A method and device for obtaining primary particle size of ternary positive electrode material

Abstract

The present invention discloses a method for obtaining the primary particle size of a ternary positive electrode material. The method pre-processes the primary particle SEM image to remove image noise and small particle interference. In view of the uncertain edge direction of the primary particle, the image gradient is calculated using a fractional gradient operator of twelve edge directions, the maximum value of the twelve edge gradients is taken, the most correct gradient direction in the image gradient map is found, and the image gradient map is obtained. The image gradient map obtained by calculation is morphologically calculated to find the potential position of the primary particle, and the foreground marking map is marked to obtain the foreground marking map. The image gradient map is combined with the foreground marking map to obtain the foreground marking gradient map, the foreground marking gradient map is subjected to watershed segmentation calculation, and area threshold processing is performed to obtain the primary particle segmentation result, and finally the result is subjected to connected domain statistical analysis to obtain the primary particle size distribution. The particle segmentation of the ternary positive electrode material of the present invention is accurate, and the primary particle size distribution is accurately obtained.

Description

Technical Field

[0001] The invention relates to the technical field of obtaining microscopic indicators in a ternary cathode material preparation process, and in particular discloses a method and a device for obtaining a primary particle size in a growth reaction stage in a ternary cathode material preparation process based on a SEM image. Background Art

[0002] With the progress and development of society, the population is continuously increasing at this stage, while energy is indeed limited, and the energy crisis is gradually intensifying. Secondly, the energy waste and environmental pollution generated by people in the production process are also important problems that have been difficult to solve. Therefore, people need to continuously optimize and improve the way of energy utilization and production and living. Lithium-ion batteries (LIB, Lithium-Ion Battery) play an important role in solving problems such as environmental pollution and resource scarcity. They have the advantages of high specific capacity and pollution-free. They are one of the most important energy storage devices in the future, thus causing a research boom. In lithium-ion batteries, positive electrode materials play an important role and are the key to their development. The preparation of high-capacity, good thermal stability, good rate performance, long service life, and low-cost high-performance positive electrode materials has become an important research and development task in recent years. At the same time, the application and development of lithium-ion lithium battery ternary positive electrode materials have a vital impact on the use of new energy and the promotion and popularization of new energy vehicles.

[0003] In the industrial sintering process, factories usually use roller kilns for a long high-temperature calcination process, which includes a heating section, a constant temperature section, and a cooling section. Each temperature section is divided into multiple temperature zones, and the temperature and atmosphere between the temperature zones are coupled and affect each other. At the same time, the reactions occurring in each temperature zone are also different. The quality of the sintered product is closely related to the temperature, atmosphere, sintering time, etc. in each temperature zone. After 22 hours of calcination and complex physical and chemical reaction changes, the desired sintered product can be finally obtained. However, there is currently no research on the method of obtaining microscopic indicators of the sintering process of ternary materials, and the law of primary particle size changes during the sintering process of ternary materials is still unclear.

[0004] Understanding and analyzing the distribution and variation of primary particle size during the sintering process of positive electrode materials, establishing the relationship between primary particle size and product performance indicators, and then clarifying the relationship between reaction environment conditions and product performance indicators are of great significance for dynamically adjusting sintering heat and oxygen supply and realizing the combination of micro and macro, and are the key to improving the yield of ternary positive electrode materials.

[0005] Therefore, there is no research on the method of obtaining microscopic indicators of the sintering process of ternary materials in the prior art, which is a technical problem that needs to be solved urgently. Summary of the invention

[0006] The present invention provides a method and device for obtaining the primary particle size of a ternary positive electrode material, aiming to solve the technical problem that there is no research method for obtaining microscopic indicators of the sintering process of the ternary material in the prior art.

[0007] One aspect of the present invention relates to a method for obtaining the primary particle size of a ternary positive electrode material, comprising the following steps:

[0008] Preprocess the primary particle SEM image to remove image noise and small particle interference;

[0009] In view of the uncertainty of the edge direction of primary particles, the image gradient is calculated using the fractional gradient operator of twelve edge directions. The maximum value of the twelve edge gradients is taken to find the most correct gradient direction in the image gradient map to obtain the image gradient map.

[0010] Perform morphological calculation on the image gradient map obtained by calculation, find the potential position of the primary particle, mark it, and obtain the foreground marking map;

[0011] The image gradient map is combined with the foreground marker map to obtain the foreground marker gradient map, and the foreground marker gradient map is subjected to watershed segmentation calculation and area threshold processing to obtain a particle segmentation result. Finally, the result is subjected to connected domain statistical analysis to obtain a particle size distribution.

[0012] Furthermore, the steps of preprocessing the primary particle SEM image to remove image noise and small particle interference include:

[0013] The histogram equalization method was used to enhance the primary particle SEM image;

[0014] Perform brightness conversion on the primary particle SEM image, perform convolution operation on the converted SEM image using a Gaussian kernel, and remove particle shadows from the converted SEM image;

[0015] Perform mean filtering on the de-shadowed image to remove image noise and small particle interference.

[0016] Furthermore, in view of the uncertainty of the edge direction of the primary particle, the image gradient is calculated using the fractional gradient operator of twelve edge directions, the maximum value of the twelve edge gradients is taken, and the most correct gradient direction in the image gradient map is found. The steps of obtaining the image gradient map include:

[0017] Define twelve directions of edge gradient directions;

[0018] Perform fractional derivative difference derivation on all edge directions of the edge gradient directions in the defined twelve directions to obtain fractional differential mask operators in the twelve directions;

[0019] Use the obtained fractional differential mask operator to perform a convolution operation on the mean filtered image g(x, y) to calculate the gradient map G(x, y);

[0020] The maximum values ​​of the twelve gradient calculations are selected as the gradient values ​​of each pixel, and finally a gradient image of the primary particle SEM image is obtained.

[0021] Further, the steps of performing morphological calculation on the calculated image gradient map, finding the potential position of the primary particle, and marking it, and obtaining the foreground marking map include:

[0022] Given a square structure element O, open and close the gradient image MG(x, y) with the structure element O to smooth the gradient image as much as possible, eliminate the interference of small impurities and retain the complete edges of particles to obtain a smooth gradient image.

[0023] Binarize the smoothed gradient image to find the possible location of the primary particles, then close the binary image with the structure element O to fill the binary holes and obtain the preliminary foreground mark of the particles.

[0024] Take the central area block of the preliminary foreground mark to obtain the particle foreground mark image IG (x, y).

[0025] Further, the steps of performing morphological calculation on the calculated image gradient map, finding the potential position of the primary particle, and marking it, and obtaining the foreground marking map include:

[0026] The gradient image MG(x, y) is combined with the particle foreground mark image IMG(x, y) to obtain the gradient image IMG(x, y) with particle foreground mark. The gradient image IMG(x, y) with particle foreground mark is segmented using the watershed algorithm to obtain the watershed line WS(x, y), which is the edge contour of the particle.

[0027] The segmentation result WS(x, y) is processed by area threshold to remove under-scores and excessive interference;

[0028] Perform a connected domain statistical analysis on the watershed WS(x, y) to obtain the connected domain BW(i) and its corresponding area size S(i), i=1, 2, 3...n, n is the number of connected domain statistics, that is, the number of primary particles;

[0029] Given area thresholds thresholdH and thresholdL, perform area threshold processing on the connected domain;

[0030] By counting the area size and distribution of the connected domain BW′(i) after area threshold processing, the size distribution of the primary particles of the ternary positive electrode material during the sintering process is obtained.

[0031] Another aspect of the present invention relates to a device for obtaining the primary particle size of a ternary positive electrode material, comprising:

[0032] A preprocessing module is used to preprocess the primary particle SEM image to remove image noise and small particle interference;

[0033] The calculation module is used to calculate the image gradient using the fractional gradient operator of twelve edge directions in view of the uncertainty of the edge direction of the primary particles, take the maximum value of the twelve edge gradients, find the most correct gradient direction in the image gradient map, and obtain the image gradient map;

[0034] The marking module is used to perform morphological calculation on the image gradient map obtained by calculation, find the potential position of the primary particle, and mark it to obtain a foreground marking map;

[0035] The acquisition module is used to combine the image gradient map with the foreground marker map to obtain the foreground marker gradient map, perform watershed segmentation calculation on the foreground marker gradient map, and perform area threshold processing to obtain a particle segmentation result. Finally, the result is statistically analyzed in the connected domain to obtain a particle size distribution.

[0036] Furthermore, the preprocessing module includes:

[0037] An image enhancement unit, used for enhancing the primary particle SEM image by using a histogram equalization method;

[0038] An image operation unit is used to perform brightness conversion on a primary particle SEM image, perform convolution operation on the converted SEM image using a Gaussian kernel, and remove particle shadows from the converted SEM image;

[0039] The image filtering unit performs mean filtering on the de-shadowed image to remove image noise and small particle interference.

[0040] Furthermore, the calculation module includes:

[0041] A definition unit is used to define edge gradient directions in twelve directions;

[0042] A derivation unit, used for performing fractional-order derivative difference derivation on all edge directions of the edge gradient directions in the defined twelve directions, to obtain fractional-order differential mask operators in the twelve directions;

[0043] A calculation unit, used to perform a convolution operation on the mean filtered image g(x, y) using the obtained fractional-order differential mask operator to calculate a gradient map G(x, y);

[0044] The selection unit is used to select the maximum value of twelve gradient calculations as the gradient value of each pixel, and finally obtain a gradient image of the primary particle SEM image.

[0045] Furthermore, the marking module includes:

[0046] The first operation unit is used to give a square structure element O, and perform opening and closing operations on the gradient image MG (x, y) with the structure element O, so as to smooth the gradient image as much as possible, eliminate the interference of small impurity edges, retain the complete edges of particles, and obtain a smooth gradient image;

[0047] The second operation unit is used to perform binarization processing on the smoothed gradient image, find the possible location of the primary particles, and then perform a closing operation on the binary image with the structure element O to fill the binary holes and obtain the preliminary foreground mark of the particles;

[0048] The acquisition unit is used to obtain the central area block of the preliminary foreground mark to obtain the particle foreground mark image IG (x, y).

[0049] Furthermore, the acquisition module includes:

[0050] A segmentation unit is used to combine the gradient image MG(x, y) with the particle foreground mark image IMG(x, y) to obtain the gradient image IMG(x, y) with the particle foreground mark, and segment the gradient image IMG(x, y) with the particle foreground mark using the watershed algorithm to obtain the watershed line WS(x, y), that is, the edge contour of the particle;

[0051] The first processing unit is used to process the segmentation result WS(x, y) by area threshold to remove under-scores and excessive interference;

[0052] The analysis unit is used to perform a connected domain statistical analysis on the watershed line WS(x, y) to obtain the connected domain BW(i) and its corresponding area size S(i), i=1, 2, 3...n, n is the statistical number of connected domains, that is, the number of primary particles;

[0053] The second processing unit is used to set area thresholds thresholdH and thresholdL to perform area threshold processing on the connected domain;

[0054] The statistical unit is used to count the area size and distribution of the connected domain BW′(i) after area threshold processing, that is, to obtain the size distribution of the primary particles of the ternary positive electrode material during the sintering process.

[0055] The beneficial effects achieved by the present invention are:

[0056] The invention discloses a method and device for obtaining the size of primary particles of a ternary positive electrode material. The method comprises the following steps: preprocessing a SEM image of a primary particle to remove image noise and interference of small particles; in view of the uncertain edge direction of the primary particle, using a fractional-order gradient operator of twelve edge directions to calculate the image gradient, taking the maximum value of the twelve edge gradients, finding the most correct gradient direction in an image gradient map, and obtaining an image gradient map; performing morphological calculation on the calculated image gradient map to find the potential position of the primary particles, and marking them to obtain a foreground marking map; combining the image gradient map with the foreground marking map to obtain a foreground marking gradient map, performing watershed segmentation calculation on the foreground marking gradient map, and performing area threshold processing to obtain a primary particle segmentation result; and finally performing connected domain statistical analysis on the result to obtain the primary particle size distribution. The method and device for obtaining the primary particle size of a ternary positive electrode material disclosed in the present invention can obtain the size of the primary particles of the ternary positive electrode material during the sintering process from the SEM image, filling the gap that the microscopic indicators of the ternary positive electrode material during the sintering process are difficult to obtain; a fractional-order differential operator based on twelve edge directions is designed, which can effectively calculate the most accurate gradient direction in the image and realize accurate segmentation of particles. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 A schematic flow chart of an embodiment of a method for obtaining primary particle size of a ternary positive electrode material provided by the present invention;

[0058] Figure 2 The SEM image of the ternary positive electrode material in actual production selected from the method for obtaining the primary particle size of the ternary positive electrode material provided by the present invention;

[0059] Figure 3 The image preprocessing result in the method for obtaining the primary particle size of the ternary positive electrode material provided by the present invention;

[0060] Figure 4 The calculation result of the image gradient map in the method for obtaining the primary particle size of the ternary positive electrode material provided by the present invention;

[0061] Figure 5 The foreground marking result of the primary particles in the method for obtaining the primary particle size of the ternary positive electrode material provided by the present invention;

[0062] Figure 6 The segmentation result of the primary particles in the method for obtaining the primary particle size of the ternary positive electrode material provided by the present invention;

[0063] Figure 7 The statistical result of the primary particle area distribution in the method for obtaining the primary particle size of the ternary positive electrode material provided by the present invention;

[0064] Figure 8It is an edge trend shape diagram of the edge gradient directions in twelve directions defined in the method for obtaining the primary particle size of the ternary positive electrode material provided by the present invention;

[0065] Fig. 9 The fractional differential mask operator in twelve directions in the method for obtaining the primary particle size of the ternary positive electrode material provided by the present invention;

[0066] Fig.10 A square structural element O is given as a primary particle foreground mark in the primary particle size acquisition method of the ternary positive electrode material provided by the present invention;

[0067] Fig.11 for Figure 1 A detailed flow chart of the steps of preprocessing the primary particle SEM image to remove image noise and small particle interference is shown in ;

[0068] Fig.12 for Figure 1 As shown in the figure, in view of the uncertainty of the edge direction of primary particles, the image gradient is calculated using fractional gradient operators of twelve edge directions, the maximum value of the twelve edge gradients is taken, the most correct gradient direction in the image gradient map is found, and the detailed flow chart of the steps of obtaining the image gradient map is shown;

[0069] Fig.13 for Figure 1 A detailed flow chart of the steps of performing morphological calculation on the calculated image gradient map, finding the potential position of the primary particles, and marking them to obtain a foreground marking map as shown in FIG.

[0070] Fig.14 for Figure 1 As shown in , the image gradient map is combined with the foreground label map to obtain the foreground label gradient map, the foreground label gradient map is subjected to watershed segmentation calculation, and area threshold processing is performed to obtain a particle segmentation result, and finally the result is subjected to connected domain statistical analysis to obtain a particle size distribution step refinement flow diagram;

[0071] Fig.15 A functional block diagram of an embodiment of a device for obtaining primary particle size of a ternary positive electrode material provided by the present invention;

[0072] Fig.16 for Fig.15 The functional module diagram of the preprocessing module 1 embodiment shown in FIG.

[0073] Fig.17 for Fig.15 A functional module diagram of an embodiment of a computing module shown in FIG.

[0074] Fig.18 for Fig.15A functional module diagram of a marking module embodiment shown in FIG.

[0075] Fig.19 for Fig.15 A functional module diagram of an acquisition module embodiment shown in FIG.

[0076] Description of Figure Numbers:

[0077] 10. Preprocessing module; 20. Calculation module; 30. Marking module; 40. Acquisition module; 11. Image enhancement unit; 12. Image operation unit; 13. Image filtering unit; 21. Definition unit; 22. Derivation unit; 23. Calculation unit; 24. Selection unit; 31. First operation unit; 32. Second operation unit; 33. Acquisition unit; 41. Segmentation unit; 42. First processing unit; 43. Analysis unit; 44. Second processing unit; 45. Statistics unit. DETAILED DESCRIPTION

[0078] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0079] like Figures 1 to 10 As shown, the first embodiment of the present invention provides a method for obtaining the primary particle size of a ternary positive electrode material, comprising the following steps:

[0080] Step S100: pre-processing the primary particle SEM image to remove image noise and small particle interference.

[0081] According to the characteristics of blurry and dark primary particle SEM images, the images are preprocessed to remove image noise and small particle interference, making the particle edges and particles clearer.

[0082] Step S200: In view of the uncertainty of the edge direction of the primary particle, the image gradient is calculated using the fractional gradient operator of twelve edge directions, the maximum value of the twelve edge gradients is taken, the most correct gradient direction in the image gradient map is found, and the image gradient map is obtained.

[0083] In view of the uncertainty of the edge direction of primary particles, the fractional gradient operator of twelve edge directions is used to calculate the image gradient. The maximum value of the twelve edge gradients is taken to find the most correct gradient direction in the image.

[0084] Step S300: Perform morphological calculation on the image gradient map obtained by calculation, find the potential position of the primary particles, and mark them to obtain a foreground marking map.

[0085] Morphological calculation is performed on the calculated image gradient map to find the potential position of the primary particle and mark it to obtain the particle foreground mark of the image.

[0086] Step S400, combining the image gradient map with the foreground label map to obtain the foreground label gradient map, performing watershed segmentation calculation on the foreground label gradient map, and performing area threshold processing to obtain a particle segmentation result, and finally performing connected domain statistical analysis on the result to obtain a particle size distribution.

[0087] The image gradient map is combined with the foreground marker map to obtain the foreground marker gradient map, and the foreground marker gradient map is subjected to watershed segmentation calculation and area threshold processing to obtain a particle segmentation result. Finally, the result is subjected to connected domain statistical analysis to obtain a particle size distribution.

[0088] The present embodiment discloses a method for obtaining the size of primary particles of ternary positive electrode materials, which preprocesses the SEM image of primary particles to remove image noise and small particle interference; in view of the uncertain edge direction of primary particles, the image gradient is calculated using a fractional gradient operator of twelve edge directions, the maximum value of the twelve edge gradients is taken, the most correct gradient direction in the image gradient map is found, and the image gradient map is obtained; the image gradient map obtained by calculation is morphologically calculated to find the potential position of the primary particles, and is marked to obtain a foreground marking map; the image gradient map is combined with the foreground marking map to obtain a foreground marking gradient map, the foreground marking gradient map is watershed segmented, and area threshold processing is performed to obtain a primary particle segmentation result, and finally the result is statistically analyzed by the connected domain to obtain the primary particle size distribution. The method for obtaining the size of primary particles of ternary positive electrode materials provided in the present embodiment obtains the size of primary particles of ternary positive electrode materials during sintering from the SEM image, filling the gap that the microscopic indicators of ternary positive electrode materials during sintering are difficult to obtain; a fractional differential operator based on twelve edge directions is designed, which can effectively calculate the most accurate gradient direction in the image and realize accurate segmentation of particles.

[0089] Further, see Fig.11 , Fig.11 for Figure 1 In the detailed flow chart of step S100 shown in FIG. 1 , in this embodiment, step S100 includes:

[0090] Step S110: using a histogram equalization method to enhance the primary particle SEM image.

[0091] Let variable r k is the gray level of the pixel in the image, and the gray level is normalized so that 0≤r k ≤1, where k = 0, 1, 2, ..., n-1. Using the probability density function p r (r k ) to represent the distribution of image gray levels:

[0092]

[0093] In formula (1), n ​​is the sum of the pixels in the image, n k is the number of pixels at the current gray level, p r (r k ) is the probability density function.

[0094] Given the transformation function T(r), the grayscale value of each pixel after equalization can be directly calculated by the following formula:

[0095]

[0096] In formula (2), n is the sum of the pixels in the image, n j is the number of pixels of the current gray level, and T(r) is the given transformation function.

[0097] Step S120, performing brightness conversion on the primary particle SEM image, performing convolution operation on the converted SEM image using a Gaussian kernel, and removing particle shadows from the converted SEM image.

[0098] Assume variable I is the grayscale expression of the image after histogram equalization, and use the following function conversion relationship to convert the grayscale value of the image:

[0099]

[0100] In formula (3), I is the grayscale expression of the image after histogram equalization, and I′ is the transformed grayscale image. By formula (3), the transformed grayscale image I′ is obtained.

[0101] Define the Gaussian kernel function G(x, y) as:

[0102]

[0103] In formula (4), K is defined as a constant that satisfies the following equation:

[0104]

[0105] In formula (5), c is the Gaussian kernel scale constant, and the convolution operation on the original image I is expressed as:

[0106] I(x,y) n =I(x, y)*G(x, y) (6)

[0107] In formula (6), G(x, y) is the Gaussian kernel function defined by formula (4), and the convolution result I n Contains the brightness between adjacent pixels, and I is the original image.

[0108] By comparing it with the central pixel and performing image contrast conversion based on the comparison scale, the de-shadowed image R(x, y) can be obtained:

[0109] R(x, y) = 255·I n (x, y) r(x，y) (7)

[0110] In formula (7), the exponent r(x, y) = I″(x, y) / I(x, y), so the brightness of the central pixel increases or decreases depending on whether r(x, y) is greater than 1 or less than 1 (surrounded by brighter pixels or darker pixels), achieving the effect of image deshading.

[0111] Step S130: perform mean filtering on the de-shadowed image to remove image noise and small particle interference.

[0112] Select a 3×3 mean filter template and perform mean filtering on the image R(x, y). The pixel at the center of the image is the average value of all pixel values ​​in the template. The operation is shown in the following formula:

[0113]

[0114] In formula (8), S is the template area, R(x, y) is the image after shadow removal, and g(x, y) is the image after mean filtering. According to formula (8), the mean filtered image g(x, y) can be obtained by calculation.

[0115] The present embodiment discloses a method for obtaining the primary particle size of a ternary positive electrode material, which enhances the primary particle SEM image by using a histogram equalization method; performs brightness conversion on the primary particle SEM image, performs convolution operation on the converted SEM image by using a Gaussian kernel, and removes the particle shadows of the converted SEM image; and performs mean filtering on the de-shadowed image to remove image noise and small particle interference. The method for obtaining the primary particle size of a ternary positive electrode material provided in the present embodiment performs mean filtering on the de-shadowed image to obtain the mean filtered image g(x, y), which is convenient for calculating the gradient map G(x, y), and the primary particle size distribution is accurately obtained.

[0116] Preferably, see Fig.12 , Fig.12 for Figure 1 In the detailed flow chart of step S200 shown in FIG. 1 , in this embodiment, step S200 includes:

[0117] Step S210: define twelve directions of edge gradient directions.

[0118] To calculate the image gradient map, we first define the edge gradient directions in twelve directions. The shape of the edge direction is as follows: Figure 8 shown.

[0119] Step S220 , performing fractional-order derivative difference derivation on all edge directions in the defined twelve directions of edge gradient directions to obtain fractional-order differential mask operators in the twelve directions.

[0120] The fractional derivative difference form under the GL definition can be approximately expressed as follows:

[0121]

[0122] In formula (9), D v is a fractional differential operator, v is the order, which takes a value between 0 and 1, and f(x) is the function to be fractionally differentiated. is the Newton binomial, and Γ(x) is the gamma function.

[0123] Extending to two dimensions, we get:

[0124]

[0125]

[0126] In formula (10) and formula (11), h is the sampling interval, which is generally taken as 1 in digital image discrete processing. Figure 8 Taking the edge direction as an example, using the fractional derivative difference form to deduce, we can get:

[0127]

[0128] In formula (12), v is the order (0<v<1), G is the value of the fractional differential, and f(x+1, y) and f(x-1, y) are the next pixel and the previous pixel of the pixel in the x-axis direction, respectively. Figure 8 All the edge directions shown in the figure are derived by fractional derivative difference as above, and we can get Fig. 9 Fractional differential mask operators in twelve directions are shown.

[0129] Step S230: Use the obtained fractional differential mask operator to perform a convolution operation on the mean filtered image g(x, y) to calculate a gradient map G(x, y).

[0130] Use the above fractional differential mask operator to convolve the mean filtered image g(x, y) and calculate the gradient map G(x, y):

[0131]

[0132] In formula (13), G(x, y) is the gradient map, g(x, y) is the image after mean filtering, and f i There are twelve edge-oriented fractional-order mask operators. The direction with the largest gradient value can best represent the direction of change of the grayscale of the pixel in the image.

[0133] Step S240: Select the maximum value of the twelve gradient calculations as the gradient value of each pixel, and finally obtain the gradient image MG (x, y) of the primary particle SEM image.

[0134] The maximum value of the twelve gradient calculations is selected as the gradient value of each pixel, and finally the gradient image MG (x, y) of the primary particle SEM image is obtained.

[0135] MG(x,y)=max{G(x,y)1,...,G(x,y) 12} (14)

[0136] In formula (14), G(x, y) represents the gradient map, and MG(x, y) is the gradient image.

[0137] The present embodiment provides a method for obtaining the primary particle size of a ternary positive electrode material, by defining the edge gradient directions in twelve directions; performing fractional derivative difference derivation on all edge directions of the defined edge gradient directions in the twelve directions, and obtaining the fractional differential mask operators in the twelve directions; using the obtained fractional differential mask operator to perform a convolution operation on the mean filtered image g(x, y), and calculating the gradient map G(x, y); selecting the maximum value of the twelve gradient calculations as the gradient value of each pixel, and finally obtaining the gradient image MG(x, y) of the primary particle SEM image. The method for obtaining the primary particle size of a ternary positive electrode material provided in the present embodiment selects the maximum value of the twelve gradient calculations as the gradient value of each pixel, and obtains the gradient image MG(x, y) of the primary particle SEM image, which is convenient for calculating the gradient image IMG(x, y) with particle foreground marks, and the primary particle size distribution is accurately obtained.

[0138] Further, see Fig.13 , Fig.13 for Figure 1 In the detailed flow chart of step S300 shown in FIG. 1 , in this embodiment, step S300 includes:

[0139] Step S310: Given a square structure element O, the gradient image MG(x, y) is opened and closed with the structure element O to smooth the gradient image as much as possible, eliminate the interference of small impurity edges, retain the complete edges of particles, and obtain a smooth gradient image.

[0140] The foreground marking of the primary particles in the primary particle SEM image is given by a square structure element O, such as Fig.10As shown:

[0141] The gradient image MG(x, y) is opened with the structure element O, and then closed to smooth the gradient image as much as possible, eliminate the interference of small impurities and retain the complete edges of the particles to obtain a smooth gradient image.

[0142] Step S320: Binarize the smoothed gradient image to find the possible locations of primary particles, and then perform a closing operation on the binary image with the structure element O to fill the binary holes and obtain a preliminary foreground mark of the particles.

[0143] The smoothed gradient image MG (x, y) is binarized to obtain a binary image and find the possible location of the primary particles. The binary image is then closed with the structure element O to fill the binary holes and obtain the preliminary foreground mark of the particles.

[0144] Step S330 , taking the central area block of the preliminary foreground mark to obtain a particle foreground mark image IG (x, y).

[0145] The central area block of the preliminary foreground mark is taken from the binary image to obtain the particle foreground mark image IG (x, y).

[0146] The present embodiment provides a method for obtaining the primary particle size of a ternary positive electrode material. By giving a square structure element O, the structure element O is used to perform an opening operation and a closing operation on the gradient image MG (x, y), so as to smooth the gradient image as much as possible, eliminate the interference of small impurity edges, retain the complete edges of the particles, and obtain a smooth gradient image direction; the smoothed gradient image is binarized to find the possible location of the primary particles, and then the structure element O is used to perform a closing operation on the binary image to fill the binary holes and obtain the preliminary foreground mark of the particles; the central area block of the preliminary foreground mark is taken to obtain the particle foreground mark image IG (x, y). The method for obtaining the primary particle size of a ternary positive electrode material provided in the present embodiment takes the central area block of the preliminary foreground mark to obtain the particle foreground mark image IG (x, y), which is convenient for calculating the gradient image IMG (x, y) with the particle foreground mark, and the primary particle size distribution is accurately obtained.

[0147] Preferably, see Fig.14 , Fig.14 for Figure 1 In the detailed flow chart of step S400 shown in FIG. 1 , in this embodiment, step S400 includes:

[0148] Step S410: Combine the gradient image MG(x, y) with the particle foreground mark image IMG(x, y) to obtain the gradient image IMG(x, y) with particle foreground mark, and segment the gradient image IMG(x, y) with particle foreground mark using the watershed algorithm to obtain the watershed line WS(x, y), i.e., the edge contour of the particle.

[0149] The gradient image MG(x, y) is combined with the particle foreground marked image IMG(x, y) to obtain the gradient image IMG(x, y) with particle foreground marks. The gradient image IMG(x, y) with particle foreground marks is segmented using the watershed algorithm to obtain the watershed line WS(x, y), which is the edge contour of the particle.

[0150] Step S420: Perform area threshold processing on the segmentation result WS(x, y) to remove under-segmentation and excessive interference.

[0151] Since over-segmentation, under-segmentation and high adhesion of primary particles during the sintering process are inevitable during watershed segmentation, it is necessary to perform area threshold processing on the segmentation result WS(x, y) to remove under-segmentation and over-interference.

[0152] Step S430, perform connected domain statistical analysis on the watershed WS(x, y) to obtain the connected domain BW(i) and its corresponding area size S(i), i=1, 2, 3...n, n is the statistical number of connected domains, that is, the number of primary particles.

[0153] Perform a connected domain statistical analysis on the watershed WS(x, y) to obtain the connected domain BW(i) and its corresponding area size S(i), i=1, 2, 3...n, n is the statistical number of connected domains, that is, the number of primary particles.

[0154] Step S440: Given area thresholds thresholdH and thresholdL, area threshold processing is performed on the connected domain.

[0155] Given area thresholds thresholdH and thresholdL, perform area threshold processing on the connected domain:

[0156]

[0157] In formula (15), BW′(i) is the connected domain after area threshold processing, thresholdH, thresholdL are given area thresholds, and S(i) is the area size corresponding to the connected domain.

[0158] Step S450, the area size and distribution of the connected domain BW′(i) after area threshold processing are counted, that is, the size distribution of the primary particles of the ternary positive electrode material during the sintering process is obtained.

[0159] By counting the area size and distribution of the connected domain BW′(i) after area threshold processing, the size distribution of the primary particles of the ternary positive electrode material during the sintering process is obtained.

[0160] The present embodiment provides a method for obtaining the primary particle size of a ternary positive electrode material, which combines a gradient image MG(x, y) with a particle foreground mark image IMG(x, y) to obtain a gradient image IMG(x, y) with a particle foreground mark, and uses a watershed algorithm to segment the gradient image IMG(x, y) with the particle foreground mark to obtain a watershed line WS(x, y), that is, the edge contour of the particle; the segmentation result WS(x, y) is processed by an area threshold to remove under-scores and excessive interference; a connected domain statistical analysis is performed on the watershed line WS(x, y) to obtain a connected domain BW(i) and its corresponding area size S(i), i=1, 2, 3...n, n is the statistical number of connected domains, that is, the number of primary particles; area thresholds thresholdH, thresholdL are given, and area threshold processing is performed on the connected domain; the area size and distribution of the connected domain BW′(i) after the area threshold processing are statistically analyzed, that is, the size distribution of the primary particles of the ternary positive electrode material during the sintering process is obtained. The method for obtaining the primary particle size of the ternary positive electrode material provided in this embodiment obtains the size of the primary particles of the ternary positive electrode material during the sintering process from the SEM image, filling the gap that the microscopic indicators of the ternary positive electrode material during the sintering process are difficult to obtain; a fractional-order differential operator based on twelve edge directions is designed, which can effectively calculate the most accurate gradient direction in the image and realize accurate segmentation of particles.

[0161] like Fig.15 As shown, Fig.15 The present invention provides a functional block diagram of an embodiment of a device for obtaining the size of primary particles of a ternary positive electrode material. In this embodiment, the device for obtaining the size of primary particles of a ternary positive electrode material comprises a preprocessing module 10, a computing module 20, a marking module 30 and an acquisition module 40, wherein the preprocessing module 10 is used to preprocess the SEM image of the primary particles to remove image noise and interference from small particles; the computing module 20 is used to calculate the image gradient using a fractional-order gradient operator of twelve edge directions in view of the uncertain edge direction of the primary particles, take the maximum value of the twelve edge gradients, find the most correct gradient direction in the image gradient map, and obtain the image gradient map; the marking module 30 is used to perform morphological calculation on the image gradient map obtained by calculation, find the potential position of the primary particles, and mark them to obtain a foreground marking map; the acquisition module 40 is used to combine the image gradient map with the foreground marking map to obtain a foreground marking gradient map, perform watershed segmentation calculation on the foreground marking gradient map, perform area threshold processing on the foreground marking gradient map, obtain the primary particle segmentation result, and finally perform connected domain statistical analysis on the result to obtain the primary particle size distribution.

[0162] The preprocessing module 10 preprocesses the image according to the blurry and dark characteristics of the primary particle SEM image, removes image noise and small particle interference, and makes the particle edges and particles clearer.

[0163] In view of the uncertainty of the edge direction of primary particles, the calculation module 20 uses fractional gradient operators of twelve edge directions to calculate the image gradient, takes the maximum value of the twelve edge gradients, and finds the most correct gradient direction in the image.

[0164] The marking module 30 performs morphological calculation on the calculated image gradient map, finds the potential position of the primary particle, and marks it to obtain the particle foreground mark of the image.

[0165] The acquisition module 40 combines the image gradient map with the foreground marker map to obtain the foreground marker gradient map, performs watershed segmentation calculation on the foreground marker gradient map, and performs area threshold processing to obtain a particle segmentation result, and finally performs connected domain statistical analysis on the result to obtain a particle size distribution.

[0166] The present embodiment discloses a device for obtaining the size of primary particles of ternary positive electrode materials. The device preprocesses the SEM image of primary particles to remove image noise and small particle interference. In view of the uncertainty of the edge direction of primary particles, the image gradient is calculated by using the fractional order gradient operator of twelve edge directions, the maximum value of the twelve edge gradients is taken, the most correct gradient direction in the image gradient map is found, and the image gradient map is obtained. The image gradient map obtained by calculation is morphologically calculated to find the potential position of the primary particles, and is marked to obtain a foreground marking map. The image gradient map is combined with the foreground marking map to obtain a foreground marking gradient map, and the foreground marking gradient map is subjected to watershed segmentation calculation, and area threshold processing is performed to obtain the primary particle segmentation result. Finally, the result is subjected to connected domain statistical analysis to obtain the primary particle size distribution. The device for obtaining the size of primary particles of ternary positive electrode materials provided in the present embodiment obtains the size of primary particles of ternary positive electrode materials during sintering from the SEM image, filling the gap that the microscopic indicators of ternary positive electrode materials during sintering are difficult to obtain. The fractional order differential operator based on twelve edge directions is designed, which can effectively calculate the most accurate gradient direction in the image and realize accurate segmentation of particles.

[0167] Further, see Fig.16 , Fig.16 for Fig.15, in which the preprocessing module 10 includes an image enhancement unit 11, an image operation unit 12 and an image filtering unit 13, wherein the image enhancement unit 11 is used to enhance the primary particle SEM image by adopting a histogram equalization method; the image operation unit 12 is used to perform brightness conversion on the primary particle SEM image, perform convolution operation on the converted SEM image by using a Gaussian kernel, and remove particle shadows from the converted SEM image; the image filtering unit 13 performs mean filtering on the de-shadowed image to remove image noise and small particle interference.

[0168] The image enhancement unit 11 sets the variable r k is the gray level of the pixel in the image, and the gray level is normalized so that 0≤r k ≤1, where k = 0, 1, 2, ..., n-1. Using the probability density function p r (r k ) to represent the distribution of image gray levels:

[0169]

[0170] In formula (16), n is the sum of the pixels in the image, n k is the number of pixels at the current gray level, p r (r k ) is the probability density function.

[0171] Given the transformation function T(r), the grayscale value of each pixel after equalization can be directly calculated by the following formula:

[0172]

[0173] In formula (17), n is the sum of the pixels in the image, n j is the number of pixels of the current gray level, and T(r) is the given transformation function.

[0174] The image operation unit 12 sets the variable I as the grayscale expression of the image after histogram equalization, and uses the following function conversion relationship to convert the grayscale value of the image:

[0175]

[0176] In formula (18), I is the grayscale expression of the image after histogram equalization, and I′ is the transformed grayscale image. By formula (18), the transformed grayscale image I′ is obtained.

[0177] Define the Gaussian kernel function G(x, y) as:

[0178]

[0179] In formula (19), K is defined as a constant that satisfies the following equation:

[0180]

[0181] In formula (20), c is the Gaussian kernel scale constant, and the convolution operation on the original image I is expressed as:

[0182] I″=I*G (21)

[0183] In formula (21), the convolution result I″ contains the brightness between adjacent pixels, and I is the original image.

[0184] By comparing it with the central pixel and performing image contrast conversion based on the comparison scale, the de-shadowed image R(x, y) can be obtained:

[0185] R(x, y) = 255·I(x, y) r(x，y) (twenty two)

[0186] In formula (22), the exponent r(x, y) = I″(x, y) / I(x, y), so the brightness of the central pixel increases or decreases depending on whether r(x, y) is greater than 1 or less than 1 (surrounded by brighter pixels or darker pixels), achieving the effect of image deshading.

[0187] The image filtering unit 13 selects a 3×3 mean filter template and performs mean filtering on the image R(x, y). The pixel at the center of the image is the average value of the pixel values ​​of all points in the template. The operation is shown in the following formula:

[0188]

[0189] In formula (23), S is the template area, R(x, y) is the image after shadow removal, and g(x, y) is the image after mean filtering. According to formula (23), the mean filtered image g(x, y) can be obtained by calculation.

[0190] The present embodiment discloses a device for obtaining the primary particle size of a ternary positive electrode material, which enhances the primary particle SEM image by adopting a histogram equalization method; performs brightness conversion on the primary particle SEM image, performs convolution operation on the converted SEM image by using a Gaussian kernel, removes the particle shadows of the converted SEM image; and performs mean filtering on the de-shadowed image to remove image noise and small particle interference. The device for obtaining the primary particle size of a ternary positive electrode material provided in the present embodiment performs mean filtering on the de-shadowed image to obtain a mean-filtered image g(x, y), which is convenient for calculating the gradient map G(x, y), and the primary particle size distribution is accurately obtained.

[0191] Preferably, see Fig.17 , Fig.17 for Fig.15 , in which the calculation module 20 includes a definition unit 21, a derivation unit 22, a calculation unit 23 and a selection unit 24, wherein the definition unit 21 is used to define edge gradient directions in twelve directions; the derivation unit 22 is used to perform fractional derivative difference derivation on all edge directions of the edge gradient directions in the defined twelve directions to obtain fractional differential mask operators in the twelve directions; the calculation unit 23 is used to use the obtained fractional differential mask operator to perform a convolution operation on the mean filtered image g(x, y) to calculate the gradient map G(x, y); the selection unit 24 is used to select the maximum value of the twelve gradient calculations as the gradient value of each pixel, and finally obtain the gradient image of the primary particle SEM image.

[0192] The definition unit 21 calculates the image gradient map and first defines the edge gradient directions in twelve directions. The shape of the edge direction is as follows: Figure 8 shown.

[0193] The derivation unit 22 can be approximately expressed as follows through the fractional derivative difference form defined by GL:

[0194]

[0195] Extending to two dimensions, we get:

[0196]

[0197]

[0198] by Figure 8 Taking the edge direction as an example, using the fractional derivative difference form to deduce, we can get:

[0199]

[0200] In formula (27), v is the order (0<v<1), Figure 8 All the edge directions shown in the figure are derived by fractional derivative difference as above, and we can get Fig. 9 Fractional differential mask operators in twelve directions are shown.

[0201] The calculation unit 23 uses the above-mentioned fractional differential mask operator to convolve the mean filtered image g(x, y) to calculate the gradient map G(x, y):

[0202]

[0203] In formula (28), f iThere are twelve edge-oriented fractional-order mask operators. The direction with the largest gradient value can best represent the direction of change of the grayscale of the pixel in the image.

[0204] The selection unit 24 selects the maximum value of the twelve gradient calculations as the gradient value of each pixel, and finally obtains the gradient image MG (x, y) of the primary particle SEM image.

[0205] MG(x,y)=max{G(x,y)1,...,G(x,y) 12} (29)

[0206] In formula (29), G(x, y) represents the gradient map, and MG(x, y) is the gradient image.

[0207] The present embodiment provides a device for obtaining the primary particle size of a ternary positive electrode material, by defining the edge gradient directions in twelve directions; performing fractional derivative difference deduction on all edge directions of the defined edge gradient directions in the twelve directions, and obtaining the fractional differential mask operators in the twelve directions; using the obtained fractional differential mask operator to perform a convolution operation on the mean filtered image g(x, y), and calculating the gradient map G(x, y); selecting the maximum value of the twelve gradient calculations as the gradient value of each pixel, and finally obtaining the gradient image MG(x, y) of the primary particle SEM image. The device for obtaining the primary particle size of a ternary positive electrode material provided in the present embodiment selects the maximum value of the twelve gradient calculations as the gradient value of each pixel, and obtains the gradient image MG(x, y) of the primary particle SEM image, which is convenient for calculating the gradient image IMG(x, y) with particle foreground marks, and the primary particle size distribution is accurately obtained.

[0208] Further, see Fig.18 , Fig.18 for Fig.15 , in this embodiment, the marking module 30 includes a first operation unit 31, a second operation unit 32 and an acquisition unit 33, wherein the first operation unit 31 is used to give a square structure element O, and perform opening and closing operations on the gradient image MG (x, y) with the structure element O, so as to smooth the gradient image as much as possible, eliminate the interference of small impurity edges, retain the complete edges of particles, and obtain a smooth gradient image; the second operation unit 32 is used to binarize the smoothed gradient image, find the possible location of the primary particles, and then perform closing operations on the binary image with the structure element O to fill the binary holes and obtain the preliminary foreground mark of the particles; the acquisition unit 33 is used to take the central area block of the preliminary foreground mark to obtain the particle foreground mark image IG (x, y).

[0209] The first operation unit 31 performs foreground marking on the primary particles in the primary particle SEM image, and gives a square structure element O, such as Fig.10 As shown:

[0210] The gradient image MG(x, y) is opened with the structure element O, and then closed to smooth the gradient image as much as possible, eliminate the interference of small impurities and retain the complete edges of the particles to obtain a smooth gradient image.

[0211] The second operation unit 32 performs binarization processing on the smoothed gradient image MG (x, y) to obtain a binary image and find the possible location of the primary particles; then performs a closing operation on the binary image with the structure element O to fill the binary holes and obtain a preliminary foreground mark of the particles.

[0212] The acquisition unit 33 takes the central area block of the preliminary foreground mark in the binary image to obtain the particle foreground mark image IG (x, y).

[0213] The present embodiment provides a device for obtaining the primary particle size of a ternary positive electrode material. By giving a square structure element O, the structure element O is used to perform an opening operation and a closing operation on the gradient image MG (x, y), so as to smooth the gradient image as much as possible, eliminate the interference of small impurity edges, retain the complete edges of the particles, and obtain a smooth gradient image direction; the smoothed gradient image is binarized to find the possible location of the primary particles, and then the structure element O is used to perform a closing operation on the binary image to fill the binary holes and obtain the preliminary foreground mark of the particles; the central area block of the preliminary foreground mark is taken to obtain the particle foreground mark image IG (x, y). The device for obtaining the primary particle size of a ternary positive electrode material provided in the present embodiment takes the central area block of the preliminary foreground mark to obtain the particle foreground mark image IG (x, y), which is convenient for calculating the gradient image IMG (x, y) with the particle foreground mark, and the primary particle size distribution is accurately obtained.

[0214] Preferably, see Fig.19 , Fig.19 for Fig.15, in this embodiment, the acquisition module 40 includes a segmentation unit 41, a first processing unit 42, an analysis unit 43, a second processing unit 44 and a statistical unit 45, wherein the segmentation unit 41 is used to combine the gradient image MG (x, y) with the particle foreground mark image IMG (x, y) to obtain the gradient image IMG (x, y) with the particle foreground mark, and use the watershed algorithm to segment the gradient image IMG (x, y) with the particle foreground mark to obtain the watershed line WS (x, y), that is, the edge contour of the particle; the first processing unit 42 is used to analyze the segmentation result WS (x, y) area threshold processing is performed to remove under-division and excessive interference; an analysis unit 43 is used to perform connected domain statistical analysis on the watershed WS(x, y) to obtain the connected domain BW(i) and its corresponding area size S(i), i=1, 2, 3...n, n is the statistical number of connected domains, that is, the number of primary particles; a second processing unit 44 is used to give area thresholds thresholdH, thresholdL, and perform area threshold processing on the connected domain; a statistical unit 45 is used to count the area size and distribution of the connected domain BW′(i) after area threshold processing, that is, to obtain the size distribution of the primary particles of the ternary positive electrode material during the sintering process.

[0215] The segmentation unit 41 combines the gradient image MG(x, y) with the particle foreground mark image IMG(x, y) to obtain the gradient image IMG(x, y) with the particle foreground mark, and uses the watershed algorithm to segment the gradient image IMG(x, y) with the particle foreground mark to obtain the watershed line WS(x, y), that is, the edge contour of the particles.

[0216] Since over-segmentation, under-segmentation and high adhesion of primary particles during the sintering process are inevitable during watershed segmentation, it is necessary for the first processing unit 42 to perform area threshold processing on the segmentation result WS(x, y) to remove under-segmentation and over-interference.

[0217] The analysis unit 43 performs a connected domain statistical analysis on the watershed line WS(x, y) to obtain a connected domain BW(i) and its corresponding area size S(i), i=1, 2, 3...n, n is the statistical number of connected domains, that is, the number of primary particles.

[0218] The second processing unit 44 sets area thresholds thresholdH and thresholdL, and performs area threshold processing on the connected domain:

[0219]

[0220] In formula (30), BW′(i) is the connected domain after area threshold processing, thresholdH, thresholdL are given area thresholds, and S(i) is the area size corresponding to the connected domain.

[0221] The statistical unit 45 counts the area size and distribution of the connected domain BW′(i) after the area threshold processing, that is, obtains the size distribution of the primary particles of the ternary positive electrode material during the sintering process.

[0222] The present embodiment provides a device for obtaining the size of primary particles of a ternary positive electrode material. The device combines a gradient image MG(x, y) with a particle foreground mark image IMG(x, y) to obtain a gradient image IMG(x, y) with a particle foreground mark. The gradient image IMG(x, y) with a particle foreground mark is segmented using a watershed algorithm to obtain a watershed line WS(x, y), i.e., the edge contour of the particle. The segmentation result WS(x, y) is processed by an area threshold to remove under-scores and excessive interference. The watershed line WS(x, y) is statistically analyzed for a connected domain to obtain a connected domain BW(i) and its corresponding area size S(i), where i=1, 2, 3...n, and n is the statistical number of connected domains, i.e., the number of primary particles. Area thresholds thresholdH and thresholdL are given, and area threshold processing is performed on the connected domain. The area size and distribution of the connected domain BW′(i) after the area threshold processing are statistically analyzed to obtain the size distribution of primary particles of the ternary positive electrode material during the sintering process. The device for obtaining the primary particle size of the ternary positive electrode material provided in this embodiment obtains the size of the primary particles of the ternary positive electrode material during the sintering process from the SEM image, filling the gap that the microscopic indicators of the ternary positive electrode material during the sintering process are difficult to obtain; a fractional-order differential operator based on twelve edge directions is designed, which can effectively calculate the most accurate gradient direction in the image and realize accurate segmentation of particles.

[0223] Although preferred embodiments of the present invention have been described, additional changes and modifications may be made to these embodiments by those skilled in the art once the basic inventive concepts are known. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention. Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for obtaining the primary particle size of a ternary positive electrode material, characterized in that: The following steps are involved: Preprocess the primary particle SEM image to remove image noise and small particle interference; In view of the uncertainty of the edge direction of primary particles, the image gradient is calculated using the fractional gradient operator of twelve edge directions. The maximum value of the twelve edge gradients is taken to find the most correct gradient direction in the image gradient map to obtain the image gradient map. Perform morphological calculation on the image gradient map obtained by calculation, find the potential position of the primary particle, mark it, and obtain the foreground marking map; The image gradient map is combined with the foreground marker map to obtain the foreground marker gradient map, and the foreground marker gradient map is subjected to watershed segmentation calculation and area threshold processing to obtain a particle segmentation result. Finally, the result is subjected to connected domain statistical analysis to obtain a particle size distribution. The step of preprocessing the primary particle SEM image to remove image noise and small particle interference includes: The histogram equalization method was used to enhance the primary particle SEM image; Perform brightness conversion on the primary particle SEM image, perform convolution operation on the converted SEM image using a Gaussian kernel, and remove particle shadows from the converted SEM image; Perform mean filtering on the de-shadowed image to remove image noise and small particle interference; In view of the uncertainty of the edge direction of the primary particle, the image gradient is calculated by using the fractional gradient operator of twelve edge directions, the maximum value of the twelve edge gradients is taken, and the most correct gradient direction in the image gradient map is found. The steps of obtaining the image gradient map include: Define twelve directions of edge gradient directions; Perform fractional derivative difference derivation on all edge directions of the edge gradient directions in the defined twelve directions to obtain fractional differential mask operators in the twelve directions; Use the obtained fractional differential mask operator to filter the image after mean filtering. Perform convolution operation and calculate the gradient map ; The maximum value of the twelve gradient calculations is selected as the gradient value of each pixel, and finally the gradient image of the particle SEM image is obtained. ; The step of performing morphological calculation on the image gradient map obtained by calculation, finding the potential position of the primary particle, and marking it to obtain the foreground marking map comprises: Given a square structure element O, use the structure element O to transform the gradient image Perform opening and closing operations to smooth the gradient map as much as possible, eliminate the interference of small impurity edges, retain the complete edges of particles, and obtain a smooth gradient map; Binarize the smoothed gradient image to find the possible location of the primary particles, then close the binary image with the structure element O to fill the binary holes and obtain the preliminary foreground mark of the particles. Take the central area block of the preliminary foreground mark to obtain the particle foreground mark image .

2. The method for obtaining the primary particle size of the ternary positive electrode material according to claim 1, characterized in that: The step of performing morphological calculation on the image gradient map obtained by calculation, finding the potential position of the primary particle, and marking it to obtain the foreground marking map comprises: The gradient image Label the image with the particle foreground Combined, we get a gradient image with particle foreground markers , for the gradient image with particle foreground marker Use the watershed algorithm to segment and get the watershed , i.e., the edge profile of the particle; The segmentation result Perform area threshold processing to remove under-scores and excessive interference; The waterline Perform statistical analysis on the connected domain and obtain the connected domain And its corresponding area size , , is the number of connected domains, that is, the number of primary particles; Given area threshold , perform area threshold processing on the connected domain; Statistical area of ​​connected domains after threshold processing The area size and distribution of the ternary positive electrode material can be obtained by the size distribution of the primary particles during the sintering process.

3. A device for obtaining the primary particle size of a ternary positive electrode material, characterized in that: include: A preprocessing module (10) is used to preprocess the primary particle SEM image to remove image noise and small particle interference; A calculation module (20) is used to calculate the image gradient using a fractional gradient operator of twelve edge directions in view of the uncertainty of the edge direction of the primary particles, take the maximum value of the twelve edge gradients, find the most correct gradient direction in the image gradient map, and obtain the image gradient map; A marking module (30) is used to perform morphological calculation on the image gradient map obtained by calculation, find the potential position of the primary particle, and mark it to obtain a foreground marking map; An acquisition module (40) is used to combine the image gradient map with the foreground label map to obtain the foreground label gradient map, perform watershed segmentation calculation on the foreground label gradient map, perform area threshold processing on the foreground label gradient map, obtain a primary particle segmentation result, and finally perform connected domain statistical analysis on the result to obtain a primary particle size distribution; The preprocessing module (10) comprises: An image enhancement unit (11) is used to enhance the primary particle SEM image using a histogram equalization method; An image operation unit (12) is used to perform brightness conversion on a primary particle SEM image, perform convolution operation on the converted SEM image using a Gaussian kernel, and remove particle shadows from the converted SEM image; An image filtering unit (13) performs mean filtering on the de-shadowed image to remove image noise and small particle interference; The calculation module (20) comprises: A definition unit (21), used to define twelve directions of edge gradient directions; A derivation unit (22) is used to perform fractional-order derivative difference derivation on all edge directions in the edge gradient directions of the defined twelve directions, so as to obtain fractional-order differential mask operators in the twelve directions; A computing unit (23) is used to use the obtained fractional differential mask operator to perform mean filtering on the image after mean filtering. Perform convolution operation to calculate the gradient map ; A selection unit (24) is used to select the maximum value of the twelve gradient calculations as the gradient value of each pixel, and finally obtain a gradient image of the primary particle SEM image. ; The marking module (30) comprises: The first operation unit (31) is used to give a square structure element O and use the structure element O to calculate the gradient image. Perform opening and closing operations to smooth the gradient map as much as possible, eliminate the interference of small impurity edges, retain the complete edges of particles, and obtain a smooth gradient map; The second operation unit (32) is used to perform a binary process on the smoothed gradient image to find the possible location of the primary particles, and then perform a closing operation on the binary image with a structure element O to fill the binary holes and obtain a preliminary foreground mark of the particles; The acquisition unit (33) is used to obtain the central area block of the preliminary foreground mark to obtain a particle foreground mark image. .

4. The device for obtaining the primary particle size of the ternary positive electrode material according to claim 3, characterized in that: The acquisition module (40) comprises: A segmentation unit (41) is used to segment the gradient image Label the image with the particle foreground Combined, we get a gradient image with particle foreground markers , for the gradient image with particle foreground marker Use the watershed algorithm to segment and get the watershed , i.e., the edge profile of the particle; The first processing unit (42) is used to process the segmentation result Perform area threshold processing to remove under-scores and excessive interference; An analysis unit (43) is used to analyze the watershed Perform statistical analysis on the connected domain and obtain the connected domain And its corresponding area size , , is the number of connected domains, that is, the number of primary particles; The second processing unit (44) is used to set an area threshold , perform area threshold processing on the connected domain; Statistics unit (45), used to count the connected domains after area threshold processing The area size and distribution of the ternary positive electrode material can be obtained by the size distribution of the primary particles during the sintering process.

Images