Panoramic CBCT brightness equalization method based on Gaussian low-frequency modeling and Poisson reconstruction combined bias estimation

Abstract

The invention discloses a panoramic CBCT (cone beam computed tomography) brightness equalization method based on Gaussian low-frequency modeling and Poisson reconstruction combined bias estimation, which relates to the technical field of medical image equalization and comprises the following steps: converting a Raw format of an image data file to be processed into a 32-bit floating point data format according to an ini configuration file; creating a Gaussian kernel according to the mathematical expression; a Laplace residual error is constructed; and superposing the images corrected by the bias field according to the proportion. According to the method, the brightness low-frequency trend and the structure gradient change are captured at the same time through a double-branch bias field learning mechanism, rapid and accurate estimation of the bias field is achieved, and the high-robustness bias correction model is obtained through adaptive fusion. And a gray scale stabilization strategy is further introduced, the brightness consistency and contrast optimization of the CBCT panoramic image are realized under the condition that the window width and the window level do not need to be manually adjusted, and the organization detail recognizable degree and the cross-device imaging consistency are effectively improved, so that the clinical diagnosis quality and stability are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of medical image equalization technology, and in particular to a panoramic CBCT brightness equalization method based on Gaussian low-frequency modeling and Poisson reconstruction combined with bias estimation. Background Technology

[0002] In CBCT panoramic imaging, the consistency of brightness and contrast directly affects the accuracy of lesion identification and clinical diagnosis. However, due to imaging factors such as differences in X-ray attenuation, scattering effects, limited projection geometry, and detector response inhomogeneity, CBCT panoramic images generally exhibit global or regional brightness bias. This bias typically manifests as a gradual change in low-frequency brightness, causing significant grayscale differences in the same anatomical tissue at different locations, thereby obscuring periodontal details, reducing image quality, and increasing the difficulty of clinical interpretation. Existing image equalization techniques mainly employ window width and level adjustments or global grayscale mapping (such as contrast stretching and histogram equalization), relying on manual parameter settings and failing to eliminate brightness bias from the perspective of imaging physics. Furthermore, while bias field models such as N4ITK can estimate low-frequency brightness changes, they are based on variational solutions and maximum likelihood inference, resulting in computational complexity, long iteration times, and difficulty in real-time application. Simultaneously, these methods primarily focus on low-frequency trends, and their ability to correct brightness in high-gradient structural regions such as teeth and jawbones remains insufficient.

[0003] Currently, there is a severe imbalance in local tissue contrast, and there is a lack of adaptive modeling for changes in spatial brightness and differences in structural gradients. This easily leads to local overexposure or underexposure, blurring details such as tooth roots and periodontal ligament cavities. The effect depends on equipment and manual adjustment. Different equipment and imaging parameters result in significant bias differences. The fixed parameters of traditional methods are difficult to adapt, resulting in insufficient image consistency and repeatability, as well as insufficient real-time performance. The complex iteration of bias estimation algorithms limits the development needs of rapid clinical imaging and automatic diagnosis. Summary of the Invention

[0004] In view of the problems existing in the panoramic CBCT brightness equalization method based on Gaussian low-frequency modeling and Poisson reconstruction joint bias estimation, this invention is proposed.

[0005] Therefore, the problem to be solved by this invention is that the uneven spatial brightness and contrast imbalance of panoramic CBCT images caused by scattering effects, uneven detector response and differences in X-ray attenuation affect the clear presentation of key anatomical structures such as tooth roots and alveolar bone and the reliability of clinical diagnosis.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, embodiments of the present invention provide a panoramic CBCT brightness equalization method based on Gaussian low-frequency modeling and Poisson reconstruction joint bias estimation, which includes converting the image data file to be processed in Raw format into 32-bit floating-point data format according to the ini configuration file; A Gaussian kernel is created based on the mathematical expression, wherein the Gaussian kernel is a two-dimensional normal distribution; Construct the Laplacian residual and update the image using gradient descent to reconstruct the brightness gradient distribution; Perform pixel-by-pixel division to obtain the image after bias field correction. Then, stack the bias field-corrected images proportionally to obtain the brightness-corrected image and output it.

[0007] As a preferred embodiment of the panoramic CBCT brightness equalization method based on Gaussian low-frequency modeling and Poisson reconstruction joint bias estimation described in this invention, the step of converting the image data file to be processed in Raw format to 32-bit floating-point data format according to the ini configuration file includes the following steps: Open the handle to the image data to be processed and read it line by line in binary format; Convert binary data to decimal data, with each pixel represented by 16 bits; Read the ini configuration file in the same directory to obtain the image's width W and height H; Convert the data to decimal format and arrange it in rows and columns to create an image with a width of W and a height of H.

[0008] As a preferred embodiment of the panoramic CBCT brightness equalization method based on Gaussian low-frequency modeling and Poisson reconstruction bias estimation described in this invention, wherein: the creation of the Gaussian kernel according to the mathematical expression includes the following expression of the two-dimensional Gaussian distribution function: in, Represents the x-axis, Represents the ordinate, This represents pi, which is 3.1415926. is the base of the natural logarithm. Indicates variance; Substituting the Gaussian kernel size into the formula yields a fixed normal distribution Gaussian kernel. Using a normally distributed Gaussian kernel The image to be processed is copied and converted into a floating-point array, then convolved to obtain the low-frequency brightness variation distribution, which is the Gaussian reconstructed brightness field. .

[0009] As a preferred embodiment of the panoramic CBCT brightness equalization method based on Gaussian low-frequency modeling and Poisson reconstruction bias estimation described in this invention, the reconstructed brightness gradient distribution includes the following steps: The image data to be processed, obtained by converting the Raw format of the image data file to 32-bit floating-point data format according to the ini configuration file, will be copied and converted. Set up the iteration process; A single loop completes the iterative cycle, reconstructing the brightness gradient distribution.

[0010] As a preferred embodiment of the panoramic CBCT brightness equalization method based on Gaussian low-frequency modeling and Poisson reconstruction joint bias estimation described in this invention, wherein: the copying and conversion of the image data to be processed includes copying the image data to be processed and converting it into a floating-point array, denoted as the initial iteration image. ; The iteration setting process includes setting the number of iterations. Normalized stable term Update step size and termination threshold ; The iterative loop includes the following steps: Calculate the first-order gradient; Calculate the gradient magnitude and normalize the gradient direction, and normalize the gradient vector per pixel to obtain a unit direction vector; Calculate the Laplace operator; The image is updated using the Laplace value as the residual term of the Poisson equation in the form of gradient descent. Determine whether to stop the iteration process prematurely; After the iteration is complete, the Poisson reconstructed brightness field is obtained. As the result, output it.

[0011] As a preferred embodiment of the panoramic CBCT brightness equalization method based on Gaussian low-frequency modeling and Poisson reconstruction joint bias estimation described in this invention, wherein: the calculation of the first-order gradient includes calculating the current image respectively The numerical gradients in the horizontal and vertical directions are expressed by the following formulas: in, Representing an image gradient vector, Indicates the first The image of the next iteration, This represents the first-order partial derivative of the image in the horizontal x-direction. This represents the first-order partial derivative of the image in the vertical y-direction; The formula for calculating the gradient magnitude and normalizing the gradient direction is as follows: in, Indicates the first Gradient magnitude of the image in the next iteration. This represents the second-order partial derivative of the image in the horizontal direction. This represents the second-order partial derivative of the image in the vertical direction; This represents the normalized gradient vector. This represents the magnitude of the gradient vector. Represents a small constant; The calculation of the Laplacian operator includes processing the current image. Calculate Laplace To quantify the second-order changes in local brightness; The formula for updating the image using the Laplacian value as the residual term of the Poisson equation, in the form of gradient descent, is as follows: in, Indicates the first The image of the next iteration, Step size factor; The determination of whether to stop the iteration process early includes the iteration process. Determine whether to stop early when the difference is less than a threshold. It may be terminated early.

[0012] As a preferred embodiment of the panoramic CBCT brightness equalization method based on Gaussian low-frequency modeling and Poisson reconstruction joint bias estimation described in this invention, wherein: the step of obtaining a brightness-corrected image and outputting the obtained corrected image includes... and The final brightness-corrected image is obtained by proportionally superimposing the images. The specific formula is as follows: in, For Gaussian image factors, Poisson impact factor This is the final brightness-corrected image. The image is after Gaussian reconstruction and brightness field correction. This is the image after Poisson reconstruction brightness field correction.

[0013] Secondly, embodiments of the present invention provide a panoramic CBCT brightness equalization system based on Gaussian low-frequency modeling and Poisson reconstruction joint bias estimation, comprising: a data preprocessing module, which converts the raw format of the image data file to be processed into a 32-bit floating-point data format according to an ini configuration file; a Gaussian low-frequency bias modeling module, which creates a Gaussian kernel based on a mathematical expression, wherein the Gaussian kernel is a two-dimensional normal distribution; a Poisson gradient reconstruction module, which constructs a Laplacian residual and updates the image in a gradient descent manner to reconstruct the brightness gradient distribution; and a bias field correction and fusion module, which performs pixel-by-pixel division to obtain an image after bias field correction, and superimposes the images after bias field correction according to a ratio to obtain a brightness-corrected image and outputs it.

[0014] Thirdly, embodiments of the present invention provide a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement any step of the above-described panoramic CBCT brightness equalization method based on Gaussian low-frequency modeling and Poisson reconstruction joint bias estimation.

[0015] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program thereon, wherein: when the computer program is executed by a processor, it implements any step of the above-described panoramic CBCT brightness equalization method based on Gaussian low-frequency modeling and Poisson reconstruction joint bias estimation.

[0016] The beneficial effects of this invention are as follows: This invention simultaneously captures low-frequency trends in brightness and structural gradient changes through a dual-branch bias field learning mechanism, achieving rapid and accurate estimation of the bias field, and adaptively fusing them into a highly robust bias correction model. Furthermore, a grayscale stabilization strategy is introduced to achieve brightness consistency and contrast optimization in CBCT panoramic images without requiring manual adjustment of window width and level, effectively improving the discernibility of tissue details and cross-device imaging consistency, thereby significantly enhancing the quality and stability of clinical diagnosis. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 The flowchart illustrates an algorithm for a panoramic CBCT brightness equalization method based on joint bias estimation using Gaussian low-frequency modeling and Poisson reconstruction, as provided in an embodiment of the present invention.

[0018] Figure 2This is a schematic diagram before enhancement, illustrating a panoramic CBCT brightness equalization method based on Gaussian low-frequency modeling and Poisson reconstruction joint bias estimation, provided in an embodiment of the present invention.

[0019] Figure 3 This is a schematic diagram showing the enhancement process before and after Gaussian low-frequency modeling and Poisson reconstruction joint bias estimation of a panoramic CBCT brightness equalization method based on Gaussian low-frequency modeling and Poisson reconstruction, as provided in an embodiment of the present invention. Detailed Implementation

[0020] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0023] This invention is described in detail with reference to the schematic diagrams. When describing the embodiments of this invention, for ease of explanation, the cross-sectional views illustrating the device structure may be partially enlarged, not according to the usual scale. Furthermore, the schematic diagrams are merely examples and should not be construed as limiting the scope of protection of this invention. In actual fabrication, the three-dimensional spatial dimensions of length, width, and depth should be included.

[0024] Furthermore, in the description of this invention, it should be noted that the terms "upper," "lower," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used solely for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. In addition, the terms "first," "second," or "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0025] Unless otherwise explicitly specified and limited, the terms "installation," "connection," and "joining" in this invention should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; similarly, they can refer to mechanical connections, electrical connections, or direct connections, or indirect connections through an intermediate medium, or internal connections between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0026] Example Reference Figures 1-3 This is the first embodiment of the present invention, which provides a panoramic CBCT brightness equalization method based on joint bias estimation of Gaussian low-frequency modeling and Poisson reconstruction, including: S1: Convert the image data file to be processed from Raw format to 32-bit floating-point data format according to the ini configuration file.

[0027] The process of converting the image data file from Raw format to 32-bit floating-point data format according to the ini configuration file includes the following steps: Open the handle to the image data to be processed and read it line by line in binary format; Convert binary data to decimal data, with each pixel represented by 16 bits; Read the ini configuration file in the same directory to obtain the image's width W and height H; Convert the data to decimal format and arrange it in rows and columns to create image data with a width of W and a height of H.

[0028] S2: Create a Gaussian kernel based on the mathematical expression. The Gaussian kernel is a two-dimensional normal distribution.

[0029] The creation of the Gaussian kernel based on the mathematical expression includes the following two-dimensional Gaussian distribution function expression: in, Represents the x-axis, Represents the ordinate, This represents pi, which is 3.1415926. is the base of the natural logarithm. Indicates variance; Substituting the Gaussian kernel size into the formula yields a fixed normal distribution Gaussian kernel. Using a normally distributed Gaussian kernel The image to be processed is copied and converted into a floating-point array, then convolved to obtain the low-frequency brightness variation distribution, which is the Gaussian reconstructed brightness field. .

[0030] Furthermore, the Gaussian kernel size is set to 5 x 5, and the variance is set to 255. These values ​​are then substituted into formula (1) to calculate a fixed normally distributed Gaussian kernel G. The image to be processed obtained in step 1 is copied using the Gaussian kernel G and converted to a floating-point array. A convolution operation is then performed (a sliding window is created, moving from the upper left of the image row by row to the lower right, with the window size matching the Gaussian kernel size. The values ​​within the sliding window are multiplied pixel by pixel with the values ​​of the Gaussian kernel, resulting in a final output. This output is used as the result of this sliding window operation. Moving from the upper left to the lower right row constitutes one convolution operation. The results obtained from the sliding window and the Gaussian kernel calculation form a result matrix of the same size as the original image, which is the convolution result). This yields the low-frequency brightness variation distribution, which is the Gaussian reconstructed brightness field. .

[0031] S3: Construct the Laplacian residual and update the image using gradient descent to reconstruct the brightness gradient distribution.

[0032] The reconstruction of the brightness gradient distribution includes the following steps: The image data to be processed, obtained by converting the Raw format of the image data file to 32-bit floating-point data format according to the ini configuration file, will be copied and converted. Set up the iteration process; A single loop completes the iterative cycle, reconstructing the brightness gradient distribution.

[0033] Furthermore, set the iteration count T (default value 100) and the normalization stability term. ( And update the step size (learning rate). (Default is 0.1, balancing convergence speed and stability), Termination threshold ( ).

[0034] S3.1: Copying and converting the image data to be processed includes copying the image data to be processed and converting it into a floating-point array, denoted as the initial iterative image. ; Setting up the iteration process includes setting the number of iterations. Normalized stable term Update step size and termination threshold ; The iterative loop includes the following steps: Calculate the first-order gradient; Calculate the gradient magnitude and normalize the gradient direction, and normalize the gradient vector per pixel to obtain a unit direction vector; Calculate the Laplace operator; The image is updated using the Laplace value as the residual term of the Poisson equation in the form of gradient descent. Determine whether to stop the iteration process prematurely; After the iteration is complete, the Poisson reconstructed brightness field is obtained. As the result, output it.

[0035] S3.2: Calculating the first-order gradient includes calculating the gradient of the current image separately. The numerical gradients in the horizontal and vertical directions are expressed by the following formulas: in, Representing an image gradient vector, Indicates the first The image of the next iteration, This represents the first-order partial derivative of the image in the horizontal x-direction. This represents the first-order partial derivative of the image in the vertical y-direction; The formula for calculating the gradient magnitude and normalizing the gradient direction is as follows: in, Indicates the first Gradient magnitude of the image in the next iteration. This represents the second-order partial derivative of the image in the horizontal direction. This represents the second-order partial derivative of the image in the vertical direction; This represents the normalized gradient vector. This represents the magnitude of the gradient vector. Represents a small constant; Calculating the Laplacian operator involves processing the current image. Calculate Laplace To quantify the second-order changes in local brightness; The formula for updating the image using the Laplacian value as the residual term of the Poisson equation, in the form of gradient descent, is as follows: in, Indicates the first The image of the next iteration, Step size factor; Determining whether to stop the iteration process prematurely includes the iteration process itself. Determine whether to stop early when the difference is less than a threshold. It may be terminated early.

[0036] Furthermore, for a single loop .

[0037] S4: Perform pixel-by-pixel division to obtain the image after bias field correction. Then, superimpose the images after bias field correction according to the ratio to obtain the brightness-corrected image and output it.

[0038] This involves obtaining a brightness-corrected image and outputting the corrected image. and The final brightness-corrected image is obtained by proportionally superimposing the images. The specific formula is as follows: in, For Gaussian image factors, Poisson impact factor This is the final brightness-corrected image. The image is after Gaussian reconstruction and brightness field correction. This is the image after Poisson reconstruction brightness field correction.

[0039] Furthermore, among them The Gaussian image factor is set to 0.8 here. The Poisson impact factor is set to 0.2 here.

[0040] In a preferred embodiment, a panoramic CBCT brightness equalization system based on Gaussian low-frequency modeling and Poisson reconstruction joint bias estimation includes a data preprocessing module that converts the raw format of the image data file to be processed into a 32-bit floating-point data format according to an ini configuration file; a Gaussian low-frequency bias modeling module that creates a Gaussian kernel based on a mathematical expression, wherein the Gaussian kernel is a two-dimensional normal distribution; a Poisson gradient reconstruction module that constructs a Laplacian residual and updates the image in a gradient descent manner to reconstruct the brightness gradient distribution; and a bias field correction and fusion module that performs pixel-by-pixel division to obtain an image after bias field correction, and then superimposes the bias field corrected images proportionally to obtain a brightness-corrected image and outputs it.

[0041] The aforementioned unit modules can be embedded in the processor of a computer device in hardware form or independent of it, or they can be stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each of the above modules.

[0042] In one embodiment, a computer device is provided, which may be a terminal. The computer device includes a processor, memory, a communication interface, a display screen, and an input device connected via a system bus. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage medium. The communication interface of the computer device is used for wired or wireless communication with external terminals. Wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen of the computer device may be an LCD screen or an e-ink display screen. The input device of the computer device may be a touch layer covering the display screen, or buttons, a trackball, or a touchpad located on the casing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0043] In summary, this invention simultaneously captures low-frequency trends in brightness and structural gradient changes through a dual-branch bias field learning mechanism, achieving rapid and accurate estimation of the bias field, and adaptively fusing them into a highly robust bias correction model. Furthermore, a grayscale stabilization strategy is introduced to achieve brightness uniformity and contrast optimization in CBCT panoramic images without requiring manual adjustment of window width and level, effectively improving the discernibility of tissue details and cross-device imaging consistency, thereby significantly enhancing the quality and stability of clinical diagnosis.

[0044] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A panoramic CBCT brightness equalization method based on joint bias estimation of Gaussian low-frequency modeling and Poisson reconstruction, characterized in that: include, Based on the ini configuration file, the image data file to be processed in Raw format is converted to 32-bit floating-point data format; A Gaussian kernel is created based on the mathematical expression, wherein the Gaussian kernel is a two-dimensional normal distribution; Construct the Laplacian residual and update the image using gradient descent to reconstruct the brightness gradient distribution; Perform pixel-by-pixel division to obtain the image after bias field correction. Then, stack the bias field-corrected images proportionally to obtain the brightness-corrected image and output it.

2. The panoramic CBCT brightness equalization method based on joint bias estimation of Gaussian low-frequency modeling and Poisson reconstruction as described in claim 1, characterized in that: The process of converting the image data file to be processed from Raw format to 32-bit floating-point data format according to the ini configuration file includes the following steps: Open the handle to the image data to be processed and read it line by line in binary format; Convert binary data to decimal data, with each pixel represented by 16 bits; Read the ini configuration file in the same directory to obtain the image's width W and height H; Convert the data to decimal format and arrange it in rows and columns to create image data with a width of W and a height of H.

3. The panoramic CBCT brightness equalization method based on joint bias estimation of Gaussian low-frequency modeling and Poisson reconstruction as described in claim 2, characterized in that: The creation of the Gaussian kernel based on the mathematical expression includes the following two-dimensional Gaussian distribution function expression: in, Represents the x-axis, Represents the ordinate, This represents pi, which is 3.1415926. is the base of the natural logarithm. Indicates variance; Substituting the Gaussian kernel size into the formula yields a fixed normal distribution Gaussian kernel. Using a normally distributed Gaussian kernel The image to be processed is copied and converted into a floating-point array, then convolved to obtain the low-frequency brightness variation distribution, which is the Gaussian reconstructed brightness field. .

4. The panoramic CBCT brightness equalization method based on joint bias estimation of Gaussian low-frequency modeling and Poisson reconstruction as described in claim 3, characterized in that: The reconstructed brightness gradient distribution includes the following steps: The image data to be processed, obtained by converting the Raw format of the image data file to 32-bit floating-point data format according to the ini configuration file, will be copied and converted. Set up the iteration process; A single loop completes the iterative cycle, reconstructing the brightness gradient distribution.

5. The panoramic CBCT brightness equalization method based on joint bias estimation of Gaussian low-frequency modeling and Poisson reconstruction as described in claim 4, characterized in that: The copying and conversion of the image data to be processed includes copying the image data to be processed and converting it into a floating-point array, denoted as the initial iterative image. ; The iteration setting process includes setting the number of iterations. Normalized stable term Update step size and termination threshold ; The iterative loop includes the following steps: Calculate the first-order gradient; Calculate the gradient magnitude and normalize the gradient direction, and normalize the gradient vector per pixel to obtain a unit direction vector; Calculate the Laplace operator; The image is updated using the Laplace value as the residual term of the Poisson equation in the form of gradient descent. Determine whether to stop the iteration process prematurely; After the iteration is complete, the Poisson reconstructed brightness field is obtained. As the result, output it.

6. The panoramic CBCT brightness equalization method based on joint bias estimation of Gaussian low-frequency modeling and Poisson reconstruction as described in claim 5, characterized in that: The calculation of the first-order gradient includes calculating the current image respectively. The numerical gradients in the horizontal and vertical directions are expressed by the following formulas: in, Representing an image gradient vector, Indicates the first The image of the next iteration, This represents the first-order partial derivative of the image in the horizontal x-direction. This represents the first-order partial derivative of the image in the vertical y-direction; The formula for calculating the gradient magnitude and normalizing the gradient direction is as follows: in, Indicates the first Gradient magnitude of the image in the next iteration. This represents the second-order partial derivative of the image in the horizontal direction. This represents the second-order partial derivative of the image in the vertical direction; This represents the normalized gradient vector. This represents the magnitude of the gradient vector. Represents a small constant; The calculation of the Laplacian operator includes processing the current image. Calculate Laplace To quantify the second-order changes in local brightness; The formula for updating the image using the Laplacian value as the residual term of the Poisson equation, in the form of gradient descent, is as follows: in, Indicates the first The image of the next iteration, Step size factor; The determination of whether to stop the iteration process early includes the iteration process. Determine whether to stop early when the difference is less than a threshold. It may be terminated early.

7. The panoramic CBCT brightness equalization method based on joint bias estimation of Gaussian low-frequency modeling and Poisson reconstruction as described in claim 6, characterized in that: The process of obtaining and outputting the brightness-corrected image includes... and The final brightness-corrected image is obtained by proportionally superimposing the images. The specific formula is as follows: in, For Gaussian image factors, Poisson impact factor This is the final brightness-corrected image. The image is after Gaussian reconstruction and brightness field correction. This is the image after Poisson reconstruction brightness field correction.

8. A panoramic CBCT brightness equalization system based on joint bias estimation of Gaussian low-frequency modeling and Poisson reconstruction, based on the panoramic CBCT brightness equalization method based on joint bias estimation of Gaussian low-frequency modeling and Poisson reconstruction as described in any one of claims 1 to 7, characterized in that: include, The data preprocessing module converts the image data file to be processed from Raw format to 32-bit floating-point data format according to the ini configuration file. The Gaussian low-frequency bias modeling module creates a Gaussian kernel based on a mathematical expression, wherein the Gaussian kernel is a two-dimensional normal distribution. The Poisson gradient reconstruction module constructs the Laplacian residual and updates the image using gradient descent to reconstruct the brightness gradient distribution. The offset field correction and fusion module performs pixel-by-pixel division to obtain the image after offset field correction, and then superimposes the offset field corrected images according to the ratio to obtain the brightness corrected image and output it.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the panoramic CBCT brightness equalization method based on Gaussian low-frequency modeling and Poisson reconstruction joint bias estimation as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the panoramic CBCT brightness equalization method based on Gaussian low-frequency modeling and Poisson reconstruction bias estimation as described in any one of claims 1 to 7.

Images