Method and system for evaluating doping uniformity of particles in material

Abstract

The invention discloses a method and system for evaluating the doping uniformity of particles in a material, and the method comprises the steps: 1) obtaining a material doping image, carrying out the preprocessing, and recognizing the mass center of doped particles; 2) processing the mass center of the doped particles by using a Bowyer-Watson algorithm, and constructing a Voronoi mosaic diagram; and 3) calculating an area and perimeter variation coefficient of each Voronoi region in the Voronoi mosaic graph, and calculating a unified measurement standard F1 measurement of the area and perimeter variation coefficient, thereby evaluating the particle doping uniformity. The system comprises a material sample collection device, a visual sample collection module and a visual processing module, according to the method, the distribution condition of the doped particles in the sample can be quantitatively and objectively evaluated.

Description

Technical Field

[0001] The present invention relates to the field of materials science, and particularly to a method and system for evaluating the uniformity of particle doping in materials. Background Art

[0002] In materials science, the uniformity of doping is a very important requirement. The following are several key points for explanation:

[0003] 1. Performance consistency: Uniformly distributed doping can ensure that the material has consistent physical properties in different regions, which is crucial for manufacturing devices with expected performance. For example, in semiconductor devices, uniform doping can ensure the consistency of carrier concentration, thus affecting the electrical properties of the device.

[0004] 2. Avoidance of defects: Non-uniform doping may lead to local stress concentration or defects in the material, which may reduce the mechanical properties of the material or cause device failure.

[0005] 3. Improvement of efficiency: In luminescent materials, uniform doping can improve the luminescence efficiency and change the emission wavelength, which is very important for manufacturing devices such as light-emitting diodes (LEDs).

[0006] 4. Crystal quality: Uniform doping helps to maintain the good crystalline quality of the crystal. For example, X-ray fluorescence spectroscopy shows that the Al element is uniformly distributed on the crystal cross-section, indicating that the crystal has good crystalline quality and doping uniformity.

[0007] 5. Avoidance of phase separation: In some cases, non-uniform doping may lead to phase separation in the material, which will affect the physical properties of the material. For example, some studies have shown that the electromagnetic inhomogeneity and electronic phase separation in highly doped oxides may be related to the local structure or atomic distribution.

[0008] 6. Process control: Modern semiconductor manufacturing technologies, such as molecular beam epitaxy, can precisely control the doping concentration and layer thickness, enabling the accurate comparison of theoretical and experimental results, which promotes the progress of theoretical and applied research.

[0009] Uniformly distributed doping is crucial for ensuring the consistency and reliability of material properties and can be achieved by precisely controlling the doping process during manufacturing. Non-uniform doping may have the following effects on the material properties:

[0010] 1. Affect the electrical properties of the material: Non-uniform doping may lead to non-uniform current distribution, especially during avalanche breakdown, which may cause local overheating and device damage. In semiconductor materials, non-uniform doping affects the drift and diffusion of carriers, thus affecting the conductivity and Hall effect of the material.

[0011] 2. Altering the microstructure of materials: Non-uniform doping may lead to local stress concentration or defects in the material, affecting the microstructure and phase distribution of the material. For example, in composite materials, non-uniformly distributed reinforcements will result in larger matrix grain sizes, and the grain fracture mechanism is quasi-cleavage fracture.

[0012] 3. Affecting the optical properties of materials: Non-uniform doping will affect the optical anisotropy of the material, changing its refractive index, extinction coefficient, absorption coefficient, and reflection spectrum. For example, in chromium-doped zinc selenide, the non-uniform distribution of Cr2+ ions will cause the crystal to change from isotropic to anisotropic, affecting its optical properties.

[0013] 4. Reducing the uniformity of materials: Non-uniform doping will reduce the uniformity of the material, affecting the overall performance of the material. For example, in high-entropy alloys, the non-uniformity of element distribution may manifest as concentration waves or short-range ordered structures, which breaks the understanding that high-entropy alloys are ideal solid solutions and has an important impact on the mechanical properties of the material.

[0014] 5. Affecting the electrochemical properties of materials: In electrochemical materials, non-uniform doping will affect the charge transfer impedance and conductivity, thus affecting the electrochemical properties of the material. For example, the preparation and electrochemical performance study of B-doped MnO2 show that doping can improve the migration speed of Li+ ions inside the electrode, thereby improving the electrochemical properties of the material.

[0015] 6. Affecting the photocatalytic activity of materials: In photocatalytic materials, non-uniform doping can greatly improve the activity. For example, the Mn non-uniformly doped TiO2 thin film has the highest photocatalytic activity at the optimal doping concentration.

[0016] In summary, non-uniform doping has significant effects on the electrical, optical, mechanical, and electrochemical properties of materials.

[0017] Ductile iron, as a special spheroidized cast iron material, is renowned for its excellent mechanical properties, including high strength and toughness, outstanding wear resistance, and excellent shock absorption characteristics. During the production of ductile iron, by adding spheroidizing agents (such as magnesium or rare earth elements) to the molten iron, the carbonaceous matter in the cast iron exists in the form of spherical graphite rather than the traditional flaky graphite. This unique spherical graphite structure not only enhances the impact toughness of ductile iron but also improves its shock absorption performance, making it an ideal material for manufacturing automotive parts, pipes, valves, and other products. Due to its excellent comprehensive performance, ductile iron occupies a crucial position in the field of industrial manufacturing. However, the performance of ductile iron depends to a large extent on the uniform distribution of spherical graphite. If the graphite distribution is uneven, it may lead to problems such as cracks and stress concentration inside the material, which will seriously affect the overall mechanical properties and toughness of the material and may even cause brittle fracture of the material. At present, China has not formulated a unified standard to regulate the distribution of spherical graphite in ductile iron. Traditional detection methods mainly rely on laboratory sampling, observing the cast iron samples through a microscope, and manually judging the distribution of graphite by the naked eye. This method not only has subjective errors but also is time-consuming, inefficient, costly, and lacks consistency and repeatability.

[0018] Therefore, it is particularly urgent to develop an automated evaluation method and system for spatially statistical particle distribution. This method can objectively and accurately grasp the distribution state of particles, significantly improving the doping efficiency of materials and product quality. Summary of the Invention

[0019] The object of the present invention is to provide a method for evaluating the uniformity of particle doping in a material, including the following steps:

[0020] 1) Obtain the material doping image and perform preprocessing to identify the centroid of the doping particles;

[0021] 2) Use the Bowyer-Watson algorithm to process the centroids of the doping particles and construct a Voronoi tessellation;

[0022] 3) Calculate the area and perimeter coefficient of variation of each Voronoi region in the Voronoi tessellation, and calculate the unified measure standard F1 metric of the area and perimeter coefficient of variation to evaluate the uniformity of particle doping.

[0023] Further, in step 1), the steps of preprocessing include:

[0024] 1.1) Convert the material doping image into a grayscale image;

[0025] 1.2) Perform binarization processing on the grayscale image to obtain a binarized image;

[0026] 1.3) Perform Canny edge detection on the binary image to identify the edges of the material particles;

[0027] 1.4) Perform multiple topological closing operations on the edges of the material particles to form multiple contours;

[0028] 1.5) Calculate the centroid of each contour as the centroid of the doped particles.

[0029] Further, in step 1.5), the x c coordinate and y c coordinate of the centroid of the contour are as follows:

[0030]

[0031] where, ∫∫ R represents the double integral over the region R, and dA is the area element; ∫∫ R xρ(x, y)dA, ∫∫ R yρ(x, y)dA are the first-order moments in the x and y directions; ∫∫ R ρ(x, y)dA is the area of the contour.

[0032] Further, in step 2), the steps of constructing the Voronoi tessellation include: using all the centroids of the doped particles as the generating points, and dividing multiple Voronoi regions according to the distances between these points, with each region enclosing a generating point, thereby constructing the Voronoi tessellation.

[0033] Further, the Voronoi region V(s i ) corresponding to the i-th generating point s i is as follows:

[0034]

[0035] where, ‖P - s i ‖ represents the Euclidean distance from point P to the generating point s i , and P is the set of generating points.

[0036] Further, the Voronoi tessellation V(S) is as follows:

[0037]

[0038] The Voronoi edge E(s i , s j ) is as follows:

[0039]

[0040] where, ‖P - s i‖ represents the Euclidean distance from point P to the generating point s i .

[0041] The Voronoi vertex V(s i , s j , s k ) is as follows:

[0042]

[0043] Furthermore, in step 3), the steps of calculating the area and perimeter variation coefficients of each Voronoi region in the Voronoi tessellation include:

[0044] 3.1) Calculate the area and perimeter of each Voronoi region;

[0045] Among them, the area and perimeter distributions of the i-th Voronoi region are as follows:

[0046]

[0047] In the formula, A(V(s i )) and P(V(s i )) are the area and perimeter distributions; (x i , y i ), (x i+1 , y i+1 ) are the coordinates;

[0048] 3.2) Calculate the mean and variance of the areas and perimeters of all Voronoi regions, that is:

[0049]

[0050] Among them, n is the number of generating points, that is, the number of doped particles identified in the image sample of material doping; Var(A i ) is the mean and variance of the areas of Voronoi regions; Var(P i ) is the mean and variance of the perimeters of Voronoi regions;

[0051] 3.3) Calculate the variation coefficients of the areas and perimeters of Voronoi regions, that is:

[0052]

[0053] In the formula, CV(A) is the area variation coefficient of Voronoi regions; CV(P) is the perimeter variation coefficient of Voronoi regions.

[0054] Furthermore, the unified measurement standard F1 of the coefficient of variation of area and perimeter is measured as follows:

[0055]

[0056] Furthermore, the material doping image is the material doping image of nodular cast iron.

[0057] A system applying the above method includes a material sample collection device, a visual sample collection module, and a visual processing module;

[0058] The material sample collection device collects material samples;

[0059] The visual sample collection module collects the material doping image of the material sample and transmits it to the visual processing module;

[0060] The visual processing module preprocesses the material doping image, identifies the centroids of the doping particles, processes the centroids of the doping particles using the Bowyer-Watson algorithm, constructs a Voronoi tessellation, and calculates the unified measurement standard F1 of the coefficient of variation of the areas and perimeters of all Voronoi regions in the Voronoi tessellation, so as to evaluate the uniformity of particle doping.

[0061] The technical effect of the present invention is beyond doubt. The present invention provides an end-to-end complete workflow, focuses on considering the spatial relationship between the centroids of the doping particle samples, has excellent robustness to the change of point density, and provides the uniformity measured by double indicators, thus reasonably realizing the objective evaluation of the distribution of doping particles in the sample. Brief Description of the Drawings

[0062] Figure 1 is the hardware overview diagram of the system for evaluating the uniformity of particle doping in the material;

[0063] Figure 2 is the overall workflow diagram of the system for evaluating the uniformity of particle doping in the material;

[0064] Figure 3 is the workflow diagram of machine vision preprocessing;

[0065] Figure 4 is the demonstration of the construction process of the Voronoi diagram of two generated points;

[0066] Figure 5 is the demonstration of the construction process of the Voronoi diagram of multiple generated points;

[0067] Figure 6 is the collected image sample;

[0068] Figure 7 is the grayscale image obtained after conversion;

[0069] Figure 8 is the binarization result;

[0070] Figure 9 is the Canny edge detection result;

[0071] Figure 10 is the result of topological closing operation;

[0072] Figure 11 is the result of calculating the significant object contour;

[0073] Figure 12 is the result of calculating the contour centroid;

[0074] Figure 13 is to create a Voronoi tessellation with all centroids as the set of generating points;

[0075] Figure 14 is the distribution of the areas and perimeters of the Voronoi regions. Detailed implementation manner

[0076] The present invention will be further described below in conjunction with embodiments, but it should not be understood that the above-mentioned subject matter scope of the present invention is limited to the following embodiments. Without departing from the above technical idea of the present invention, various substitutions and changes made according to common general technical knowledge and conventional means in the art shall be included within the protection scope of the present invention.

[0077] Embodiment 1:

[0078] Refer to Figures 1 to 14 , a method for evaluating the particle doping uniformity in a material, comprising the following steps:

[0079] 1) Obtain the material doping image, perform preprocessing, and identify the centroids of the doping particles;

[0080] 2) Use the Bowyer-Watson algorithm to process the centroids of the doping particles and construct a Voronoi tessellation;

[0081] 3) Calculate the coefficient of variation of the area and perimeter of each Voronoi region in the Voronoi tessellation, and calculate the unified measure F1 metric of the coefficient of variation of the area and perimeter, so as to evaluate the doping uniformity of the particles.

[0082] In step 1), the steps of performing preprocessing include:

[0083] 1.1) Convert the material doping image into a grayscale image;

[0084] 1.2) Perform binarization processing on the grayscale image to obtain a binarized image;

[0085] 1.3) Perform Canny edge detection on the binary image to identify the edges of the material particles;

[0086] 1.4) Perform multiple topological closing operations on the edges of the material particles to form multiple contours;

[0087] 1.5) Calculate the centroid of each contour as the centroid of the doped particles.

[0088] In step 1.5), the x - coordinate and y - coordinate of the centroid of the contour are obtained by dividing the first - order moments in the x and y directions by the area of the contour, that is:

[0089]

[0090] where, ∫∫ R represents the double integral over the region R, dA is the area element; ∫∫ R xρ(x,y)dA, ∫∫ R yρ(x,y)dA are the first - order moments in the x and y directions; ∫∫ R ρ(x,y)dA is the area of the contour.

[0091] In step 2), the steps to construct the Voronoi tessellation include: using all the centroids of the doped particles as the generating points, dividing into multiple Voronoi regions according to the distances between these points, and each region encloses a generating point, thus constructing the Voronoi tessellation.

[0092] The i - th generating point s i corresponding Voronoi region V(s i ) is as follows:

[0093]

[0094] where, ‖P - s i ‖ represents the Euclidean distance from point P to the generating point s i , and P is the set of generating points.

[0095] The Voronoi tessellation V(S) is as follows:

[0096]

[0097] The Voronoi edge E(s i ,s j ) is as follows:

[0098]

[0099] where, ‖P - s i ‖ represents the Euclidean distance from point P to the generating point s i .

[0100] The Voronoi vertex V(s i , s j , s k ) is as follows:

[0101]

[0102] In step 3), the steps of calculating the area and perimeter variation coefficients of each Voronoi region in the Voronoi tessellation include:

[0103] 3.1) Calculate the area and perimeter of each Voronoi region;

[0104] Among them, the area and perimeter of the i-th Voronoi region are as follows:

[0105]

[0106] In the formula, A(V(s i )) and P(V(s i )) are the area and perimeter distributions; (x i , y i ), (x i+1 , y i+1 ) are coordinates;

[0107] 3.2) Calculate the mean and variance of the areas and perimeters of all Voronoi regions, that is:

[0108]

[0109] Among them, n is the number of generated points, that is, the number of doped particles identified in the image sample of material doping; Var(A i ) is the mean and variance of the areas of Voronoi regions; Var(P i ) is the mean and variance of the perimeters of Voronoi regions;

[0110] 3.3) Calculate the variation coefficients of the areas and perimeters of Voronoi regions, that is:

[0111]

[0112] In the formula, CV(A) is the area variation coefficient of Voronoi regions; CV(P) is the perimeter variation coefficient of Voronoi regions.

[0113] The unified measurement standard F1 of the area and perimeter variation coefficients is as follows:

[0114]

[0115] The material doping image is the material doping image of ductile cast iron.

[0116] Example 2:

[0117] A method for evaluating the uniformity of particle doping in a material, comprising the following steps:

[0118] 1) Obtain the material doping image, perform preprocessing, and identify the centroids of the doping particles;

[0119] 2) Use the Bowyer-Watson algorithm to process the centroids of the doping particles and construct a Voronoi tessellation;

[0120] 3) Calculate the coefficient of variation of the area and perimeter of each Voronoi region in the Voronoi tessellation, and calculate the unified measure F1 metric of the coefficient of variation of the area and perimeter, so as to evaluate the uniformity of particle doping.

[0121] Example 3:

[0122] A method for evaluating the uniformity of particle doping in a material, the technical content is the same as that in Example 2. Further, in step 1), the steps of preprocessing include:

[0123] 1.1) Convert the material doping image into a grayscale image;

[0124] 1.2) Perform binarization processing on the grayscale image to obtain a binarized image;

[0125] 1.3) Perform Canny edge detection on the binarized image to identify the edges of the material particles;

[0126] 1.4) Perform multiple topological closing operations on the edges of the material particles to form multiple contours;

[0127] 1.5) Calculate the centroid of each contour as the centroid of the doping particle.

[0128] Example 4:

[0129] A method for evaluating the uniformity of particle doping in a material, the technical content is the same as any one of Examples 2-3. Further, in step 1.5), the x-coordinate and y-coordinate of the centroid of the contour are respectively obtained by dividing the first-order moments in the x and y directions by the area of the contour.

[0130]

[0131] where, ∫∫ R represents the double integral over the region R, and dA is the area element.

[0132] Example 5:

[0133] A method for evaluating the uniformity of particle doping in a material, the technical content is the same as any one of Embodiments 2-4. Further, in step 2), the steps of constructing a Voronoi tessellation include: using the centroids of all doped particles as generating points, and dividing multiple Voronoi regions according to the distances between these points, with each region enclosing a generating point, thereby constructing a Voronoi tessellation.

[0134] Embodiment 6:

[0135] A method for evaluating the uniformity of particle doping in a material, the technical content is the same as any one of Embodiments 2-5. Further, the i-th generating point s i The corresponding Voronoi region V(s i ) is as follows:

[0136]

[0137] Where, ‖P - s i ‖ represents the Euclidean distance from point P to the generating point s i , and P is the set of generating points.

[0138] Embodiment 7:

[0139] A method for evaluating the uniformity of particle doping in a material, the technical content is the same as any one of Embodiments 2-6. Further, the Voronoi tessellation is as follows:

[0140]

[0141] The Voronoi edge E(s i , s j ) is as follows:

[0142]

[0143] Where, ‖P - s i ‖ represents the Euclidean distance from point P to the generating point s i .

[0144] The Voronoi vertex V(s i , s j , s k ) is as follows:

[0145]

[0146] Embodiment 8:

[0147] A method for evaluating the particle doping uniformity in a material, the technical content being the same as any one of Embodiments 2-7. Further, in step 3), the steps of calculating the area and the coefficient of variation of the perimeter of each Voronoi region in the Voronoi tessellation diagram include:

[0148] 3.1) Calculate the area and the perimeter of each Voronoi region;

[0149] Among them, the area and the perimeter of the i-th Voronoi region are as follows:

[0150]

[0151] 3.2) Calculate the mean and variance of the areas and perimeters of all Voronoi regions, that is:

[0152]

[0153] Among them, n is the number of generated points, that is, the number of doping particles identified in the image sample of the material doping.

[0154] 3.3) Calculate the coefficient of variation of the area and the perimeter of the Voronoi region, that is:

[0155]

[0156] In the formula, CV(A) is the coefficient of variation of the Voronoi region area; CV(P) is the coefficient of variation of the Voronoi region perimeter.

[0157] Embodiment 9:

[0158] A method for evaluating the particle doping uniformity in a material, the technical content being the same as any one of Embodiments 2-8. Further, the unified measurement standard F1 of the area and the coefficient of variation of the perimeter is measured as follows:

[0159]

[0160] Embodiment 10:

[0161] A method for evaluating the particle doping uniformity in a material, the technical content being the same as any one of Embodiments 2-9. Further, the material doping image is an image of a ductile iron material doping.

[0162] Embodiment 11:

[0163] A system applying the method described in any one of Embodiments 1-10 includes a material sample collection device, a visual sample collection module, and a visual processing module;

[0164] The material sample collection device collects a material sample;

[0165] The visual sample acquisition module acquires the material doping image of the material sample and transmits it to the visual processing module;

[0166] The visual processing module preprocesses the material doping image, identifies the centroid of the doping particles, processes the centroid of the doping particles using the Bowyer-Watson algorithm, constructs a Voronoi tessellation, and calculates the unified measure F1 metric of the area and perimeter coefficient of variation of all Voronoi regions in the Voronoi tessellation, thereby evaluating the uniformity of particle doping.

[0167] Example 11:

[0168] A method for evaluating the uniformity of particle doping in a material is as follows:

[0169] Machine vision preprocessing step: Use machine vision technology to preprocess the material doping image sample, identify, locate, and segment the particles in the sample; calculate the centroid of the identified particles;

[0170] Step of constructing Voronoi tessellation: Based on the centroid calculated in step b, use the Bowyer-Watson algorithm to construct a Voronoi tessellation;

[0171] Statistical evaluation step: For the Voronoi tessellation constructed in b, calculate the area and perimeter of each Voronoi region; calculate the two coefficients of variation of the obtained area and perimeter; calculate the unified measure F1 metric of the two coefficients of variation, and finally evaluate the uniformity of particle doping through the obtained F1 metric.

[0172] According to the method described in claim 1, wherein the machine vision preprocessing step includes:

[0173] Extraction and preparation of the original image, ensuring that the boundary between grains and grain boundaries in the image is clearly visible;

[0174] Convert the extracted image into a grayscale image, and the weights of the red, green, and blue channels used are 0.300, 0.550, and 0.150 respectively;

[0175] Perform binarization processing on the grayscale image, and the threshold used is 120;

[0176] Perform Canny edge detection, and the applied convolution kernel size is 5×5;

[0177] Perform multiple topological closing operations, and the number of iterations of the combined operation of dilation and erosion used is 3 times;

[0178] Calculate the significant object contour, noting that smaller contours should be defined as noise and removed;

[0179] Calculate the centroid of the contour, and calculate the centroid of each particle contour by dividing the first-order moments in the x and y directions by the area of the contour.

[0180] The method according to claim 1, wherein the step of constructing the Voronoi tessellation includes:

[0181] Use all the centroids as the set of generating points to create a Voronoi tessellation through the Bowyer-Watson algorithm.

[0182] The step of statistical evaluation includes:

[0183] Apply the shoelace formula and the accumulation of Euclidean distances between vertices to calculate the area and perimeter of each Voronoi region;

[0184] Calculate the mean and variance of the areas and perimeters of all Voronoi regions through standard statistical methods;

[0185] Calculate the coefficient of variation CV(A) and CV(P) of the area and perimeter of the Voronoi region;

[0186] Calculate a unified measure of the two coefficients of variation CV(A) and CV(P). The present invention uses the F1 metric, which is the harmonic mean of CV(A) and CV(P).

[0187] The system includes:

[0188] A material sample collection device;

[0189] A visual sample collection device for collecting visual images of the material sample;

[0190] A visual processing computer for executing the evaluation method.

[0191] The visual sample collection device includes an optical microscope, an electron microscope, or a metallographic microscope.

[0192] The visual processing computer includes a network device, an input / output device, and a report printer for data exchange and result output.

[0193] The method evaluates the uniformity of particle doping by calculating the coefficient of variation and its harmonic mean of the area and perimeter of the Voronoi region.

[0194] The method determines the distribution uniformity of particle centroids by comparing the coefficients of variation of the area and perimeter of the Voronoi region.

[0195] The method is applicable to the doping of graphite particles in ductile iron and the doping of other particulate materials.

[0196] This method provides uniformity measured by dual indicators to achieve an objective evaluation of the distribution of doped particles in a sample.

[0197] This method focuses on considering the spatial relationship between the centroids of doped particle samples and has excellent robustness to changes in point density.

[0198] In summary, this method combines machine vision, the Voronoi tessellation method in computational geometry, and statistical methods. Through preprocessing of material doping image samples, it identifies, locates, and segments particles, calculates centroids, constructs a Voronoi tessellation using the Bowyer-Watson algorithm, and statistically analyzes the area and perimeter distributions of Voronoi polygons, calculates the coefficient of variation and its combined harmonic mean F1 metric, so as to achieve an objective evaluation of the particle doping uniformity. The system includes hardware such as a material sample collection device, a visual sample collection device, and a visual processing computer, as well as an end-to-end automatic recognition system that combines software and hardware. This method and system can objectively and accurately evaluate the particle distribution state, improve the material doping efficiency and product quality, and are applicable to ductile iron and other materials science fields related to particle material doping.

[0199] Example 12:

[0200] A method for evaluating the uniformity of particle doping in a material is as follows:

[0201] This invention takes the graphite particle doping in ductile iron as an example, but the applicability of this method is not limited to graphite particle doping and can be extended to other materials science fields related to particle material doping.

[0202] The hardware part of the system in this example consists of a material sample collection device, a visual sample collection device (an optical microscope in this example), a visual processing computer, a network device, an input / output device, a report printer, and accessories such as data transmission cables.

[0203] Original material sample collection

[0204] In the ductile iron casting process, a small amount of high-temperature cast iron solution is extracted from a high-temperature furnace using a crucible, and the cast iron in the solution is corroded with a corrosive agent to obtain doped graphite particles.

[0205] Visual sample collection

[0206] The doped graphite particles are observed and photographed with an optical microscope to obtain an image sample.

[0207] Machine vision preprocessing

[0208] Original image extraction and preparation

[0209] First, observe the image samples. If the clarity and contrast of the images are not ideal, perform corresponding image enhancement to ensure that the grains and boundaries in the images are clearly visible. The results are as Figure 6 shown.

[0210] Convert to grayscale image

[0211] After that, convert the original image to a grayscale image. The weights of the red, green, and blue channels used are 0.300, 0.550, and 0.150 respectively. The resulting grayscale image is as Figure 7 shown.

[0212] After binarizing the grayscale image, perform binarization on the grayscale image using a threshold of 120. The results are as Figure 8 shown.

[0213] Perform Canny edge detection

[0214] After that, apply a 5×5 convolution kernel to perform Canny edge detection on the binarization result. The results are as Figure 9 shown.

[0215] Perform multiple topological closing operations

[0216] After that, perform a topological closing operation on the Canny edge detection result through a combination of dilation and erosion with 3 iterations. The results are as Figure 10 shown.

[0217] Calculate the significant object contours

[0218] After that, calculate the contours of the significant objects for the result of the topological closing operation. The results are as Figure 11 shown. Here, note that the smaller contours need to be defined as noise and removed.

[0219] Calculate the centroid of the contour. After that, calculate the centroid of each particle contour for the result of calculating the significant object contours by dividing the first-order moments in the x and y directions by the area of the contour. The results are as Figure 12 shown.

[0220] Construct a Voronoi tessellation

[0221] After obtaining the centroids of all significant object contours, use all the centroids as the set of generating points and create a Voronoi tessellation through the Bowyer-Watson algorithm. The results are as Figure 13 shown, and the Voronoi regions are filled with colors for distinction, and the edge Voronoi regions are filled with white.

[0222] Statistical evaluation

[0223] After constructing a Voronoi diagram with the centroids of all irregularly doped particles in the material doping image sample as the set of generation points, the particle doping uniformity of the material can be objectively evaluated based on this through statistical methods.

[0224] First, apply the shoelace formula and the accumulation of Euclidean distances between vertices to calculate the area and perimeter of each Voronoi region. In this example, the size of the visual sample is 1280×960; the number of recognized significant target contours is 71; therefore, there are 71 centroids in total, and these 71 centroids are used as the set of generation points for creating the Voronoi tessellation; the created Voronoi tessellation also contains 71 Voronoi regions. The area and perimeter distribution results of the Voronoi regions calculated for the example sample are as Figure 14 shown.

[0225] Then, calculate the mean and variance of the areas and perimeters of all Voronoi regions to obtain:

[0226]

[0227] where n is the number of generation points, which is 73 in this example. The two means are as shown by the red broken lines in the two subgraphs in Figure 14 shown.

[0228] After that, calculate the coefficient of variation of the areas and perimeters of the Voronoi regions

[0229]

[0230] Finally, calculate the F1 measure of the two coefficients of variation

[0231]

[0232] This method finally evaluates the particle doping uniformity in the material through the F1 measure. The smaller the F1 measure, the more consistent the distribution of the Voronoi tessellation polygons in terms of size and shape. Correspondingly, it also means that the distribution of particle centroids is more uniform. In the analysis of this example sample, the F1 measure obtained by this method is approximately 0.96. Although this value is not the theoretical minimum, based on empirical judgment, it is still within an acceptable range. Therefore, it can be considered that the particle doping uniformity in the sample material meets the specified standard.

[0233] So far, a complete workflow from start to finish has been completed, comprehensively evaluating the uniformity of particle doping in a sample material. This invention particularly focuses on the spatial relationship between the centroids of doped particle samples, demonstrates high robustness to changes in point density, and achieves a balanced evaluation using the dual-index F1 metric. Through these methods, it is possible to quantitatively and objectively evaluate the distribution of doped particles in the sample, providing an innovative evaluation tool for the field of materials science.

Claims

1. A method for evaluating the uniformity of particle doping in a material, characterized in that, It includes the following steps: 1) Obtain the material doping image, perform preprocessing, and identify the centroid of the doping particles. 2) Use the Bowyer-Watson algorithm to process the centroids of the doping particles and construct a Voronoi tessellation; 3) Calculate the coefficient of variation of the area and perimeter of each Voronoi region in the Voronoi tessellation, and calculate the unified measurement standard F1 metric of the coefficient of variation of the area and perimeter, so as to evaluate the uniformity of particle doping.

2. The method for evaluating the particle doping uniformity in a material according to claim 1, wherein In step 1), the steps for preprocessing include: 1.1) Convert the material doping image into a grayscale image; 1.2) Perform binarization on the grayscale image to obtain a binary image; 1_3) Perform Canny edge detection on the binary image to identify the edges of the material particles; 1.4) Perform multiple topological closing operations on the edges of the material particles to form multiple contours; 1.5) Calculate the centroid of each contour as the centroid of the doping particles.

3. A method for evaluating the uniformity of particle doping in a material according to claim 2, characterized in that, In step 1.5), the x c coordinate and the y c coordinate of the centroid of the contour are as follows: where, ∫∫ R represents the double integral over the region R, and dA is the area element; ∫∫ R xρ(x, y)dA, ∫∫ R yρ(x, y)dA are the first moments in the x and y directions; ∫∫ R ρ(x, y)dA is the area of the contour.

4. A method for evaluating the uniformity of particle doping in a material according to claim 1, characterized in that In step 2), the steps for constructing the Voronoi tessellation include: using all the centroids of the doping particles as generating points, dividing multiple Voronoi regions according to the distances between these points, and each region encloses a generating point, thereby constructing the Voronoi tessellation.

5. A method for evaluating the uniformity of particle doping in a material according to claim 4, characterized in that, The i-th generation point s i The corresponding Voronoi region V(s i ) is as follows: Among them, ||P - s i || represents the Euclidean distance from point P to the generated point s i by the Euclidean distance method, and P is the set of generated points.

6. A method for evaluating the uniformity of particle doping in a material according to claim 4, characterized in that, The Voronoi tessellation V(S) is as follows: Voronoi edge E(s i , s j ) is as follows: Among them, ||P - s i || represents the Euclidean distance from point P to the generated point s i ; Voronoi vertex V(s i , s j , s k ) is as follows:

7. A method for evaluating the uniformity of particle doping in a material according to claim 1, characterized in that In step 3), the steps for calculating the coefficient of variation of the area and perimeter of each Voronoi region in the Voronoi tessellation include: 3.1) Calculate the area and perimeter of each Voronoi region; Among them, the area and perimeter of the i-th Voronoi region are distributed as follows: Wherein, A(V(s i )) and P(V(s i )) are the area and perimeter distributions; (x i , y i ), (x i+1 , y i+1 ) are coordinates; 3.2) Calculate the mean and variance of the areas and perimeters of all Voronoi regions, that is: where n is the number of generated points, that is, the number of doped particles identified in the image sample of material doping; Var(A i ) is the mean and variance of the Voronoi region area; Var(P i ) is the mean and variance of the Voronoi region perimeter; 3.3) Calculate the coefficient of variation of the area and perimeter of the Voronoi region, that is: In the formula, CV(A) is the coefficient of variation of the area of the Voronoi region; CV(P) is the coefficient of variation of the perimeter of the Voronoi region.

8. A method for evaluating the uniformity of particle doping in a material according to claim 1, characterized in that, The unified measurement standard F1 metric of the coefficient of variation of the area and perimeter is as follows:

9. The method for evaluating the particle doping uniformity in a material according to claim 1, wherein The material doping image is a material doping image of ductile iron.

10. A system applying the method according to any one of claims 1-9, characterized in that, It includes a material sample collection device, a visual sample collection module, and a visual processing module; The material sample collection device collects material samples; The visual sample collection module collects the material doping image of the material sample and transmits it to the visual processing module; The visual processing module performs preprocessing on the material doping image, identifies the centroid of the doping particles, uses the Bowyer-Watson algorithm to process the centroid of the doping particles, constructs a Voronoi tessellation, and calculates the unified measurement standard F1 metric of the coefficient of variation of the areas and perimeters of all Voronoi regions in the Voronoi tessellation, so as to evaluate the uniformity of particle doping.

Images