X-ray based method and system for detecting homogeneity of lead zirconium containing samples

Abstract

The present application relates to product uniformity detection technical field, specifically to a lead zirconium sample uniformity detection analysis method and system based on X-ray, the method comprises the following steps: screening abnormal areas based on the gray scale distribution of the pixel points in each concentric ring band in the X-ray transmission image of the lead zirconium sample; clustering all abnormal areas according to the relative gray scale difference of adjacent pixel points in the same direction of each abnormal area to obtain each target area; dividing each target area into quadrilateral regions, and obtaining the distortion correction factor of each quadrilateral region according to the morphological distribution difference between each quadrilateral region and the quadrilateral regions in its neighborhood; correcting the gray scale value of the corresponding pixel points using the distortion correction factor to obtain a composition difference map; and detecting the uniformity of the lead zirconium sample based on the composition difference map. The present application improves the accuracy of the lead zirconium sample uniformity detection result.

Description

Technical Field

[0001] This invention relates to the field of product uniformity testing technology, specifically to a method and system for uniformity testing and analysis of lead-zirconium samples based on X-rays. Background Technology

[0002] Lead-zirconium materials are widely used in key structural components in the nuclear industry, aerospace, and high-end medical devices due to their excellent high-temperature strength, corrosion resistance, and neutron absorption capacity. The uniformity of lead and zirconium composition directly determines the mechanical properties and safety of the material during use. Therefore, rapid and non-destructive uniformity testing of lead-zirconium samples is a crucial step in production quality control. In existing technologies, X-ray penetration imaging is widely used to detect material uniformity because it is sensitive to differences in internal density and atomic number. Its basic principle is to use a uniform region of a reference sample as a benchmark to correct for grayscale variations in the projected image caused by lead composition inhomogeneity in the lead-zirconium sample. This ensures that the grayscale differences at different pixel locations after correction accurately reflect the compositional differences within the sample. This method works well for samples of uniform thickness or with slow thickness variations.

[0003] Existing flat-field correction methods can only address the non-uniformity of detector response and the non-uniformity of grayscale levels caused by uneven X-ray source intensity distribution. The inherent cone-beam illumination characteristics of cone-beam X-ray imaging cause the image received by the detector to exhibit pincushion geometric distortion. The pincushion distortion caused by path length differences in the projected image is coupled with the grayscale changes caused by compositional differences, masking the true compositional differences and thus resulting in low accuracy of the uniformity detection results for lead-zirconium samples. Summary of the Invention

[0004] To address the issue of low accuracy in existing methods for detecting the homogeneity of lead-zirconium samples, the present invention aims to provide an X-ray-based method and system for detecting and analyzing the homogeneity of lead-zirconium samples. The specific technical solution adopted is as follows: In a first aspect, the present invention provides an X-ray-based method for detecting and analyzing the homogeneity of lead-zirconium-containing samples, the method comprising the following steps: Obtain X-ray transmission images of the lead-zirconium sample to be tested; Based on the grayscale distribution of pixels within each concentric ring in the X-ray transmission image, abnormal regions are screened; based on the relative grayscale difference of adjacent pixels in the same direction within each abnormal region, all abnormal regions are clustered to obtain each target region. Each target region is divided into quadrilateral regions. Based on the morphological distribution differences between each quadrilateral region and its neighboring quadrilateral regions, the distortion correction factor for each quadrilateral region is obtained. The grayscale values ​​of the corresponding pixels are corrected using the distortion correction factor to obtain a compositional difference map; the uniformity of the lead-zirconium sample is then detected based on the compositional difference map.

[0005] Preferably, the acquisition of the concentric annular bands includes: establishing a polar coordinate system with the geometric position corresponding to the center point of the stage as the origin, calculating the radial distance of each pixel in the X-ray transmission image; and dividing the X-ray transmission image into multiple concentric annular bands according to the radial distance.

[0006] Preferably, the step of filtering out abnormal regions based on the grayscale distribution of pixels within each concentric annular band in the X-ray transmission image includes: For any annular zone: Based on the grayscale difference between adjacent pixels within any annular band and the average grayscale value of all pixels within any annular band, a grayscale anomaly index for any annular band is determined. If the grayscale anomaly index is greater than a preset grayscale threshold, then the connected region formed by pixels whose grayscale value is greater than a preset deviation threshold than the difference between the grayscale value of any pixel in any annular band and the average grayscale value of all pixels in any annular band is considered an abnormal region.

[0007] Preferably, the step of clustering all abnormal regions to obtain each target region based on the relative grayscale difference of adjacent pixels in the same direction within each abnormal region includes: For any abnormal region, within the abnormal region, the line segments in each direction with the center of the circle corresponding to all concentric rings as the endpoints of the line segments are recorded as feature line segments in each direction; the edge blurring degree in each direction is obtained based on the relative gray level difference of adjacent pixels on the feature line segments in the same direction; and the region distortion parameters of the abnormal region are obtained by combining the edge blurring degree in all directions of the abnormal region. Based on the regional distortion parameters and the center point location of the abnormal regions, a hierarchical clustering algorithm is used to cluster all abnormal regions to obtain each target region.

[0008] Preferably, the step of obtaining the edge blurring degree in each direction based on the relative grayscale difference of adjacent pixels on the feature line segment in the same direction includes: For a feature line segment in any direction, the pixels on the feature line segment in any direction are arranged in order of increasing distance from the pixel on the feature line segment to the center of all concentric rings to obtain a pixel sequence. The ratio between the grayscale difference between two adjacent pixels in the pixel sequence and the grayscale value of the preceding pixel in the two adjacent pixels is taken as the relative grayscale difference between the two adjacent pixels. The average value of the relative grayscale difference between all two adjacent pixels in the pixel sequence is determined as the edge blur level in the corresponding direction.

[0009] Preferably, the step of obtaining the distortion correction factor for each quadrilateral region based on the morphological distribution difference between each quadrilateral region and its neighboring quadrilateral regions includes: For each quadrilateral region, the lengths of the two diagonals of the quadrilateral region are obtained; the ratio of the maximum to the minimum length of the two diagonals is used as the deformation factor of each quadrilateral region, and the deformation factor is used to characterize the morphological distribution of the quadrilateral region. The distortion correction factor of the candidate quadrilateral region is obtained based on the difference between the deformation factor of the candidate quadrilateral region and the deformation factor of its neighboring quadrilateral regions. The candidate quadrilateral region can be any quadrilateral region.

[0010] Preferably, obtaining the distortion correction factor of the candidate quadrilateral region based on the difference between the deformation factor of the candidate quadrilateral region and the deformation factor of its neighboring quadrilateral regions includes: Calculate the average deformation factor of all quadrilateral regions within the neighborhood of the candidate quadrilateral region; Obtain the first difference between the deformation factor of the candidate quadrilateral region and the deformation factor of each quadrilateral region in its neighborhood, and use the ratio of the first difference to the average value of the deformation factor as the deformation index corresponding to each quadrilateral region in the neighborhood of the candidate quadrilateral region. The average value of the deformation index corresponding to all quadrilateral regions in the neighborhood of the candidate quadrilateral region is summed with a constant 1 as the distortion correction factor for the candidate quadrilateral region.

[0011] Preferably, the step of correcting the grayscale values ​​of the corresponding pixels using the distortion correction factor to obtain a component difference map includes: Calculate the product of the gray value of each pixel in each quadrilateral region and the distortion correction factor of the quadrilateral region it belongs to, and round up the result. Take the minimum value between the rounded result and the constant 255 as the corrected gray value of the pixel. A component difference map is generated based on the corrected grayscale values.

[0012] Preferably, the step of dividing each target region into quadrilateral regions includes: A variable density adaptive mesh is used to divide each target region into meshes, and each mesh region is treated as a quadrilateral region.

[0013] Secondly, the present invention provides an X-ray-based homogeneity detection and analysis system for lead-zirconium samples, comprising a memory and a processor, wherein the processor executes a computer program stored in the memory to implement the aforementioned X-ray-based homogeneity detection and analysis method for lead-zirconium samples.

[0014] The present invention has at least the following beneficial effects: This invention first filters out anomalous regions based on the grayscale distribution of pixels within concentric annular bands in the X-ray transmission image of the lead-zirconium sample to be tested. Then, it clusters adjacent pixels in the same direction within the anomalous regions according to their relative grayscale differences, thereby obtaining each target region. This process fully considers the radial distortion distribution caused by the divergence characteristics of X-ray cone-beam imaging, effectively distinguishing between grayscale changes caused by differences in true composition and artifact interference caused by geometric distortion, thus improving the accuracy and specificity of anomalous region identification. Furthermore, each target region is divided into quadrilateral regions, and a distortion correction factor is calculated for each quadrilateral region based on the morphological distribution differences between each quadrilateral region and its neighboring quadrilateral regions. This distortion correction factor reflects the degree of distortion in a local area, avoiding overcorrection or undercorrection in traditional global correction methods. This method addresses the problem of fine-grained and localized processing of image distortion. It then uses a distortion correction factor to correct the grayscale values ​​of the corresponding pixels, obtaining a compositional difference map that reflects the true differences in components. Based on this map, homogeneity detection of lead-zirconium samples is performed. This method compensates for the influence of distortion on grayscale distribution, ensuring that the corrected image accurately and reliably reflects the distribution of lead, zirconium, and other components in the sample. It solves the detection error problem caused by the coupling of cone-beam X-ray geometric distortion and compositional differences in existing methods, thus improving the accuracy of homogeneity detection results for lead-zirconium samples. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art 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.

[0016] Figure 1 This is a flowchart of an X-ray-based method for detecting and analyzing the uniformity of lead-zirconium samples, provided as an embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve the intended purpose, the following detailed description of the X-ray-based homogeneity detection and analysis method and system for lead-zirconium samples proposed in accordance with the present invention is provided in conjunction with the accompanying drawings and preferred embodiments.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] The following description, in conjunction with the accompanying drawings, details the specific scheme of the X-ray-based homogeneity detection and analysis method and system for lead-zirconium samples provided by this invention.

[0020] Example of an X-ray-based method for detecting and analyzing the homogeneity of lead-zirconium samples: This embodiment proposes an X-ray-based method for detecting and analyzing the uniformity of lead-zirconium-containing samples, such as... Figure 1 As shown, the X-ray-based homogeneity detection and analysis method for lead-zirconium samples in this embodiment includes the following steps: Step S1: Obtain an X-ray transmission image of the lead-zirconium sample to be tested.

[0021] When performing homogeneity testing on lead-zirconium samples, the first step is to acquire X-ray transmission images of the sample using a digital X-ray transmission imaging system. This system includes a cone-beam X-ray source, a stage, and a flat panel detector. The specific acquisition process for the X-ray transmission images of the lead-zirconium sample is as follows: the X-ray tube is positioned 150 cm directly above the stage, and the flat panel detector is placed 30 cm below the stage, with the center of the detector coinciding with the emission center of the X-ray source. The lead-zirconium sample is placed on the stage, and the equipment is activated to acquire the X-ray intensity distribution penetrating the sample, obtaining the original projection image. Then, the original projection image is processed by grayscale conversion and Gaussian filtering for noise reduction. This processed image is recorded as the X-ray transmission image of the lead-zirconium sample.

[0022] Step S2: Based on the grayscale distribution of pixels within each concentric ring in the X-ray transmission image, abnormal regions are screened; based on the relative grayscale difference of adjacent pixels in the same direction within each abnormal region, all abnormal regions are clustered to obtain each target region.

[0023] Given that lead in lead-zirconium samples has extremely strong absorption capacity for X-rays, even a small change in the lead content in the sample can cause a significant change in the X-ray intensity signal. However, due to the characteristics of the X-ray cone beam distribution, the image received by the detector exhibits pincushion distortion. This distortion has radial symmetry: the central region of the image is close to the optical axis and has a small degree of distortion; the edge regions (especially the four vertices) have larger distortions, accompanied by gray-level stretching / compression. This can mask the true defect of a sparse center and normal edges, especially since the gray-level gradient at the edges is amplified, affecting the accurate judgment of the lead content distribution.

[0024] A polar coordinate system is established with the geometric position corresponding to the center point of the stage as the origin, and the radial distance of each pixel in the X-ray transmission image is calculated. The X-ray transmission image is divided into multiple concentric annular bands according to the radial distance. In this embodiment, the bandwidth of each annular band is set to 5 pixels.

[0025] After obtaining multiple concentric annular bands, each annular band is analyzed separately. Considering that the greater the grayscale difference between adjacent pixels within an annular band, the more likely the annular band is to have distortion, this embodiment will evaluate the grayscale anomaly of each annular band based on the grayscale difference between adjacent pixels within a single annular band region, and obtain a grayscale anomaly index.

[0026] Specifically, for any annular band: First, the pixels within any annular band are sorted clockwise according to the polar angle of the polar coordinates to obtain a one-dimensional pixel sequence within the annular band. Then, based on the grayscale difference between adjacent pixels within the annular band and the average grayscale value of all pixels within the annular band, the grayscale anomaly index of the annular band is determined.

[0027] As a specific implementation method, a specific calculation formula for the grayscale anomaly index is given. The grayscale anomaly index of the m-th annular band can be expressed as: in, This represents the grayscale anomaly index of the m-th annular band. This represents the number of pixels within the m-th annular band. This represents the grayscale value of the i-th pixel within the m-th annular band. This represents the grayscale value of the (i+1)th pixel within the m-th annular band. This represents the average grayscale value of all pixels within the m-th annular band. This indicates the absolute value sign.

[0028] It represents the grayscale difference between the i-th pixel and the (i+1)-th pixel within the m-th annular band. The larger the value, the greater the grayscale difference between the two pixels. This characterizes the average grayscale difference among all adjacent pixels within the m-th annular band. The value represents the average gray level difference within the m-th annular band relative to the average gray level of all pixels within that annular band. The larger this value is, the more abnormal the gray level distribution of pixels within the m-th annular band is, i.e., the larger the gray level anomaly index of the m-th annular band is.

[0029] It should be noted that i and i+1 in the grayscale anomaly index are both indices of pixels in the one-dimensional pixel sequence obtained above. Specifically, if the denominator of the grayscale anomaly index calculation formula is 0, then the grayscale anomaly index of the corresponding annular band is directly set to 0.

[0030] Using the above method, the grayscale anomaly index for each annular band can be obtained. The connected regions formed by pixels whose grayscale anomaly index exceeds a preset grayscale threshold and whose difference between the grayscale value of each pixel within the annular band and the average grayscale value of all pixels within that annular band exceeds a preset deviation threshold are considered anomaly regions. The preset grayscale threshold is obtained by calibrating a known defect-free standard lead-zirconium sample. Specifically, an image of the standard lead-zirconium sample is acquired, and the grayscale anomaly index for each concentric annular band is calculated. The mean of all grayscale anomaly indices is added to three times the standard deviation to obtain the preset grayscale threshold. For example, the preset grayscale threshold is 0.45.

[0031] After identifying the abnormal regions, grayscale change analysis is performed on each abnormal region to evaluate the degree of ambiguity of the abnormal regions.

[0032] For any abnormal region, multiple directions are preset, for example: the direction can be... direction, direction, direction, direction, direction, direction, direction and Within the abnormal region, line segments in each direction, with the centers of all concentric rings as their endpoints, are denoted as feature line segments in each direction. The degree of edge blurring in each direction is determined based on the relative grayscale difference between adjacent pixels on the same feature line segment. Specifically, for any feature line segment in any direction, pixels on that feature line segment are arranged in ascending order of distance from the center of all concentric rings to obtain a pixel sequence. The ratio of the grayscale difference between two adjacent pixels in the pixel sequence to the grayscale value of the preceding pixel in those two adjacent pixels is taken as the relative grayscale difference between the two adjacent pixels. The average of the relative grayscale differences between all two adjacent pixels in the pixel sequence is determined as the degree of edge blurring in the corresponding direction.

[0033] As a specific implementation method, a specific formula for calculating the degree of edge blurring is given. The degree of edge blurring in the j-th direction within the z-th anomaly region can be expressed as: in, This indicates the degree of edge blurring in the j-th direction within the z-th anomalous region. This represents the number of pixels in the pixel sequence in the j-th direction within the z-th abnormal region. Let represent the grayscale value of the u-th pixel in the pixel sequence in the j-th direction within the z-th abnormal region. Let represent the grayscale value of the (u+1)th pixel in the pixel sequence in the j-th direction within the z-th abnormal region. This indicates the absolute value sign.

[0034] This value represents the grayscale difference between the u-th pixel and the (u+1)-th pixel in the pixel sequence in the j-th direction within the z-th abnormal region. The larger this value is, the greater the grayscale difference between the two pixels. The relative grayscale difference between the u-th and u+1-th pixels in the pixel sequence in the j-th direction within the z-th anomalous region is represented. The larger the average relative grayscale difference between two adjacent pixels in the pixel sequence in the j-th direction within the z-th anomalous region, the more severe the grayscale transition of the feature line segment in that direction, and the more likely it is to correspond to a false edge caused by a real component mutation or distortion. That is, the greater the degree of edge blurring in the j-th direction within the z-th anomalous region.

[0035] Specifically, if the formula for calculating the degree of edge blurring includes... If it is 0, then directly set The value is 0. It should be noted that if the number of pixels on any feature line segment extracted based on the preset direction is less than or equal to 1, the edge blur degree in the corresponding direction will not be calculated and will not be included; and when calculating the average edge blur degree in multiple directions of the abnormal region in the subsequent calculation, it will be removed from the total number of directions calculated.

[0036] The above method can be used to obtain the edge blurring degree in each direction within the z-th anomalous region. Furthermore, by combining the edge blurring degree in all directions of the anomalous region, the regional distortion parameter of the anomalous region can be obtained.

[0037] In one specific implementation, the average value of the edge blurring degree in all directions within the z-th anomalous region is used as the region distortion parameter of that anomalous region.

[0038] Using the above method, the regional distortion parameters of each anomalous region can be obtained. Next, based on the regional distortion parameters and the center point location of the anomalous region, a hierarchical clustering algorithm is used to cluster all anomalous regions. Specifically, the center point coordinates and regional distortion parameters of each anomalous region are normalized respectively; the normalized center point coordinates and regional distortion parameters are concatenated into a three-dimensional feature vector; the Euclidean distance between the three-dimensional feature vectors of each pair of anomalous regions is calculated, and the Euclidean distance is used as a similarity measure. The hierarchical clustering algorithm is then used to cluster all anomalous regions, and each cluster obtained by clustering is taken as a target region, thus obtaining multiple target regions. In this process, the normalization of the center point coordinates and regional distortion parameters uses a maximum-minimum value normalization method. The maximum and minimum values ​​of the center point coordinates are obtained by traversing all abnormal regions and extracting the maximum and minimum values ​​of the center point's x-coordinate and y-coordinate, respectively. Similarly, the maximum and minimum values ​​of the regional distortion parameters are obtained by traversing all abnormal regions and extracting the maximum and minimum values ​​of all regional distortion parameters. Specifically, when performing maximum-minimum value normalization on the center point coordinates and regional distortion parameters, if the obtained maximum and minimum values ​​are equal (i.e., the range is zero), the corresponding normalized feature value is directly set to a constant of 0.5. The hierarchical clustering algorithm is an existing clustering algorithm and will not be elaborated further here.

[0039] Step S3: Divide each target region into quadrilateral regions. Based on the morphological distribution differences between each quadrilateral region and its neighboring quadrilateral regions, obtain the distortion correction factor for each quadrilateral region.

[0040] Cone-beam X-ray imaging distortion exhibits a spatial distribution characteristic of weak distortion at the center and strong distortion at the edges, resulting in strong radial symmetry in the acquired X-ray transmission images. The central rays are perpendicularly incident, resulting in the shortest transmission path and minimal distortion; while the edge rays are obliquely incident, leading to significant pincushion deformation and scattering errors. Therefore, this embodiment abandons fixed equidistant grids and adopts variable density adaptive grids based on pixel gray-level gradients and distortion levels. By analyzing the distortion levels of different regions, it avoids directly mixing the strong distortion at the edges with the true signal at the center for calculation, which could lead to overcorrection or undercorrection.

[0041] Specifically, a variable-density adaptive mesh is used to divide each target region into grids, with each grid region treated as a quadrilateral region. The variable-density adaptive mesh is an existing technology and will not be elaborated upon further here.

[0042] Ideally, if X-rays are incident perpendicularly and the sample surface is parallel to the detector plane, the regular grid structure in the image should present a uniform, similarly shaped orthogonal grid. However, due to the divergence characteristics of cone-beam X-rays, the angle of incidence of X-rays changes more and more severely from the center of the image to the edge, resulting in pincushion distortion in the quadrilateral region. Therefore, the deformation of the quadrilateral region will be analyzed next to construct a distortion correction factor.

[0043] Specifically, for each quadrilateral region, the lengths of the two diagonals of the quadrilateral region are obtained; the ratio of the maximum to the minimum length of the two diagonals is used as the deformation factor of each quadrilateral region. The deformation factor is used to characterize the morphological distribution of the quadrilateral region. The larger the value, the more severe the deformation and the greater the possibility of pincushion distortion in the region.

[0044] The following embodiment uses a quadrilateral region as an example for explanation. Other quadrilateral regions can be processed using the method provided in this embodiment.

[0045] Any quadrilateral region is denoted as a candidate quadrilateral region.

[0046] The distortion correction factor for the candidate quadrilateral region is obtained based on the difference between the deformation factor of the candidate quadrilateral region and the deformation factors of its neighboring quadrilateral regions. Specifically, the average deformation factor of all quadrilateral regions within the neighborhood of the candidate quadrilateral region is calculated; the absolute value of the difference between the deformation factor of the candidate quadrilateral region and the deformation factor of each quadrilateral region within its neighborhood is calculated, and this absolute value is denoted as the first difference; the ratio of the first difference to the average deformation factor of all quadrilateral regions within the neighborhood of the candidate quadrilateral region is taken as the deformation index corresponding to each quadrilateral region within the neighborhood of the candidate quadrilateral region; the sum of the average deformation index corresponding to all quadrilateral regions within the neighborhood of the candidate quadrilateral region and a constant 1 is taken as the distortion correction factor for the candidate quadrilateral region.

[0047] As a specific implementation method, a specific formula for calculating the distortion correction factor is given. The distortion correction factor for the candidate quadrilateral region can be expressed as: in, This represents the distortion correction factor for the candidate quadrilateral region. This represents the number of quadrilateral regions within the neighborhood of a candidate quadrilateral region. Let v represent the deformation factor of the v-th quadrilateral region within the neighborhood of the candidate quadrilateral region. This represents the deformation factor of the candidate quadrilateral region. Indicates the absolute value sign. This represents the average deformation factor of all quadrilateral regions within the neighborhood of the candidate quadrilateral region.

[0048] In this embodiment, the size of the neighborhood is set to eight neighborhoods. In specific applications, the implementer can set it according to the specific situation. It should be noted that if the divided candidate quadrilateral region is an isolated grid structure and no other connecting blocks are found, that is, the number of quadrilateral regions in its neighborhood is 0, no local relative neighborhood distortion correction is performed on the candidate quadrilateral region, and the distortion correction factor of the candidate quadrilateral region is directly set to 1.

[0049] This represents the first difference between the candidate quadrilateral region and the v-th quadrilateral region in its neighborhood. By calculating the difference between the candidate quadrilateral region and its neighborhood quadrilateral regions, the distortion trend along the current direction is reflected. The larger the difference, the more obvious the deformation gradient along that direction, and the greater the possibility that the pixels in the region need to be corrected, that is, the larger the distortion correction factor of the candidate quadrilateral region.

[0050] Using the above method, the distortion correction factor for each quadrilateral region can be obtained.

[0051] Step S4: Correct the grayscale value of the corresponding pixel using the distortion correction factor to obtain a composition difference map; perform uniformity detection on the lead-zirconium sample based on the composition difference map.

[0052] In step S3 of this embodiment, the distortion correction factor for each quadrilateral region is determined. Next, the distortion correction factor will be used to correct the gray value of the pixels in the quadrilateral region, thereby achieving accurate detection of the uniformity of the lead-zirconium sample.

[0053] Specifically, the product of the gray value of each pixel within each quadrilateral region and the distortion correction factor of that quadrilateral region is calculated and rounded up. The minimum value between this rounded up result and the constant 255 is taken as the corrected gray value of the corresponding pixel. This method can correct the gray values ​​of pixels within quadrilateral regions. For pixels outside quadrilateral regions, their gray values ​​are not corrected; their original gray values ​​are directly used as the corrected gray values. The corrected gray values ​​of the pixels are then used to replace the original gray values ​​of the pixels in the image to obtain a composition difference map.

[0054] After obtaining the composition difference map, the composition difference map is denoised. Specifically, nonlocal mean filtering or BM3D algorithm can be used to process the composition difference map to remove residual noise and obtain the denoised image.

[0055] Then, an adaptive threshold segmentation algorithm is used to segment each target region in the denoised image, extracting the initial abnormal regions within the target regions. Morphological opening and closing operations are performed on the segmentation results to remove isolated noise points, connect fracture boundaries, delineate non-uniform defect boundaries, distinguish between normal homogeneous regions and abnormal regions, and mark gray-level abnormal regions in the image. The adaptive threshold segmentation algorithm can be the Otsu multi-threshold method, which is an existing technology and will not be elaborated further here.

[0056] Furthermore, the geometric features and grayscale features of each grayscale anomaly region are extracted. The geometric features include area, perimeter, roundness, aspect ratio, etc., and the grayscale features include average grayscale, grayscale variance, contrast, etc. The geometric features and grayscale features are input into the neural network model, and the uniformity detection results of the lead-zirconium sample are output. The neural network model can be a BP neural network. The network structure and training process of the neural network model are existing technologies, and will not be described in detail in this embodiment.

[0057] Thus, the method provided in this embodiment has been used to detect the uniformity of lead-zirconium samples.

[0058] This embodiment first filters out anomalous regions based on the grayscale distribution of pixels within concentric annular bands in the X-ray transmission image of the lead-zirconium sample to be detected. Then, it clusters adjacent pixels in the same direction within the anomalous region according to their relative grayscale differences, thereby obtaining each target region. This process fully considers the radial distortion distribution caused by the divergence characteristics of X-ray cone-beam imaging, effectively distinguishing between grayscale changes caused by differences in true composition and artifact interference caused by geometric distortion, thus improving the accuracy and specificity of anomalous region identification. Furthermore, each target region is divided into quadrilateral regions, and a distortion correction factor is calculated for each quadrilateral region based on the morphological distribution differences between each quadrilateral region and its neighboring quadrilateral regions. The distortion correction factor reflects the degree of distortion in a local area, avoiding overcorrection or undercorrection in traditional global correction methods. This method addresses the problem of fine-grained and localized processing of image distortion. It then uses a distortion correction factor to correct the grayscale values ​​of the corresponding pixels, obtaining a compositional difference map that reflects the true differences in components. Based on this map, homogeneity detection of lead-zirconium samples is performed. This method compensates for the influence of distortion on grayscale distribution, ensuring that the corrected image accurately and reliably reflects the distribution of lead, zirconium, and other components in the sample. It solves the detection error problem caused by the coupling of cone-beam X-ray geometric distortion and compositional differences in existing methods, thus improving the accuracy of homogeneity detection results for lead-zirconium samples.

[0059] Example of an X-ray-based homogeneity detection and analysis system for lead-zirconium samples: The X-ray-based homogeneity detection and analysis system for lead-zirconium samples in this embodiment includes a memory and a processor. The processor executes the computer program stored in the memory to implement the above-described X-ray-based homogeneity detection and analysis method for lead-zirconium samples.

[0060] Since the X-ray-based method for detecting and analyzing the uniformity of lead-zirconium samples has already been described in the embodiments of the X-ray-based method for detecting and analyzing the uniformity of lead-zirconium samples, this embodiment will not repeat the description of the X-ray-based method for detecting and analyzing the uniformity of lead-zirconium samples.

[0061] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for detecting and analyzing the homogeneity of lead-zirconium-containing samples based on X-rays, characterized in that, The method includes the following steps: Obtain X-ray transmission images of the lead-zirconium sample to be tested; Based on the grayscale distribution of pixels within each concentric ring in the X-ray transmission image, abnormal regions are screened; based on the relative grayscale difference of adjacent pixels in the same direction within each abnormal region, all abnormal regions are clustered to obtain each target region. Each target region is divided into quadrilateral regions. Based on the morphological distribution differences between each quadrilateral region and its neighboring quadrilateral regions, the distortion correction factor for each quadrilateral region is obtained. The grayscale values ​​of the corresponding pixels are corrected using the distortion correction factor to obtain a compositional difference map; the uniformity of the lead-zirconium sample is then detected based on the compositional difference map.

2. The method for detecting and analyzing the homogeneity of lead-zirconium-containing samples based on X-rays according to claim 1, characterized in that, The acquisition of the concentric annular bands includes: establishing a polar coordinate system with the geometric position corresponding to the center point of the stage as the origin, calculating the radial distance of each pixel in the X-ray transmission image; and dividing the X-ray transmission image into multiple concentric annular bands according to the radial distance.

3. The method for detecting and analyzing the homogeneity of lead-zirconium-containing samples based on X-rays according to claim 1, characterized in that, The process of filtering out abnormal regions based on the grayscale distribution of pixels within each concentric annular band in the X-ray transmission image includes: For any annular zone: Based on the grayscale difference between adjacent pixels within any annular band and the average grayscale value of all pixels within any annular band, a grayscale anomaly index for any annular band is determined. If the grayscale anomaly index is greater than a preset grayscale threshold, then the connected region formed by pixels whose grayscale value is greater than a preset deviation threshold than the difference between the grayscale value of any pixel in any annular band and the average grayscale value of all pixels in any annular band is considered an abnormal region.

4. The method for detecting and analyzing the homogeneity of lead-zirconium-containing samples based on X-rays according to claim 2, characterized in that, The step of clustering all abnormal regions to obtain each target region based on the relative grayscale difference of adjacent pixels in the same direction within each abnormal region includes: For any abnormal region, within the abnormal region, the line segments in each direction with the center of the circle corresponding to all concentric rings as the endpoints of the line segments are recorded as feature line segments in each direction; the edge blurring degree in each direction is obtained based on the relative gray level difference of adjacent pixels on the feature line segments in the same direction; and the region distortion parameters of the abnormal region are obtained by combining the edge blurring degree in all directions of the abnormal region. Based on the regional distortion parameters and the center point location of the abnormal regions, a hierarchical clustering algorithm is used to cluster all abnormal regions to obtain each target region.

5. The method for detecting and analyzing the homogeneity of lead-zirconium-containing samples based on X-rays according to claim 4, characterized in that, The step of obtaining the edge blurring degree in each direction based on the relative grayscale difference of adjacent pixels on feature line segments in the same direction includes: For a feature line segment in any direction, the pixels on the feature line segment in any direction are arranged in order of increasing distance from the pixel on the feature line segment to the center of all concentric rings to obtain a pixel sequence. The ratio between the grayscale difference between two adjacent pixels in the pixel sequence and the grayscale value of the preceding pixel in the two adjacent pixels is taken as the relative grayscale difference between the two adjacent pixels. The average value of the relative grayscale difference between all two adjacent pixels in the pixel sequence is determined as the edge blur level in the corresponding direction.

6. The method for detecting and analyzing the homogeneity of lead-zirconium-containing samples based on X-rays according to claim 1, characterized in that, The distortion correction factor for each quadrilateral region is obtained based on the morphological distribution differences between each quadrilateral region and its neighboring quadrilateral regions, including: For each quadrilateral region, the lengths of the two diagonals of the quadrilateral region are obtained; the ratio of the maximum to the minimum length of the two diagonals is used as the deformation factor of each quadrilateral region, and the deformation factor is used to characterize the morphological distribution of the quadrilateral region. The distortion correction factor of the candidate quadrilateral region is obtained based on the difference between the deformation factor of the candidate quadrilateral region and the deformation factor of its neighboring quadrilateral regions. The candidate quadrilateral region can be any quadrilateral region.

7. The method for detecting and analyzing the homogeneity of lead-zirconium-containing samples based on X-rays according to claim 6, characterized in that, The step of obtaining the distortion correction factor for a candidate quadrilateral region based on the difference between the distortion factor of the candidate quadrilateral region and the distortion factors of its neighboring quadrilateral regions includes: Calculate the average deformation factor of all quadrilateral regions within the neighborhood of the candidate quadrilateral region; Obtain the first difference between the deformation factor of the candidate quadrilateral region and the deformation factor of each quadrilateral region in its neighborhood, and use the ratio of the first difference to the average value of the deformation factor as the deformation index corresponding to each quadrilateral region in the neighborhood of the candidate quadrilateral region. The average value of the deformation index corresponding to all quadrilateral regions in the neighborhood of the candidate quadrilateral region is summed with a constant 1 as the distortion correction factor for the candidate quadrilateral region.

8. The method for detecting and analyzing the homogeneity of lead-zirconium-containing samples based on X-rays according to claim 1, characterized in that, The step of correcting the grayscale values ​​of the corresponding pixels using the distortion correction factor to obtain a component difference map includes: Calculate the product of the gray value of each pixel in each quadrilateral region and the distortion correction factor of the quadrilateral region it belongs to, and round up the result. Take the minimum value between the rounded result and the constant 255 as the corrected gray value of the pixel. A component difference map is generated based on the corrected grayscale values.

9. The method for detecting and analyzing the homogeneity of lead-zirconium-containing samples based on X-rays according to claim 1, characterized in that, The process of dividing each target region into quadrilateral regions includes: A variable density adaptive mesh is used to divide each target region into meshes, and each mesh region is treated as a quadrilateral region.

10. An X-ray-based homogeneity detection and analysis system for lead-zirconium samples, comprising a memory and a processor, characterized in that, The processor executes the computer program stored in the memory to implement the X-ray-based homogeneity detection and analysis method for lead-zirconium samples as described in claim 1.

Images