A Fast CT Ring Artifact Correction Method and System Based on Projection-Averaged Images

Abstract

This invention discloses a fast CT ring artifact correction method and system based on projected average images, belonging to the field of image processing technology. The key technical points are: First, the original projected images at different angles are superimposed and averaged to obtain a projected average image. Then, the projected average image is differentiated to obtain a differential image reflecting the gray-level abrupt changes in the projected average image. Next, the filter kernel window size at different pixel positions is calculated based on the differential image. Then, the projected average image is adaptively low-pass filtered based on the filter kernel window size to obtain a low-frequency image. Finally, the projected average image is divided by the low-frequency image to obtain a gain-corrected image. Finally, the pixel gain of the original projected images at different angles is corrected based on the gain-corrected image to obtain a projection sequence image with consistent gain. This invention extracts the gain inconsistency characteristics of all pixels from the projected average image, eliminating or mitigating ring artifacts without significantly reducing image quality.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and more specifically, to a fast CT ring artifact correction method and system based on projected average images. Background Technology

[0002] Computed tomography (CT) is a non-destructive testing technique widely used in medical diagnosis, industrial inspection, and other fields. CT acquires images of an object from different angles using various projections and employs specific image reconstruction algorithms to obtain images of the object's cross-section, which are then used for medical diagnosis or industrial defect detection. However, the presence of artifacts often significantly degrades the quality of CT images, with ring artifacts being a common type. This type of artifact is mainly caused by inconsistent detector response to X-rays, non-linearity between X-ray intensity and pixel grayscale, and detector pixel defects. Furthermore, insufficient X-ray photon counts, data acquisition and transmission failures, and detector surface contamination or damage can also cause data deviations, ultimately leading to ring artifacts in the reconstructed CT images. This complicates post-processing tasks such as image segmentation, recognition, measurement, and analysis.

[0003] Currently, there are two main types of image processing techniques for eliminating ring artifacts: preprocessing methods based on projection data and postprocessing methods based on reconstructed images. Preprocessing methods typically locate and identify vertical stripes in the projected sine wave using Huough transform, feature recognition, or the Canny operator, and then remove these stripes using filtering or interpolation to eliminate the ring artifacts. Postprocessing methods generally use polar coordinate transformation to convert the ring artifacts into bar artifacts, and then use various complex image processing methods to eliminate the bar artifacts. While these correction methods have good suppression effects on ring artifacts, they also have some shortcomings. For example, in preprocessing methods, generating a projected sine wave is necessary. When there are many pixels with inconsistent detector responses, identifying the vertical stripes in the sine wave requires significant computational resources, and overall filtering and interpolation can reduce image resolution to some extent. Postprocessing methods also suffer from low computational efficiency and image degradation; for example, the correction process requires two polar coordinate transformations and complex filtering, iterative optimization, and image reconstruction processes.

[0004] Therefore, efficiently eliminating ring artifacts without significantly reducing image quality remains a challenging task. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the present invention aims to provide a fast CT ring artifact correction method and system based on projected average images. This method extracts the gain inconsistency characteristics of all pixels from the projected average image and uses corresponding correction coefficients to correct the gain of each pixel's grayscale, thereby achieving an ideal detection image with near-consistent gain. This eliminates or reduces ring artifacts without significantly degrading image quality.

[0006] The above-mentioned technical objective of the present invention is achieved through the following technical solution:

[0007] Firstly, a fast CT ring artifact correction method based on projected averaged images is provided, including the following steps:

[0008] The original projected images at different angles are superimposed and averaged to obtain the averaged projected image.

[0009] By performing a differential operation on the projected average image, a differential image reflecting the gray-level abrupt changes in the projected average image is obtained.

[0010] Calculate the filter kernel window size at different pixel locations based on the differential image;

[0011] The projected average image is adaptively low-pass filtered based on the filter kernel window size to obtain a low-frequency image;

[0012] Divide the average projected image by the low-frequency image to obtain the gain-corrected image;

[0013] The pixel gain of the original projection images at different angles is corrected based on the gain correction image to obtain a projection sequence image with consistent gain.

[0014] Furthermore, the specific formula for calculating the average projected image is as follows:

[0015]

[0016] in, This represents the projected average image; m represents the image row number; n represents the image column number; p i (m, n) represents the i-th original projected image; N represents the number of original projected images.

[0017] Furthermore, the differentiation operation employs a first-order Sobel filter or a second-order Laplacian filter.

[0018] Furthermore, the calculation formula for the differential image using first-order differential operations is as follows:

[0019] g(m, n) = |G x (m, n)|+|G y (m, n)|

[0020] Where g(m, n) represents the differential image; G x (m, n) represents the image gradient along the column direction; G y (m, n) represents the image gradient along the row direction; m represents the image row index; n represents the image column index;

[0021] Alternatively, the calculation formula for the differential image using second-order differential operations is as follows:

[0022]

[0023] in, Represents the Laplace operator; This represents the average projected image.

[0024] Furthermore, in the spatial domain, the formula for calculating the size of the filter kernel window is as follows:

[0025] k(m, n) = k min +[g max -g(m,n)](k max -k min ) / (g max -g min )

[0026] Alternatively, in the frequency domain, the formula for calculating the filter kernel window size is as follows:

[0027] k(m, n) = k max -[g max -g(m,n)](k max -k min ) / (g max -g min )

[0028] Where k(m, n) represents the filter kernel window size of the adaptive low-pass filter; k min This represents the minimum size of the filter kernel window; k max The maximum value of the filter kernel window size is represented by g(m, n); g(m, n) represents the differential image; m represents the image row index; n represents the image column index; g min The minimum value of the differential graph is represented by g. max This represents the maximum value of the differential image.

[0029] Furthermore, the specific formula for calculating the low-frequency image is as follows:

[0030]

[0031] Where f(m, n) represents the low-frequency image; LowPassFilter represents an adaptive low-pass filter; denoted as the projected average image; k(m, n) represents the filter kernel window size of the adaptive low-pass filter; m represents the image row number; n represents the image column number.

[0032] Furthermore, the specific formula for calculating the projected image sequence is as follows:

[0033]

[0034] in, p represents the i-th projection sequence image; i (m, n) represents the i-th original projected image; h(m, n) represents the gain-corrected image; m represents the image row number; n represents the image column number.

[0035] Secondly, a fast CT ring artifact correction system based on projected averaged images is provided, including:

[0036] The averaging module is used to superimpose and average the original projected images at different angles to obtain the average projected image.

[0037] The differential operation module is used to perform differential operations on the projected average image to obtain a differential image that reflects the gray-level abrupt changes in the projected average image.

[0038] The size calculation module is used to calculate the size of the filter kernel window at different pixel positions based on the differential image;

[0039] The low-pass filter module is used to adaptively low-pass filter the projected average image according to the size of the filter kernel window to obtain a low-frequency image;

[0040] The gain correction module is used to divide the projected average image by the low-frequency image to obtain the gain-corrected image;

[0041] The gain correction module is used to correct the pixel gain of the original projection image at different angles based on the gain correction image, so as to obtain a projection sequence image with consistent gain.

[0042] Thirdly, a computer terminal is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the fast CT ring artifact correction method based on projected average images as described in any one of the first aspects.

[0043] Fourthly, a computer-readable medium is provided having a computer program stored thereon, the computer program being executed by a processor to implement the fast CT ring artifact correction method based on projected average images as described in any one of the first aspects.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] 1. The fast CT ring artifact correction method based on projection average image provided by the present invention extracts the gain inconsistency characteristics of all pixels from the projection average image, and uses the corresponding correction coefficient to correct the gain of each pixel gray level, so as to achieve an ideal detection image with near-consistent gain, thereby eliminating or reducing ring artifacts without significantly reducing image quality.

[0046] 2. This invention belongs to the preprocessing method. The effect of ring artifact correction can be adjusted by simply setting the adaptive filter kernel window size range. The entire process only involves one differential linear filtering and adaptive low-pass filtering calculation for a single image. All other calculations are pixel-by-pixel arithmetic operations, which can be performed in parallel. The computational resource consumption is low and the computational efficiency is very high.

[0047] 3. The present invention performs a simultaneous correction process that only divides the projected image, thus preserving the high-frequency detail features and spatial resolution of the original image.

[0048] 4. This invention performs calculations on the acquired projection data, has no strict requirements on the hardware system or image acquisition method, has a wide range of applications, and is effective for CT scanning methods such as parallel beam, fan beam, and cone beam. It can correct the gain inconsistency between pixels and repair abnormal pixels such as local defects or necrotic pixels. Attached Figure Description

[0049] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings:

[0050] Figure 1 This is a flowchart from an embodiment of the present invention;

[0051] Figure 2 This is an uncorrected slice image of the annular artifact in an embodiment of the present invention;

[0052] Figure 3 This is a slice image of ring artifact correction based on median filtering with a constant kernel (k=35) in an embodiment of the present invention.

[0053] Figure 4 This is a slice image of ring artifact correction based on adaptive kernel (k=5-35) median filtering in an embodiment of the present invention;

[0054] Figure 5 This is a system block diagram in an embodiment of the present invention. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0056] Example 1: A fast CT ring artifact correction method based on projected averaged images, such as... Figure 1 As shown, it is achieved through the following steps.

[0057] Step 1: Overlay and average the original projected images at different angles to obtain the average projected image.

[0058] If the detector pixel gain is consistent, the high-frequency image information at different angles accumulates and aliased with position changes, resulting in a relatively smooth low-frequency background image. However, for cases where pixel gain is inconsistent, the inconsistent pixel positions are fixed, so this inconsistency still exists in the projected average image, manifesting as high-frequency information. For locally defective or necrotic pixels (collectively referred to as anomalous pixels), this location appears as significant grainy noise in the projected average image.

[0059] The specific formula for calculating the projected average image is as follows:

[0060]

[0061] in, This represents the projected average image; m represents the image row number; n represents the image column number; p i (m, n) represents the i-th original projected image; N represents the number of original projected images.

[0062] Step 2: Perform a differentiation operation on the projected average image, such as a first-order Sobel filter or a second-order Laplacian filter, to obtain a new differential image. The size of the differential kernel is appropriately selected according to the specific scene, generally 3×3. This image can reflect the parts of the projected average image where the grayscale changes drastically, such as the boundary regions between projected images of different materials.

[0063] The specific formula for calculating the differential image using first-order differential operations is as follows:

[0064]

[0065] Where g(m, n) represents the differential image; G x (m, n) represents the image gradient along the column direction; G y (m, n) represents the image gradient along the row direction; m represents the image row number; n represents the image column number.

[0066] The specific formula for calculating the differential image using second-order differential operations is as follows:

[0067]

[0068] in, Represents the Laplace operator; This represents the average projected image.

[0069] Step 3: Calculate the size of the filter kernel window at different pixel locations based on the differential image to prepare for subsequent adaptive low-pass filtering.

[0070] The filter kernel window size k(m, n) is determined based on the differential image grayscale value at the minimum kernel (minimum value) k. min and the maximum kernel (maximum value) k max A linear mapping is performed between them. When the gray value of the differential image reaches its maximum value, the minimum kernel is used; when it reaches its minimum value, the maximum kernel is used; and for intermediate values, the nearest smallest odd integer is used based on the calculation result.

[0071] Adaptive low-pass filtering can be performed in the spatial domain, including but not limited to adaptive mean filtering, adaptive median filtering, and adaptive Gaussian filtering, as well as in the frequency domain, including but not limited to adaptive ideal frequency domain low-pass filtering, adaptive Butterworth low-pass filtering, and adaptive Gaussian low-pass filtering.

[0072] In the spatial domain, the formula for calculating the filter kernel window size is as follows:

[0073] k(m, n) = k min +[g max -g(m,n)](k max -k min ) / (g max -g min )

[0074] Alternatively, in the frequency domain, the formula for calculating the filter kernel window size is as follows:

[0075] k(m, n) = k max -[g max -g(m,n)](k max -k min ) / (g max -g min )

[0076] Where k(m, n) represents the filter kernel window size of the adaptive low-pass filter; k min This represents the minimum size of the filter kernel window; k max This represents the maximum value of the filter kernel window size; g min The minimum value of the differential graph is represented by g. maxRepresents the maximum value of the differential image. The filter kernel window size range [k] min k max The size of the adaptive filtering window should be selected based on the specific scenario and the desired correction effect. The window size should be the nearest odd number to the value calculated by the above formula.

[0077] Step 4: Perform adaptive low-pass filtering on the projected average image based on the filter kernel window size to obtain a low-frequency image.

[0078] The specific formula for calculating low-frequency images is as follows:

[0079]

[0080] Where f(m, n) represents the low-frequency image; LowPassFilter represents an adaptive low-pass filter; The image represents the projected average image; k(m, n) represents the size of the adaptive low-pass filter kernel window; m represents the image row index; n represents the image column index. The adaptive low-pass filter window can be a two-dimensional rectangular window or a one-dimensional window arranged by rows or columns. Compared with conventional kernel low-pass filtering, adaptive low-pass filtering can better eliminate new artifacts caused by areas with large gray-level changes in the projected average image (such as the boundaries of different materials or areas with abrupt gray-level changes).

[0081] Step 5: Divide the projected average image by the low-frequency image to obtain the gain-corrected image.

[0082] The specific formula for calculating the gain-corrected image is as follows:

[0083]

[0084] Where h(m, n) represents the gain correction image.

[0085] Step 6: Based on the gain correction image, perform pixel gain correction on the original projection images at different angles to obtain projection sequence images with consistent gain, which are used for CT reconstruction to eliminate or reduce ring artifacts.

[0086] The specific formula for calculating the projected image sequence is as follows:

[0087]

[0088] in, p represents the i-th projection sequence image; i (m, n) represents the i-th original projected image; h(m, n) represents the gain-corrected image; m represents the image row number; n represents the image column number.

[0089] The correction method described in this invention is a preprocessing method. It only requires setting the adaptive filter kernel window size range to adjust the effect of annular artifact correction. The entire process involves only one differential linear filtering and adaptive low-pass filtering calculation for a single image; all other calculations are pixel-by-pixel arithmetic operations, allowing for parallel computation with low computational resource consumption and very high computational efficiency. Furthermore, the correction process only performs division operations on the projected image, preserving the high-frequency detail features and spatial resolution of the original image. This method performs calculations on the acquired projection data, has no strict requirements on the hardware system or image acquisition method, and has a wide range of applications. It is effective for parallel beam, fan beam, and cone beam CT scanning methods, correcting gain inconsistencies between pixels and repairing abnormal pixels such as local defects or necrotic pixels.

[0090] like Figure 2 As shown, an industrial cone-beam CT scanner was used to scan the part, obtaining the original projected image data. Dark-field and flat-field corrections were then performed on the original projected image to obtain a projected image with background subtraction and inconsistent light field. Due to the inconsistency in pixel response, conventional CT reconstruction processes exhibit severe ring artifacts. To verify the correction effect of this method on anomalous pixels, two anomalous pixels were pre-selected in the projected image. The projection values ​​of these two pixels are constant across all angles. It can be seen that the two anomalous pixels in the uncorrected image cause concentric ring artifacts with very prominent contrast.

[0091] After processing in steps 1-6, the adaptive filter kernel size at each pixel is calculated in the spatial domain, with the maximum filter kernel set to 35×35 and the minimum kernel set to 5×5.

[0092] The ring artifacts in the CT reconstruction slices of the parts were largely eliminated, such as Figure 4 As shown, the high-contrast rings caused by the two abnormal pixels disappear, significantly improving image quality and preserving good spatial details. While median filtering based on a constant kernel can also eliminate ring artifacts, a large constant kernel can introduce new artifacts, such as… Figure 3 The raised circular ring shown is not a feature of the image itself, but a novel artifact introduced by large kernel median filtering. The adaptive median filtering method effectively suppresses this type of artifact, while also largely eliminating the circular artifact.

[0093] Example 2: A fast CT ring artifact correction system based on projected average images. This system is used to implement the fast CT ring artifact correction method based on projected average images described in Example 1, such as... Figure 5 As shown, it includes an averaging module, a differential operation module, a size calculation module, a low-pass filtering module, a gain correction module, and a gain adjustment module.

[0094] The system includes: an averaging module for superimposing and averaging the original projected images at different angles to obtain an average projected image; a differentiation module for performing differentiation on the average projected image to obtain a differential image reflecting the gray-level abrupt changes in the average projected image; a size calculation module for calculating the size of the filter kernel window at different pixel positions based on the differential image; a low-pass filtering module for adaptively low-pass filtering the average projected image based on the filter kernel window size to obtain a low-frequency image; a gain correction module for dividing the average projected image by the low-frequency image to obtain a gain-corrected image; and a gain adjustment module for adjusting the pixel gain of the original projected images at different angles based on the gain-corrected image to obtain a projection sequence image with consistent gain.

[0095] Working principle: This invention extracts the gain inconsistency characteristics of all pixels from the projected average image and uses the corresponding correction coefficient to correct the gain of each pixel grayscale, thereby achieving an ideal detection image with near-consistent gain, thus eliminating or reducing ring artifacts without significantly reducing image quality.

[0096] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0097] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0098] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1The function specified in one or more boxes.

[0099] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0100] The above specific embodiments further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A fast CT ring artifact correction method based on projected averaged images, characterized in that, Includes the following steps: The original projected images at different angles are superimposed and averaged to obtain the averaged projected image. By performing a differential operation on the projected average image, a differential image reflecting the gray-level abrupt changes in the projected average image is obtained. Calculate the filter kernel window size at different pixel locations based on the differential image. Based on the gray level of the differential image in the minimum kernel and the largest kernel A linear mapping is performed between them. When the gray value of the differential image reaches its maximum value, the minimum kernel is taken; when it reaches its minimum value, the maximum kernel is taken; and for intermediate values, the nearest smallest odd integer is taken based on the calculation result. Adaptive low-pass filtering can be performed in the spatial domain as adaptive mean filtering, adaptive median filtering, or adaptive Gaussian filtering, or in the frequency domain as adaptive ideal frequency domain low-pass filtering, adaptive Butterworth low-pass filtering, or adaptive Gaussian low-pass filtering. In the spatial domain, the formula for calculating the filter kernel window size is as follows: ； Alternatively, in the frequency domain, the formula for calculating the filter kernel window size is as follows: ； in, This indicates the size of the filter kernel window for adaptive low-pass filtering; This represents the minimum size of the filter kernel window; This represents the maximum value of the filter kernel window size; This represents the minimum value of the differential graph; Represents the maximum value of the differential image; the range of filter kernel window size. The size of the adaptive filter window should be selected based on the specific scenario and the desired correction effect. The window size should be the nearest odd number to the value calculated by the above formula. The projected average image is adaptively low-pass filtered based on the filter kernel window size to obtain a low-frequency image; Divide the average projected image by the low-frequency image to obtain the gain-corrected image; The pixel gain of the original projection images at different angles is corrected based on the gain correction image to obtain a projection sequence image with consistent gain.

2. The fast CT ring artifact correction method based on projected averaged images according to claim 1, characterized in that, The specific formula for calculating the average projected image is as follows: ； in, Represents the projected average image; Indicates the row number of the image; Indicates image column labels; Indicates the first Original projection image; This indicates the number of original projected images.

3. The fast CT ring artifact correction method based on projected averaged images according to claim 1, characterized in that, The differentiation operation uses either a first-order Sobel filter or a second-order Laplacian filter.

4. The fast CT ring artifact correction method based on projected averaged images according to claim 1, characterized in that, The calculation formula for the differential image using first-order differential operations is as follows: ； in, Represents the differential image; This represents the image gradient along the column direction; This represents the image gradient along the row direction; Indicates the row number of the image; Indicates image column labels; Alternatively, the calculation formula for the differential image using second-order differential operations is as follows: ； in, Represents the Laplace operator; This represents the average projected image.

5. The fast CT ring artifact correction method based on projection averaging image according to claim 1, characterized in that, The specific formula for calculating the low-frequency image is as follows: ； in, Represents low-frequency images; This indicates an adaptive low-pass filter; Represents the projected average image; This indicates the size of the filter kernel window for adaptive low-pass filtering; Indicates the row number of the image; Indicates the column header of the image.

6. The fast CT ring artifact correction method based on projected averaged images according to claim 1, characterized in that, The specific formula for calculating the projected image sequence is as follows: ； in, Indicates the first Aspect-projected sequence of images; Indicates the first Original projection image; Represents a gain-corrected image; Indicates the row number of the image; Indicates the column header of the image.

7. A fast CT ring artifact correction system based on projected averaged images, characterized in that, include: The averaging module is used to superimpose and average the original projected images at different angles to obtain the average projected image. The differential operation module is used to perform differential operations on the projected average image to obtain a differential image that reflects the gray-level abrupt changes in the projected average image. The size calculation module is used to calculate the size of the filter kernel window at different pixel positions based on the differential image. Based on the gray level of the differential image in the minimum kernel and the largest kernel A linear mapping is performed between them. When the gray value of the differential image reaches its maximum value, the minimum kernel is taken; when it reaches its minimum value, the maximum kernel is taken; and for intermediate values, the nearest smallest odd integer is taken based on the calculation result. Adaptive low-pass filtering can be performed in the spatial domain as adaptive mean filtering, adaptive median filtering, or adaptive Gaussian filtering, or in the frequency domain as adaptive ideal frequency domain low-pass filtering, adaptive Butterworth low-pass filtering, or adaptive Gaussian low-pass filtering. In the spatial domain, the formula for calculating the filter kernel window size is as follows: ； Alternatively, in the frequency domain, the formula for calculating the filter kernel window size is as follows: ； in, This indicates the size of the filter kernel window for adaptive low-pass filtering; This represents the minimum size of the filter kernel window; This represents the maximum value of the filter kernel window size; This represents the minimum value of the differential graph; Represents the maximum value of the differential image; the range of filter kernel window size. The size of the adaptive filter window should be selected based on the specific scenario and the desired correction effect. The window size should be the nearest odd number to the value calculated by the above formula. The low-pass filter module is used to adaptively low-pass filter the projected average image according to the size of the filter kernel window to obtain a low-frequency image; The gain correction module is used to divide the projected average image by the low-frequency image to obtain the gain-corrected image; The gain correction module is used to correct the pixel gain of the original projection image at different angles based on the gain correction image, so as to obtain a projection sequence image with consistent gain.

8. A computer terminal comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the fast CT ring artifact correction method based on projection averaging image as described in any one of claims 1-6.

Images