated high-brightness target. Figure 3 This invention provides an image after curvature flow processing. Figure 4 This invention provides a thermal radiation feature map after suppressing a bright target through morphological top-hat transformation. Figure 5 This invention provides a control point map obtained by downsampling a thermal radiation feature image; Figure 6 A filtered control point map provided in an embodiment of the present invention; Figure 7 This invention provides a reconstructed thermal radiation surface map as an embodiment of the invention. Figure 8 This is a diagram corresponding to the reconstructed thermal radiation surface provided in an embodiment of the present invention; Figure 9 This is an image provided by an embodiment of the present invention after surface correction to remove thermal radiation; Figure 10 This invention provides a three-dimensional surface diagram corresponding to a surface correction diagram after removing thermal radiation. Figure 11 This is a calibration diagram for enhancing display effect provided in an embodiment of the present invention; Figure 12 This invention provides a three-dimensional surface diagram corresponding to a calibration diagram with enhanced display effect, as provided in an embodiment of the invention. Figure 13 This is a schematic diagram of the hardware structure of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. All other embodiments obtained by those skilled in the art based on the embodiments provided by this invention without inventive effort are within the scope of protection of this invention.

[0027] Obviously, the accompanying drawings described below are merely some examples or embodiments of the present invention. Those skilled in the art can apply the present invention to other similar scenarios based on these drawings without any inventive effort. Furthermore, it is understood that although the efforts made in this development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this invention, modifications to design, manufacturing, or production based on the technical content disclosed in this invention are merely conventional technical means and should not be construed as insufficient disclosure of the present invention.

[0028] In this invention, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this invention may be combined with other embodiments without conflict.

[0029] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "a," "an," "an," "the," and similar words used in this invention do not indicate quantity limitation and may indicate singular or plural. The terms "comprising," "including," "having," and any variations thereof used in this invention are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units not listed, or may include other steps or units inherent to these processes, methods, products, or devices. The terms "connected," "linked," "coupled," and similar words used in this invention are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "A plurality" used in this invention refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships may exist; for example, "A and / or B" can represent: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following objects have an "or" relationship. The terms "first," "second," and "third" used in this invention are merely to distinguish similar objects and do not represent a specific ordering of the objects.

[0030] This invention provides a method for reconstructing and correcting thermal radiation surfaces based on curvature flow and control point selection, comprising the following steps: First, input a thermal radiation degradation image and processing parameters, and preprocess the thermal radiation degradation image, including grayscale conversion, normalization, and boundary filling; then, iteratively update the image surface height by using the local average curvature and gradient magnitude of the preprocessed image pixels to filter out and smooth abrupt details on the surface, while suppressing local abnormal extrema in the image through morphological top-hat transformation, thereby extracting thermal radiation feature images; obtain control points of the thermal radiation surface through downsampling, and extract local surface normal vectors based on the control point normal vector estimation method; then calculate the difference between the control point and its neighboring normal vectors to obtain the Laplacian curvature of the control point, and select low curvature points and boundary points; based on this, construct a continuous and smooth thermal radiation surface using Delaunay triangulation interpolation surface reconstruction and Lowess smoothing algorithm; finally, remove this surface from the input image to correct the thermal radiation effect of the degradation image, and simultaneously perform highlight target suppression and shadow enhancement on the corrected image to achieve better display effects.

[0031] The method of this invention is simple and effective, and can be used to correct images degraded by high dynamic range aero-optical thermal radiation effects, especially those with saturated bright targets. By implementing the method of this invention, the defects of abnormal extrema affecting the correction quality in traditional correction algorithms can be avoided, effectively removing the thermal radiation effect of the image and improving image quality and clarity.

[0032] like Figure 1 As shown, the thermal radiation surface reconstruction and correction method based on curvature flow and control point screening provided by the present invention includes the following steps: S1. Preprocess the image, including grayscale conversion, normalization, and boundary filling.

[0033] S2. Calculate the local average curvature and gradient magnitude of each point in the image, iteratively update the height of each point on the surface through curvature flow operation, and adaptively terminate the iteration by combining the maximum number of iterations and the convergence criterion, so as to filter out and smooth the details of abrupt changes on the image surface.

[0034] Specifically, the local average curvature and gradient magnitude of each point in the three-dimensional surface of the preprocessed image are calculated. At each surface point, its height value is updated based on its current local curvature features and time step. The entire surface is iteratively updated while numerical truncation is applied to prevent numerical overflow. The iteration is adaptively terminated by combining the maximum number of iterations and the convergence criterion. This achieves the filtering and smoothing of abrupt details on the image surface and initially extracts thermal radiation feature images.

[0035] S3. Next, the image obtained in step S2 is subjected to morphological top-hat transformation to suppress local abnormal extrema in the image, thereby extracting the thermal radiation feature image and downsampling to obtain thermal radiation surface control points.

[0036] Specifically, for the initially extracted thermal radiation feature image, a morphological structuring element is first constructed as a local operation template for the image. Using this template, the TopHat algorithm (i.e., subtracting the morphological opening operation result from the original image) is used to extract local abnormal extreme value regions in the image whose size is smaller than the structuring element. Then, this region is removed from the original image. After smoothing the edges and details of the scene target, only the thermal radiation feature image of the background brightness change trend caused by thermal effect is retained. Downsampling is then performed to select pixel points on a uniform grid as control points for the thermal radiation surface.

[0037] S4. Extract the neighborhood local point set of the control point of the thermal radiation surface using the radius search method based on Euclidean distance, fit the local plane based on principal component analysis, and extract the local plane normal vector.

[0038] Specifically, the Euclidean distances between each control point on the thermal radiation surface and other points are calculated, and points within a set radius are selected as their neighborhood point set. A normal vector estimation method based on local neighborhood principal component analysis is employed. Through eigenvalue decomposition of the covariance matrix, the eigenvector corresponding to the minimum eigenvalue is extracted as the local tangent plane normal vector, which is used to calculate the subsequent normal vector difference.

[0039] S5. Calculate the difference between the normal vector of the control point and its neighboring control points to obtain the Laplace curvature. Based on the set curvature threshold, select the low curvature points and boundary points as the final control points.

[0040] Specifically, the difference between the normal vector of each control point and the normal vectors in its neighborhood is calculated, i.e., the Laplacian curvature, which serves as the local curvature index for that point. This index is used to filter out high-curvature control points located on scene targets and details, while forcibly retaining control points at image edges, thereby constructing a stable and complete set of background control points for subsequent thermal radiation surface reconstruction.

[0041] S6. Based on the selected control points, the thermal radiation surface is reconstructed by Delaunay triangulation interpolation. At the same time, the Lowess smoothing algorithm is used to eliminate local noise and jumps, resulting in a continuous and smooth thermal radiation surface.

[0042] Specifically, using the selected control points as vertices, a Delaunay triangular mesh is constructed, dividing the entire plane into several non-overlapping triangles. For each regular mesh point, its Delaunay triangle is found, and the gray value of the mesh point is calculated by weighted averaging based on the area using the coordinates and gray values ​​of the triangle's vertices, thus forming a continuous surface. Finally, the Lowess smoothing algorithm is used to eliminate local noise and jumps in the continuous surface, resulting in a continuous and smooth thermal radiation surface.

[0043] S7. The reconstructed thermal radiation surface is removed from the degraded image to correct the degraded image. At the same time, the corrected image is subjected to suppression of bright targets and enhancement of dark areas to achieve better display effect.

[0044] Specifically, the reconstructed thermal radiation surface is removed from the degraded image to obtain an image corrected for thermal radiation effects. Then, a high brightness threshold is set to truncate and suppress abnormally bright areas. Finally, the image is linearly stretched to enhance the visibility of low-brightness details, thereby improving the dynamic range balance of the image and enhancing visual quality.

[0045] By implementing the method of this invention, when processing high dynamic range degraded images, the influence of saturated bright targets and outliers on the thermal radiation surface calculation process can be avoided. Furthermore, before surface reconstruction, the selected control points are screened based on their curvature characteristics, avoiding the use of unreasonable control points that could affect the correction results. Simultaneously, by using irregular control points and employing a weighted interpolation method for surface reconstruction, the interpolation points are only affected by neighboring control points, avoiding constraints from the mathematical model. This method can correct different types of aero-optical thermal radiation effects, and it is particularly effective for high dynamic range thermal radiation degraded images with abnormally bright targets or significant noise and targets, improving the image signal-to-noise ratio and preparing for target recognition and detection operations.

[0046] This invention also provides a method for reconstructing and correcting thermal radiation surfaces based on curvature flow and control point selection. The algorithm flow is as follows: Figure 1 As shown, it includes the following steps: S1. Preprocess the image, including grayscale conversion, normalization, and boundary filling; S2. Calculate the local average curvature and gradient magnitude of each point in the image, iteratively update the height of each point on the surface through curvature flow operation, and adaptively terminate the iteration by combining the maximum number of iterations and the convergence criterion, so as to filter out and smooth the details of abrupt changes on the image surface. S3. Next, the image obtained in step S2 is subjected to morphological top-hat transformation to suppress local abnormal extrema in the image, thereby extracting the thermal radiation feature image and downsampling to obtain thermal radiation surface control points. S4. Extract the neighborhood local point set of the control point of the thermal radiation surface by the radius search method based on Euclidean distance, fit the local plane based on principal component analysis, and extract the local plane normal vector. S5. Calculate the difference between the normal vector of the control point and its neighboring control points to obtain the Laplace curvature. Based on the set curvature threshold, select the low curvature points and boundary points as the final control points. S6. Based on the selected control points, the thermal radiation surface is reconstructed by Delaunay triangulation interpolation. At the same time, the Lowess smoothing algorithm is used to eliminate local noise and jumps, resulting in a continuous and smooth thermal radiation surface. S7. The reconstructed thermal radiation surface is removed from the degraded image to correct the degraded image. At the same time, the corrected image is subjected to suppression of bright targets and enhancement of dark areas to achieve better display effect.

[0047] In some embodiments, the method in step S1 specifically involves: for example... Figure 2The 16-bit degraded image of size 256×256 shown is preprocessed, including grayscale conversion, normalization and boundary filling. The image data with 16-bit grayscale values ​​of integers 0 to 65536 needs to be normalized to a 0 to 1 double-precision floating-point number type to facilitate subsequent processing and calculation.

[0048] In some embodiments, the method in step S2 specifically involves: performing curvature flow iterative processing on the preprocessed image Z to suppress high-frequency noise and local anomalies, thereby filtering out and smoothing abrupt details on the image surface while preserving the smooth change trend in the background caused by thermal radiation.

[0049] First, calculate the points of the Z-shaped 3D surface of the image. Local mean curvature H and gradient magnitude :

[0050]

[0051] in, H and It is a matrix containing the average curvature and gradient magnitude of each point in image Z; , , , , Let Z be the first and second partial derivatives of each point in image Z. Since the calculation of derivatives is usually based on the spatial difference of the image's neighborhood, that is, using a 3×3 neighborhood formed by the pixel's top, bottom, left, right, and diagonal positions for approximation, this means that the local average curvature of a point is... H In fact, it is approximately calculated by the second derivative composed of the grayscale values ​​of the pixels in its local neighborhood, reflecting the curvature characteristics of the local area around the point.

[0052] Then, taking each point as a unit, combine the gradient magnitude. Mean curvature and the set time step (In this embodiment, the value is set to 0.4), iteratively update the surface height:

[0053] in, For the new round of surface iteration, For the current surface (initial values ​​are the normalized grayscale values ​​of each point in the preprocessed image Z), this process represents the slight change in the height value (i.e., the normalized grayscale value) of each point in the image in each iteration, based on its local curvature shape. When this occurs, it indicates a localized depression, and the surface height will increase; when When it is, it indicates a local protrusion, and the surface height will decrease; when When the surface height is not updated, it indicates that the surface is locally flat.

[0054] In addition, numerical truncation needs to be applied during iteration to ensure that pixel values ​​remain within the range. Internal measures to prevent numerical overflow:

[0055] That is, by calculating the maximum magnitude of the surface change between two adjacent iterations. When the change is less than the set threshold When the time reaches 1e-5 (in this embodiment, the maximum number of iterations is set to 20), the algorithm is considered to have converged and terminated early, ultimately obtaining the image after curvature flow processing. like Figure 3 As shown.

[0056] It should be noted that the maximum magnitude of surface change between two adjacent iterations is less than a set threshold. At that time, the change in all points was less than the set threshold. .

[0057] In some embodiments, the method in step S3 specifically involves: setting a morphological structuring element (in this embodiment, a circular template with a radius of 15) to define the scale of local image operations, and then applying the local operation scale to the image. The TopHat morphological transformation is performed by subtracting the morphological opening operation result from the original image, thereby finding the image... The local extrema regions smaller than the structuring element are identified. The opening operation involves first eroding the image, then dilating it. This removes bright areas larger than the structuring element, retaining only local extrema smaller than the structuring element.

[0058] Then from the image Subtracting this local extremum from the image removes bright targets or anomalous extrema regions that curvature flow operations cannot filter out, resulting in a thermal radiation feature image that retains only the brightness variation trend caused by thermal radiation in the background, free from target interference. Figure 4 As shown, downsampling is performed on a regular uniform grid, that is, sampling at a fixed step size n (set to 0.3 in this embodiment) at a fixed ratio every approximately 10*n pixels at equal intervals in the row and column directions to construct a regular uniform network, i.e., a set of control points for the thermal radiation surface, as shown. Figure 5 As shown.

[0059] In some embodiments, the method in step S4 specifically involves: for each control point of the thermal radiation surface... Choose a radius with its center as r (In this embodiment, it is set to 2) the set of points within the neighborhood. Principal component analysis was performed on the neighborhood point set. Using the geometric center of the neighborhood point set (i.e., the average position of these points) as a reference, the offset of each neighborhood point relative to the center was calculated, and a covariance matrix was constructed. :

[0060]

[0061]

[0062] in, Control points and neighboring points The Euclidean distance between them; , represents the geometric center of the neighborhood points, that is, the average position of the neighborhood points; It is the covariance matrix of the local point set, which reflects the spatial distribution characteristics of the local point set in three directions (x, y, z axes).

[0063] Subsequently, the covariance matrix was analyzed. Perform eigenvalue decomposition and take the eigenvector corresponding to the smallest eigenvalue as the control point. Normal vector of the local plane By iterating through each point, we can obtain the set of normal vectors for all points. .

[0064] In some embodiments, the method in step S5 specifically involves: calculating the mean Euclidean difference between the normal vector of each control point and its neighboring normal vectors. That is, the Laplace curvature, which serves as a local curvature index at that point:

[0065] According to the preset threshold (In this embodiment, the value is set to 0.7). Points with smaller curvature are selected, while the boundary point set is retained to supplement the edge constraints of the control points, thus obtaining the final control point set of the selected thermal radiation surface. ,like Figure 6 As shown:

[0066] Final control point set of thermal radiation surface It includes all Laplace curvatures Less than the curvature threshold control points It also includes the set of boundary points.

[0067] In some embodiments, the method in step S6 specifically involves: processing the final control point set of the selected thermal radiation surface. The Delaunay triangulation method is used to construct a Delaunay triangular mesh with control points as vertices, dividing the entire plane into several non-overlapping triangles.

[0068] For each grid point in the regular grid (a 256×256 grid of degraded image size) In other words, interpolation points (i.e., all pixels in the original degraded image that do not correspond to the final control points) are located in the Delaunay triangle based on their coordinates. Then, using the coordinates and grayscale values ​​of the three vertices of the triangle containing that point, a weighted average is calculated based on the area to determine the grayscale value of that grid point. :

[0069] in, This represents the grayscale values ​​of the three vertices of the triangle containing that point. For the interpolation weights corresponding to each vertex, Let be the area of ​​the sub-triangle formed by the point and two vertices of the triangle.

[0070] After traversing all interpolation points, a corresponding grayscale matrix is ​​generated for the entire regular mesh, forming a continuous surface. This is the initial thermal radiation surface. Furthermore, through interpolation, the surface size is restored to the original image size.

[0071] Finally, the surface is smoothed using Lowess to eliminate local noise and abrupt changes, resulting in a continuous and smooth thermal radiation surface. bias ,like Figure 7 and Figure 8 As shown. Lowess smoothing is performed as follows:

[0072] This expression represents the expression for each point in its local neighborhood. The inner (set to 5 in this example) uses a Gaussian kernel weight. Calculate a weighted average and assign different weights to each point in the neighborhood based on its distance from the center point, thereby smoothing out local noise and jumps.

[0073] In some embodiments, the method in step S6 specifically involves: removing the calculated thermal radiation surface from the input degraded image, thereby correcting the thermal radiation effect and obtaining a corrected image, such as... Figure 9 and Figure 10As shown. Since the high pixel values ​​of saturated, bright targets in the image suppress the display range of low-brightness areas, the brightness range is adjusted to improve the visualization quality of the corrected image. First, a brightness threshold is set (0.7 in this embodiment), and the portion exceeding this threshold is truncated to prevent local overexposure from affecting the overall display effect. Then, linear grayscale stretching is used to enhance the low-brightness areas in the image, stretching the grayscale range from [0, 0.5] to [0, 1] to enhance the display effect of dark details. The final corrected image achieves thermal radiation effect correction and foreground target brightness optimization, as shown. Figure 11 and Figure 12 As shown.

[0074] It should be noted that the steps shown in the above process or in the flowchart of the accompanying figures can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0075] Combination Figure 1 The thermal radiation surface reconstruction and correction method based on curvature flow and control point screening described in this invention can be implemented by a computer device. Figure 13 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Figure 13 As shown, the device may include a processor 301 and a memory 302 storing computer program instructions.

[0076] Specifically, the processor 301 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of the present invention.

[0077] Memory 302 may include a large-capacity memory for data or instructions. For example, and not limitingly, memory 302 may include a hard disk drive (HDD), a floppy disk drive, a solid-state drive (SSD), flash memory, an optical disk drive, a magneto-optical disk drive, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 302 may include removable or non-removable (or fixed) media. Where appropriate, memory 302 may be internal or external to a data processing device. In a particular embodiment, memory 302 is non-volatile memory. In a particular embodiment, memory 302 includes read-only memory (ROM) and random access memory (RAM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable read-only memory (PROM), an erasable read-only memory (EPROM), an electrically erasable read-only memory (EEPROM), an electrically alterable read-only memory (EAROM), or flash memory, or a combination of two or more of these. Where appropriate, the RAM can be Static Random-Access Memory (SRAM) or Dynamic Random-Access Memory (DRAM). DRAM can be Fast Page Mode Dynamic Random-Access Memory (FPMDRAM), Extended Data Out Dynamic Random-Access Memory (EDODRAM), Synchronous Dynamic Random-Access Memory (SDRAM), etc.

[0078] The memory 302 can be used to store or cache various data files that need to be processed and / or communicated, as well as possible computer program instructions executed by the processor 301.

[0079] The processor 301 reads and executes computer program instructions stored in the memory 302 to implement any of the thermal radiation surface reconstruction and correction methods based on curvature flow and control point screening in the above embodiments.

[0080] In some embodiments, the computer device may further include a communication interface 303 and a bus 300. For example, Figure 13 As shown, the processor 301, memory 302, and communication interface 303 are connected through bus 300 and complete communication with each other.

[0081] The communication interface 303 is used to enable communication between the various modules, devices, units, and / or equipment in the embodiments of the present invention. The communication interface 303 can also enable data communication with other components such as external devices, image / data acquisition devices, databases, external storage, and image / data processing workstations.

[0082] Bus 300 includes hardware, software, or both, that couples components of a computer device together. Bus 300 includes, but is not limited to, at least one of the following: data bus, address bus, control bus, expansion bus, and local bus. For example, and not as a limitation, bus 300 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a MicroChannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses, or a combination of two or more of these. Where appropriate, bus 300 may include one or more buses. Although specific buses are described and illustrated in embodiments of the invention, the invention contemplates any suitable bus or interconnect.

[0083] The computer device can execute the thermal radiation surface reconstruction and correction method based on curvature flow and control point screening in the embodiments of the present invention, thereby achieving a combination of Figure 1 The method described is a thermal radiation surface reconstruction and correction method based on curvature flow and control point screening.

[0084] Furthermore, in conjunction with the thermal radiation surface reconstruction and correction method based on curvature flow and control point screening in the above embodiments, this invention can be implemented using a computer-readable storage medium. This computer-readable storage medium stores computer program instructions; when executed by a processor, these computer program instructions implement any of the thermal radiation surface reconstruction and correction methods based on curvature flow and control point screening in the above embodiments.

[0085] In summary, this invention provides a thermal radiation surface reconstruction and correction method based on curvature flow and control point selection. It can be used to correct images with high dynamic range aero-optical-thermal effects that degrade the image due to saturated bright targets. It avoids the defects of abnormal extreme values ​​affecting the correction quality in traditional correction algorithms, and also avoids the situation where improper selection of control points affects the correction effect in traditional correction methods. It effectively removes the thermal radiation effect of the image and improves the image quality and clarity.

[0086] It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments have been described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification. In addition, depending on the implementation needs, the various steps / components described in this invention can be broken down into more steps / components, or two or more steps / components or parts of steps / components can be combined into new steps / components to achieve the purpose of this invention.

[0087] It will be readily understood by those skilled in the art that the above-described embodiments merely illustrate several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this patent should be determined by the appended claims.

Claims

1. A method for reconstructing and correcting thermal radiation surfaces based on curvature flow and control point selection, characterized in that, The method includes: S1. Acquire thermal radiation degradation images and preprocess them; S2. Based on the preprocessed image, calculate the local average curvature and gradient magnitude of each point in the image. Iteratively update the height of each point on the surface through curvature flow operation. At the same time, combine the maximum number of iterations and the convergence criterion to adaptively terminate the iteration and obtain the image after curvature flow processing. S3. Perform morphological top-hat transformation on the image after curvature flow processing to suppress local abnormal extrema in the image, extract the thermal radiation feature image, and downsample the thermal radiation feature image to obtain thermal radiation surface control points. S4. Determine the local plane normal vectors of all control points on the thermal radiation surface; S5. Calculate the difference between the local plane normal vector of each control point and the local plane normal vector of its neighboring control points to obtain the Laplacian curvature of the control point. Based on the set curvature threshold, select control points whose Laplacian curvature is less than the curvature threshold and form the final control points of the thermal radiation surface together with the boundary points. S6. Based on the final control points of the thermal radiation surface, the thermal radiation surface is reconstructed by Delaunay triangulation interpolation to obtain a preliminary thermal radiation surface. The preliminary thermal radiation surface is then smoothed by Lowess to obtain a continuous and smooth thermal radiation surface. S7. Subtract the continuous and smooth thermal radiation surface from the thermal radiation degradation image to correct the thermal radiation degradation image and obtain the corrected image.

2. The method for reconstructing and correcting thermal radiation surfaces based on curvature flow and control point selection according to claim 1, characterized in that, Preprocessing includes grayscale conversion, normalization, and boundary filling.

3. The method for reconstructing and correcting thermal radiation surfaces based on curvature flow and control point selection according to claim 1, characterized in that, Step S2 specifically includes: Based on the preprocessed image Z, calculate the points in the image. Local mean curvature H and gradient magnitude : In the formula, H and Each point in image Z The local mean curvature and gradient magnitude; , , , , Each point in image Z The first and second partial derivatives; Taking each point as a unit, combined with the local mean curvature of that point H and gradient magnitude Iteratively update the surface height: In the formula, The surface after one iteration update; This is the current surface, and its initial value is the value of each point in the preprocessed image Z; For time step; Calculate the maximum magnitude of surface change between two adjacent iterations. If it is less than a set threshold, or the current iteration count reaches the preset maximum iteration count, terminate the iteration and obtain the image after curvature flow processing; otherwise, recalculate the local average curvature and gradient magnitude of each point in the image based on the iteratively updated surface, and continue iterating.

4. The method for reconstructing and correcting thermal radiation surfaces based on curvature flow and control point selection according to claim 3, characterized in that, After each iteration update, check if the value of each point is between 0 and 1. If it is, keep the original value; if it is greater than 1, set it to 1; if it is less than 0, set it to 0.

5. The method for reconstructing and correcting thermal radiation surfaces based on curvature flow and control point selection according to claim 1, characterized in that, Step S5 specifically includes: Calculate the mean Euclidean difference between the local plane normal vector of each control point and the local plane normal vectors of its neighboring control points. , serving as a local curvature index for this control point: In the formula, This represents the Laplace curvature of the control point; This represents the set of neighboring control points of the control point. Indicates the number of centralized control points in the neighborhood control points; This represents the local plane normal vector of the control point. This represents the local plane normal vector of the centralized control points in the neighborhood control points; Based on the set curvature threshold Filter out those with Laplacian curvature less than the curvature threshold The control points, together with the boundary points, form the final control points of the thermal radiation surface: In the formula, The final control point set for the thermal radiation surface, including all Laplace curvatures. Less than the curvature threshold control points And the set of boundary points.

6. The method for reconstructing and correcting thermal radiation surfaces based on curvature flow and control point selection according to claim 1, characterized in that, Based on the final control points of the thermal radiation surface, the thermal radiation surface is reconstructed through Delaunay triangulation interpolation to obtain a preliminary thermal radiation surface, including: Based on the final control points of the thermal radiation surface, the Delaunay triangulation method is used to construct a Delaunay triangular mesh with the final control points as vertices, dividing the entire surface into several non-overlapping triangles. For all points in the thermal radiation degradation image, if a corresponding final control point exists, its value is taken; otherwise, it is used as an interpolation point. For each interpolation point, the triangle it belongs to is determined, and the gray value of the interpolation point is calculated by weighting the area using the coordinates and gray values ​​of the three vertices of the triangle. : In the formula, Indicates the interpolation point The value; This represents the values ​​of the three vertices of the triangle to which the interpolation point belongs; For the interpolation weights corresponding to each vertex, Let the area be the area of ​​the sub-triangle formed by the interpolation point and the remaining two vertices of the triangle; After traversing all interpolation points, a continuous surface is formed, namely the preliminary thermal radiation surface.

7. The method for reconstructing and correcting thermal radiation surfaces based on curvature flow and control point selection according to claim 1, characterized in that, Lowess smoothing is performed as follows: In the formula, Midpoint of the preliminary thermal radiation surface Values ​​after Lowess smoothing; Point local neighborhood All points within , For local neighborhood radius; Representing neighborhood points The value; Representing neighborhood points The weight.

8. The method for reconstructing and correcting thermal radiation surfaces based on curvature flow and control point screening according to claim 1, characterized in that, Step S7 also includes: The brightness of points in the corrected image that exceed the brightness threshold is set as the brightness threshold, and low-brightness areas in the image are enhanced by linear gray-scale stretching.

9. A computer device, characterized in that, include: The processor and memory, wherein the memory stores a program or instructions that can run on the processor, and when the program or instructions are executed by the processor, implement the steps of the thermal radiation surface reconstruction and correction method based on curvature flow and control point screening as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, It stores a program or instruction, which, when executed by a processor, implements the steps of the thermal radiation surface reconstruction and correction method based on curvature flow and control point screening as described in any one of claims 1 to 8.

Images