Recognition and statistics method for high-temperature alloy additive manufacturing microstructure

Abstract

The invention provides an identification and statistical method for a high-temperature alloy additive manufacturing microstructure, which comprises the following steps of: S1, preparing a high-temperature alloy sample piece by adopting additive manufacturing, and obtaining a microstructure image; s2, detecting the microstructure of the microstructure image by adopting an image segmentation algorithm to obtain a segmented image; s3, extracting the characteristics of the microstructure of the segmented image; and S4, carrying out classification identification and statistics according to the characteristics of the microstructure. According to the identification and statistical method for the high-temperature alloy additive manufacturing microstructure, image processing of cellular structures and columnar crystal structures of typical characteristic structures of high-temperature alloy additive manufacturing can be achieved, and then classification identification and data statistics of the structures in the images are achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of high-temperature alloy additive manufacturing, and specifically refers to an identification and statistical method for the microstructure of high-temperature alloy additive manufacturing. Background Art

[0002] High-temperature alloys based on nickel, iron and cobalt are called "superalloys" because of their good oxidation resistance, thermal corrosion resistance and high strength at high temperatures. They are widely used in aviation, aerospace and energy fields. Compared with the traditional preparation method of high-temperature alloys, additive manufacturing preparation technology can directly transform the digital design model of high-temperature alloys into entities, with the advantages of high manufacturing freedom, short processing cycle, multi-level and multi-material combination, and complex shape precision manufacturing. In the additive manufacturing process of high-temperature alloys, some microstructures different from those of traditional processes will be produced in the high-temperature alloys. Specifically, the additive manufacturing process is accompanied by rapid heating and cooling and complex thermal cycles, which makes the main microstructure of the high-temperature alloy form a fine columnar crystal structure epitaxially grown along the bottom of the molten pool and a cellular structure with dislocation entanglement. The change in the microstructure has a direct impact on the performance of the high-temperature alloy. The study of ZG Zhu et al. shows that a cellular structure of a certain size and fraction is beneficial to the strength improvement of the additive high-temperature alloy ("Hierarchical microstructure and strengthening mechanisms of a CoCrFeNiMn high entropy alloy additively manufactured by selective laser melting", Scripta Materialia 154 (2018) 20-24).

[0003] At present, there are two main statistical methods for the microstructure of additively manufactured high-temperature alloys. One is to manually identify and classify and calibrate the statistics one by one, and the other is to use high-end testing methods such as electron backscatter diffraction EBSD to assist in the characterization of the microstructure. However, manual identification and statistics are inefficient and have large precision errors, while high-end testing methods such as EBSD are expensive and cannot automatically realize the classification, identification and calibration of cellular and columnar structures.

[0004] Therefore, there is an urgent need to provide a new identification and statistical method for the microstructure of high-temperature alloy additive manufacturing, which can improve the defects of existing statistical methods such as low calibration efficiency, large precision error, and inability to automatically realize the classification, identification and calibration statistics of cellular and columnar crystal structures. Summary of the invention

[0005] The purpose of the present invention is to provide a method for identifying and counting the microstructure of high-temperature alloy additive manufacturing, which can improve the defects of existing statistical methods such as low calibration efficiency, large precision error and inability to automatically realize the classification, identification and calibration statistics of cellular structure and columnar crystal structure.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method for identifying and counting microstructures of high-temperature alloy additive manufacturing, comprising the following steps:

[0008] S1. Prepare high-temperature alloy samples by additive manufacturing and obtain microstructure images;

[0009] S2. Using an image segmentation algorithm to detect the microscopic tissue structure of the microscopic tissue image to obtain a segmented image;

[0010] S3. extracting the characteristics of the microstructure of the segmented image;

[0011] S4. Classify and identify the microstructure according to its characteristics and perform statistics.

[0012] Optionally, S1. Prepare a high temperature alloy sample by additive manufacturing and obtain a microstructure image, including:

[0013] S1-1. The high temperature alloy sample is prepared by at least one of laser cladding, electron beam selective melting, laser selective melting or electron beam fuse deposition;

[0014] S1-2. Using metallographic sample preparation, obtain the sample to be tested of the high temperature alloy sample;

[0015] S1-3. Use a scanning electron microscope to collect images of the sample to be tested to obtain the microstructure image.

[0016] Optionally, S2. using an image segmentation algorithm to detect the microscopic tissue structure of the microscopic tissue image to obtain a segmented image, comprising:

[0017] S2-1. Using image processing software, the gray value of the tissue boundary in the microscopic tissue image is measured, and the gray value of the tissue boundary is set as the separation threshold t, wherein the tissue boundary is a cellular tissue boundary or a columnar crystal tissue boundary;

[0018] S2-2. Image preprocessing is performed using an image cropping algorithm to remove the scanning imaging information at the bottom of the microscopic tissue image;

[0019] S2-3. Use an image segmentation algorithm to screen areas where the grayscale value of pixels is greater than the separation threshold t as tissue boundaries, perform image segmentation based on the tissue boundaries, obtain a segmented image, and achieve segmentation of single cellular tissue or columnar crystal tissue.

[0020] Optionally, S2-2. Using an image cropping algorithm to perform image preprocessing to remove the scanning imaging information at the bottom of the microscopic tissue image, including:

[0021] The image[] function in Python is used for image preprocessing. The formula is as follows:

[0022] image=image[0:image.shape[0]–G0,:,:]

[0023] Among them, the original shape of the microstructure image is defined by the number of rows, columns and channels, image.shape[0] means obtaining the original number of rows of the microstructure image, image.shape[0]-G0 means the original number of rows minus the G0 row, G0 is the number of rows of scanned imaging information, the first colon means obtaining all the columns, the second colon means obtaining all the channels, and the output result of the image[] function is to return the cropped image, which contains all the columns and channels of the original image, and the bottom G0 row is removed.

[0024] Optionally, S2-3. Using an image segmentation algorithm, screening the area where the pixel gray value is greater than the separation threshold t as the tissue boundary, performing image segmentation according to the tissue boundary, obtaining a segmented image, and realizing the segmentation of a single cellular tissue or columnar crystal tissue, including:

[0025] Image segmentation is performed using the np.where() function in Python. The formula is as follows:

[0026] image=np.where(image>t,0,image)

[0027] Among them, image>t is the reference condition, that is, the grayscale value of the pixel point is greater than the segmentation threshold t. When the reference condition is met, the output result of the function is 0, that is, the grayscale value of the pixel point is replaced by 0, so as to remove the tissue boundary in the microscopic tissue image; when the reference condition is not met, the output result of the function is the grayscale value of the pixel point itself.

[0028] Optionally, S3. extracting features of the microscopic tissue structure of the segmented image comprises:

[0029] S3-1. Calculate the area ratio X of each microstructure in the segmented image. The formula is:

[0030] X=area / (crop_image.shape[0]*crop_image.shape[1])

[0031] Among them, area represents the number of pixels in the current connected area, crop_image.shape[0]*crop_image.shape[1] is the area of ​​the rectangle surrounding the current connected area, crop_image.shape[0] represents the height of the rectangle surrounding the current connected area, and crop_image.shape[1] represents the width of the rectangle surrounding the current connected area;

[0032] S3-2. Calculate the axial ratio Y of each microscopic tissue structure in the segmented image, the formula is:

[0033] Y=min(crop_image.shape[0],crop_image.shape[1]) / max(crop_image.shape[0],crop_imag e.shape[1]), min represents the minimum value and max represents the maximum value.

[0034] Optionally, before S3. extracting the features of the microstructure, the method further includes using S3'. an image erosion algorithm to separate the subtle connections between the cellular tissue or columnar crystal tissue and the tissue boundary in the microstructure image.

[0035] Optionally, S3'. An image erosion algorithm is used to separate the fine connection between the cellular tissue or columnar crystal tissue and the tissue boundary in the microscopic tissue structure image, comprising:

[0036] Use the erode() function in Python to erode the image. The formula is as follows:

[0037] dst=cv2.erode(src,kernel,iterations)

[0038] Among them, dst represents the output image after the corrosion operation, and its type is consistent with src; src represents the input image, which is a two-dimensional array representing the image pixels; kernel represents the convolution kernel, which is a two-dimensional array that defines the neighborhood of the corrosion operation, kernel = np.ones((a,b), np.uint8), np.uint8 represents all pixel values ​​in the range of unsigned 8-bit integers, ranging from 0 to 255, (a,b) represents a two-dimensional array of a×b; iterations represents the number of iterations, and the result of the function output is to convolve the convolution kernel kernel with the input image to obtain the minimum value of the pixel points of the convolution kernel kernel, and use this minimum value to replace the pixel value of the reference point, and return the image after completing the number of iterations.

[0039] Optionally, before S3. extracting the features of the microstructure, it also includes using S3". image enhancement algorithm to increase the clarity of the microstructure image after image corrosion. The S3". image enhancement algorithm uses the ImageEnhance.Sharpness() method in Python to create an image enhancement object, and calls the enhance() method to specify enhancement parameters.

[0040] Optionally, S4. classifying and identifying the microstructure according to its characteristics and performing statistics, including:

[0041] S4-1. Determine whether the area ratio X and the axis ratio Y of each of the microstructures are greater than a threshold value. If X>X0, Y>Y0 are satisfied, the microstructure is identified as a cellular structure, X0 is the threshold of the area ratio X, and Y0 is the threshold of the axis ratio Y; if X>X0, Y>Y0 are not satisfied, the microstructure is identified as a columnar crystal structure;

[0042] S4-2. Count the data information of the cellular tissue and the data information of the columnar crystal tissue, wherein the data information includes at least one of the number of tissues, the diameter of the circumscribed circle of each cellular tissue, or the length of the major and minor axes of each columnar crystal, and the statistics are all performed based on pixel units.

[0043] In summary, the present invention has at least one of the following beneficial effects:

[0044] The method for identifying and counting the microstructure of additive manufacturing of high-temperature alloys provided by the present invention processes the high-temperature alloy additive tissue images to achieve tissue segmentation, and classifies and performs data statistics of cellular tissues and columnar crystal tissues. It can realize rapid processing and data statistics of a large number of additive tissue images with higher accuracy. The data statistics are performed based on pixels, and the data contain multiple information such as tissue distribution and length, which is suitable for establishing the additive process-structure-performance relationship. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 A microstructure image of a high-temperature alloy sample obtained by a partial scanning electron microscope according to a specific embodiment of the present invention;

[0047] Figure 2 For the corresponding Figure 1 Cropped, segmented, eroded, and sharpened images;

[0048] Figure 3 For the corresponding Figure 1 (a) Data information of cellular structure and columnar crystal structure in the image;

[0049] Figure 4 For the corresponding Figure 1 (b) Data information of cellular structure and columnar crystal structure in the image. DETAILED DESCRIPTION

[0050] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0051] An embodiment of the present invention provides a method for identifying and counting microstructures of high-temperature alloy additive manufacturing, comprising the following steps:

[0052] S1. Prepare high-temperature alloy samples using additive manufacturing and obtain microstructure images, including:

[0053] S1-1. The high temperature alloy sample is prepared by at least one of laser cladding, electron beam selective melting, laser selective melting or electron beam fuse deposition;

[0054] S1-2. Using metallographic sample preparation, obtain the sample to be tested of the high temperature alloy sample;

[0055] S1-3. Use a scanning electron microscope to collect images of the sample to be tested to obtain the microstructure image.

[0056] In a specific embodiment of the present invention, laser selective melting (equipment name EOS M 290) is used to prepare nickel-based high-temperature alloy additive samples, and metallographic sample preparation processes such as cutting, grinding, polishing, and corrosion are completed. The processed samples to be tested are observed for their organizational morphology using a scanning electron microscope. The image acquisition mode is a secondary electron method 2000 times. Each sample is uniformly collected on the observation surface. No less than 30 images are collected, and multiple images can be used to calculate the average particle size and observe the particle size distribution. Figure 1 These are some original images of high-temperature alloy additive manufacturing obtained by scanning electron microscope, that is, the microstructure image obtained.

[0057] S2. Using an image segmentation algorithm to detect the microscopic tissue structure of the microscopic tissue image to obtain a segmented image, including:

[0058] S2-1. Using image processing software, the gray value of the tissue boundary in the microscopic tissue image is measured, and the gray value of the tissue boundary is set as the separation threshold t, and the tissue boundary is a cellular tissue boundary or a columnar crystal tissue boundary. In a specific embodiment of the present invention, Photoshop is used to measure the gray value of the subgrain boundary between adjacent cellular tissues, that is, the tissue boundary, and this is used as the threshold for subsequent image segmentation. In this example, the threshold for subsequent image segmentation is set to 140 based on the gray value of the tissue boundary measured according to the image information.

[0059] S2-2. Using an image cropping algorithm to perform image preprocessing to remove the scanning imaging information at the bottom of the microscopic tissue image, including:

[0060] Use the image[] function in Python to crop the image. The formula is as follows:

[0061] image=image[0:image.shape[0]–G0,:,:]

[0062] The original shape of the microscopic tissue image is defined by the number of rows, columns and channels. image.shape[0] indicates obtaining the original number of rows of the microscopic tissue image. image.shape[0]-G0 indicates obtaining the original number of rows minus the G0 row. The G0 row is the number of rows of the scanned imaging information. The first colon indicates obtaining all the columns. The second colon indicates obtaining all the channels. The output result of the image[] function is to return the cropped image, which includes all the columns and channels of the original image. The bottom G0 row is removed. In a specific embodiment of the present application, the G0 row is set to 70 rows.

[0063] S2-3. Using an image segmentation algorithm, screening the area where the pixel gray value is greater than the separation threshold t as the tissue boundary, performing image segmentation according to the tissue boundary, obtaining a segmented image, and realizing the segmentation of a single cellular tissue or columnar crystal tissue, including:

[0064] The np.where() function in Python is used for image segmentation. The formula is as follows:

[0065] image=np.where(image>t,0,image)

[0066] Among them, image>t is the reference condition, that is, the grayscale value of the pixel point is greater than the segmentation threshold t. When the reference condition is met, the output result of the function is 0, that is, the grayscale value of the pixel point is replaced by 0, so as to remove the tissue boundary in the microscopic tissue image; when the reference condition is not met, the output result of the function is the grayscale value of the pixel point itself.

[0067] In a specific embodiment of the present application, the grayscale value of the subgrain boundary between the cellular tissues measured by reference S2-1 is set to 140. The np.where function is used to traverse the pixels of the preprocessed microstructure image. When the grayscale value is greater than 140, the pixel is assigned a value of 0 to remove the cellular tissue or columnar crystal tissue boundary. When the grayscale value is not greater than 140, the output result of the function is the grayscale value of the pixel itself, thereby realizing the segmentation of a single cellular tissue or columnar crystal tissue.

[0068] S3'. An image corrosion algorithm to separate the fine connection between the cellular tissue or columnar crystal tissue and the tissue boundary in the microscopic tissue structure image, comprising:

[0069] Use the erode() function in Python to erode the image. The formula is as follows:

[0070] dst=cv2.erode(src,kernel,iterations)

[0071] Among them, dst represents the image output after the corrosion operation, and its type is consistent with src; src represents the input image, which is usually a two-dimensional array representing the image pixels, and its data type is usually uint8, that is, an unsigned 8-bit integer, ranging from 0 to 255; kernel represents the convolution kernel, which is a two-dimensional array that defines the neighborhood of the corrosion operation, kernel = np.ones((a,b), np.uint8), np.uint8 represents all pixel values ​​within the range of unsigned 8-bit integers, ranging from 0 to 255, (a,b) represents a two-dimensional array of a×b; iterations represents the number of iterations, and the result of the function output is to use the convolution kernel kernel to perform convolution calculation with the input image, obtain the minimum value of the pixel point of the convolution kernel kernel, and use this minimum value to replace the pixel value of the reference point, and return the image after completing the number of iterations. In a specific embodiment of the present invention, kernel = np.ones((3,3), np.uint8), and the number of iterations is 1.

[0072] It also includes S3". Image enhancement algorithm, which increases the clarity of the microstructure image after image corrosion. S3". Image enhancement algorithm uses the ImageEnhance.Sharpness() method in the Python algorithm to create an image enhancement object, and calls the enhance() method to specify the enhancement parameters. Specifically, the formula is:

[0073] image=Image.fromarray(image)

[0074] image_enhance=ImageEnhance.Sharpness(image)

[0075] image=image_enhance.enhance(6)

[0076] image = np.array(image)

[0077] The first step is to convert the NumPy array into a PIL Image object so that you can use PIL to provide image processing functions;

[0078] The second step is to create a Sharpness instance under the ImageEnhance module to adjust the image sharpness. Image is the Image object passed to the Sharpness constructor.

[0079] The third step is to call the enhance() method to enhance the image sharpness. 6 is the degree of increase. The larger the value, the more prominent the image edge is, and it looks clearer, which is convenient for pixel statistics.

[0080] The fourth step is to convert the processed PIL Image object into a NumPy array to facilitate subsequent pixel statistics. Figure 2 For the corresponding Figure 1 Image after cropping, segmentation, erosion, and sharpening.

[0081] S3. Extracting features of the microscopic tissue structure of the segmented image, including:

[0082] S3-1. Calculate the area ratio X of each microstructure in the segmented image. The formula is:

[0083] X=area / (crop_image.shape[0]*crop_image.shape[1])

[0084] area represents the number of pixels in the current connected area, crop_image.shape[0]*crop_image.shape[1] represents the area of ​​the rectangle surrounding the current connected area, crop_image.shape[0] represents the height of the rectangle surrounding the current connected area, and crop_image.shape[1] represents the width of the rectangle surrounding the current connected area;

[0085] S3-2. Calculate the axial ratio Y of each microscopic tissue structure in the segmented image, the formula is:

[0086] Y=min(crop_image.shape[0],crop_image.shape[1]) / max(crop_image.shape[0],crop_imag e.shape[1]), min represents the minimum value and max represents the maximum value.

[0087] S4. Classify and identify the microstructure characteristics and perform statistics, including:

[0088] S4-1. Determine whether the area ratio X and the axis ratio Y of each of the microstructures are greater than a threshold value. If X>X0, Y>Y0 are satisfied, the microstructure is identified as a cellular structure, X0 is the threshold of the area ratio X, and Y0 is the threshold of the axis ratio Y; if X>X0, Y>Y0 are not satisfied, the microstructure is identified as a columnar crystal structure;

[0089] S4-2. Count the data information of the cellular tissue and the data information of the columnar crystal tissue, wherein the data information includes at least one of the number of tissues, the diameter of the circumscribed circle of each cellular tissue, or the length of the major and minor axes of each columnar crystal, and the statistics are all performed based on pixel units.

[0090] Specifically, in step S4, tissues that simultaneously satisfy a certain area ratio X and axis ratio Y are classified as cellular tissues. In a specific embodiment of the present application, segmented tissue regions that simultaneously satisfy area ratio X>0.3 and axis ratio Y>0.3 are determined to be cellular tissues, and others are classified as columnar crystal tissues.

[0091] The number of cellular tissues, the average cell body size (the sum of the pixel area of ​​cellular tissues / the number), and the minimum circumscribed circle diameter of each cellular tissue were counted; the number of columnar crystal tissues, the average columnar crystal size (the sum of the pixel area of ​​columnar crystal tissues / the number), and the length of the major and minor axes of each columnar crystal were counted. Correspondingly, the statistics were obtained Figure 1 In (a), the number of cellular tissues is 175, the number of columnar crystal tissues is 210, the average cell size is 251, the average columnar crystal size is 1650, and the minimum circumscribed circle diameter of each cellular tissue is listed as follows: Figure 3 As shown in (a), the major and minor axes of some columnar crystals are listed as follows Figure 3 (b) is shown. Figure 1 In (b), the number of cellular tissues is 452, the number of columnar crystal tissues is 124, the average cell size is 527, the average columnar crystal size is 931, and the minimum circumscribed circle diameter of some cellular tissues is listed as follows: Figure 4 As shown in (a), the major and minor axes of some columnar crystals are listed as follows Figure 4 The statistical information is based on pixel data and can be converted later. In this example, the magnification of the image acquisition is 2000 times. By converting 37 pixels = 1 μm, the μm scale data of the cellular structure and columnar crystal structure can be obtained.

[0092] The above is a detailed introduction to the identification and statistical method of the microstructure of the high-temperature alloy additive manufacturing provided by the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present invention, the present invention can also be improved and modified, and these improvements and modifications also fall within the scope of protection of the claims of the present invention.

Claims

1. A method for identifying and counting microstructures of high-temperature alloy additive manufacturing, characterized in that: The steps include: S1. Prepare high-temperature alloy samples by additive manufacturing and obtain microstructure images; S2. Using an image segmentation algorithm to detect the microscopic tissue structure of the microscopic tissue image to obtain a segmented image; S3. extracting the characteristics of the microstructure of the segmented image; S4. Classify and identify the microstructure according to its characteristics and perform statistics.

2. The method for identifying and counting microstructures of high-temperature alloy additive manufacturing according to claim 1, characterized in that: S1. Prepare high-temperature alloy samples using additive manufacturing and obtain microstructure images, including: S1-1. The high temperature alloy sample is prepared by at least one of laser cladding, electron beam selective melting, laser selective melting or electron beam fuse deposition; S1-2. Using metallographic sample preparation, obtain the sample to be tested of the high temperature alloy sample; S1-3. Use a scanning electron microscope to collect images of the sample to be tested to obtain the microstructure image.

3. The method for identifying and counting microstructures of high-temperature alloy additive manufacturing according to claim 1, characterized in that: S2. Using an image segmentation algorithm to detect the microscopic tissue structure of the microscopic tissue image to obtain a segmented image, including: S2-1. Using image processing software, the gray value of the tissue boundary in the microscopic tissue image is measured, and the gray value of the tissue boundary is set as the separation threshold t, wherein the tissue boundary is a cellular tissue boundary or a columnar crystal tissue boundary; S2-2. Image preprocessing is performed using an image cropping algorithm to remove the scanning imaging information at the bottom of the microscopic tissue image; S2-3. Use an image segmentation algorithm to screen areas where the grayscale value of pixels is greater than the separation threshold t as tissue boundaries, perform image segmentation based on the tissue boundaries, obtain a segmented image, and achieve segmentation of single cellular tissue or columnar crystal tissue.

4. The method for identifying and counting microstructures of high-temperature alloy additive manufacturing according to claim 3, characterized in that: S2-2. Using an image cropping algorithm to perform image preprocessing to remove the scanning imaging information at the bottom of the microscopic tissue image, including: The image[] function in Python is used for image preprocessing. The formula is as follows: image=image[0:image.shape[0]–G0,:,:] Among them, the original shape of the microstructure image is defined by the number of rows, columns and channels, image.shape[0] means obtaining the original number of rows of the microstructure image, image.shape[0]-G0 means the original number of rows minus the G0 row, G0 is the number of rows of scanned imaging information, the first colon means obtaining all the columns, the second colon means obtaining all the channels, and the output result of the image[] function is to return the cropped image, which contains all the columns and channels of the original image, and the bottom G0 row is removed.

5. The method for identifying and counting microstructures of high-temperature alloy additive manufacturing according to claim 3, characterized in that: S2-3. Using an image segmentation algorithm, screening the area where the pixel gray value is greater than the separation threshold t as the tissue boundary, performing image segmentation according to the tissue boundary, obtaining a segmented image, and realizing the segmentation of a single cellular tissue or columnar crystal tissue, including: Image segmentation is performed using the np.where() function in Python. The formula is as follows: image=np.where(image>t,0,image) Among them, image>t is the reference condition, that is, the grayscale value of the pixel point is greater than the segmentation threshold t. When the reference condition is met, the output result of the function is 0, that is, the grayscale value of the pixel point is replaced by 0, so as to remove the tissue boundary in the microscopic tissue image; when the reference condition is not met, the output result of the function is the grayscale value of the pixel point itself.

6. The method for identifying and counting microstructures of high-temperature alloy additive manufacturing according to claim 1, characterized in that: S3. Extracting features of the microscopic tissue structure of the segmented image, including: S3-1. Calculate the area ratio X of each microstructure in the segmented image. The formula is: X=area / (crop_image.shape[0]*crop_image.shape[1]) Among them, area represents the number of pixels in the current connected area, crop_image.shape[0]*crop_image.shape[1] is the area of ​​the rectangle surrounding the current connected area, crop_image.shape[0] represents the height of the rectangle surrounding the current connected area, and crop_image.shape[1] represents the width of the rectangle surrounding the current connected area; S3-2. Calculate the axial ratio Y of each microscopic tissue structure in the segmented image, the formula is: Y=min(crop_image.shape[0],crop_image.shape[1]) / max(crop_image.shape[0],crop_imag e.shape[1]), min represents the minimum value and max represents the maximum value.

7. The method for identifying and counting microstructures of high-temperature alloy additive manufacturing according to claim 6, characterized in that: S3. Before extracting the features of the microstructure, it also includes using S3'. Image corrosion algorithm to separate the cellular tissue or columnar crystal tissue and the subtle connection of the tissue boundary in the microstructure image.

8. The method for identifying and counting microstructures of high-temperature alloy additive manufacturing according to claim 7, characterized in that: S3'. An image corrosion algorithm to separate the fine connection between the cellular tissue or columnar crystal tissue and the tissue boundary in the microscopic tissue structure image, comprising: Use the erode() function in Python to erode the image. The formula is as follows: dst=cv2.erode(src,kernel,iterations) Among them, dst represents the output image after the corrosion operation, and its type is consistent with src; src represents the input image, which is a two-dimensional array representing the image pixels; kernel represents the convolution kernel, which is a two-dimensional array that defines the neighborhood of the corrosion operation, kernel = np.ones((a,b), np.uint8), np.uint8 represents all pixel values ​​in the unsigned 8-bit integer range, ranging from 0 to 255, (a,b) represents a × b two-dimensional array; iterations represents the number of iterations, and the result of the function output is to convolve the convolution kernel kernel with the input image to obtain the minimum value of the pixel point of the convolution kernel kernel, and use this minimum value to replace the pixel value of the reference point, and return the image after completing the number of iterations.

9. The method for identifying and counting microstructures of high-temperature alloy additive manufacturing according to claim 8, characterized in that: S3. Before extracting the features of the microstructure, it also includes using S3". Image enhancement algorithm to increase the clarity of the microstructure image after image corrosion. S3". Image enhancement algorithm uses the ImageEnhance.Sharpness() method in Python to create an image enhancement object, and calls the enhance() method to specify enhancement parameters.

10. The method for identifying and counting microstructures of high-temperature alloy additive manufacturing according to claim 1, characterized in that: S4. Classify and identify the microstructure characteristics and perform statistics, including: S4-1. Determine whether the area ratio X and the axis ratio Y of each of the microstructures are greater than a threshold value. If X>X0, Y>Y0 are satisfied, the microstructure is identified as a cellular structure, X0 is the threshold of the area ratio X, and Y0 is the threshold of the axis ratio Y; if X>X0, Y>Y0 are not satisfied, the microstructure is identified as a columnar crystal structure; S4-2. Count the data information of the cellular tissue and the data information of the columnar crystal tissue, wherein the data information includes at least one of the number of tissues, the diameter of the circumscribed circle of each cellular tissue, or the length of the major and minor axes of each columnar crystal, and the statistics are all performed based on pixel units.

Images