Image recognition-based SMA-13 standard part manufacturing method and device

Abstract

This invention relates to the field of road engineering material testing technology, specifically to a method and apparatus for manufacturing SMA-13 ​​standard parts based on image recognition. The method includes acquiring a basic image of an SMA-13 ​​mixture specimen; determining the original standard part from the basic image; acquiring the original standard part drawing based on the binarized image of the original standard part; obtaining a pre-standard part by laser engraving based on the original standard part drawing; cutting a PET film based on the original standard part drawing to obtain a coarse aggregate area mask and an asphalt adhesive area mask; coloring the pre-standard part by dividing it into sections using the coarse aggregate area mask and the asphalt adhesive area mask; acquiring an image of the pre-standard part after coloring; performing contrast verification and error analysis based on the image of the pre-standard part after coloring; and using the pre-standard part as the final standard part when both contrast verification and error analysis pass. This method enables efficient and accurate manufacturing of SMA-13 ​​standard parts.

Description

Technical Field

[0001] This invention relates to the field of road engineering material testing technology, specifically to a method and apparatus for manufacturing SMA-13 ​​standard parts based on image recognition. Background Technology

[0002] SMA-13 ​​asphalt mastic aggregate mixture is a commonly used material for the surface layer of high-grade highways. Its aggregate gradation characteristics and particle contact state directly determine the strength, stability, and durability of the pavement structure. Currently, the gradation distribution detection of SMA-13 ​​mixture mainly relies on the traditional sieving method. This method has problems such as cumbersome operation and low detection efficiency, and it cannot achieve quantitative analysis of the microstructure of the mixture. In terms of particle contact point identification, it mostly relies on manual observation or simple image analysis methods. Existing technologies have not yet clearly defined quantitative judgment standards for contact points and lack unified standards, resulting in identification accuracy being easily affected by subjective factors and poor stability.

[0003] In the field of asphalt mixture image recognition and detection, relevant technological achievements have been explored for application. For example, the patent CN119762520B, "A Deep Learning Segmentation Method for Asphalt Mixture CT Images," achieves accurate segmentation of the aggregate-binder boundary by preprocessing CT images through equalization and noise reduction, combined with Inception convolutional modules and residual connections to optimize and improve the U-Net model. Despite these technological breakthroughs, existing image analysis software and machine learning detection systems still suffer from unavoidable systematic errors. The key issue lies in the lack of dedicated standard components for verifying and calibrating the detection results, leading to insufficient reliability of the detection data. Summary of the Invention

[0004] The purpose of this invention is to provide a method and apparatus for manufacturing SMA-13 ​​standard parts based on image recognition, which can achieve efficient and accurate manufacturing of SMA-13 ​​standard parts.

[0005] In a first aspect of the present invention, a method for manufacturing an SMA-13 ​​standard part based on image recognition is provided, comprising:

[0006] Step 1. Obtain the basic image of the SMA-13 ​​mixture specimen; determine the original standard part based on the basic image; obtain the drawing of the original standard part based on the binarized image of the original standard part;

[0007] Step 2. Obtain a pre-standard part by laser engraving based on the original standard part drawing; cut PET film based on the original standard part drawing to obtain a mask for the coarse aggregate area and a mask for the asphalt adhesive area; color the pre-standard part by dividing it into sections using the mask for the coarse aggregate area and the mask for the asphalt adhesive area;

[0008] Step 3. Obtain the pre-standard part after coloring, and perform contrast verification and error analysis based on the pre-standard part after coloring image; when both contrast verification and error analysis pass, use the pre-standard part as the final standard part; when either contrast verification or error analysis fails, return to step 2 and adjust the operation parameters of step 2 based on the error analysis results.

[0009] As a preferred embodiment of the present invention, step 1, obtaining the basic image of the SMA-13 ​​mixture specimen specifically includes:

[0010] Multiple sets of SMA-13 ​​mixture specimens were prepared, and the cured SMA-13 ​​mixture specimens were cut transversely with a cutting machine to obtain a flat internal cross-section. The internal cross-section of each SMA-13 ​​mixture specimen was photographed with a camera to obtain a basic image.

[0011] As a preferred embodiment of the present invention, step 1, determining the original standard part through the basic image, specifically includes:

[0012] Step 11. Binarize the base images to obtain a binarized image that corresponds one-to-one with each base image;

[0013] Step 12. Based on the binarized images, obtain the average number of contact points, skeleton ratio, and uniformity index corresponding to each binarized image;

[0014] Step 13. Calculate the Marshall stability and dynamic stability corresponding to each binarized image based on the average number of contact points, skeleton ratio, and uniformity index.

[0015] Step 14. Select the binarized image with the best overall performance based on the Marshall stability and dynamic stability of each binarized image, and determine the SMA-13 ​​mixture specimen corresponding to the selected binarized image as the original standard specimen.

[0016] As a preferred embodiment of the present invention, step 11, the processing steps for each base image include:

[0017] Step 111. Perform grayscale processing on the base image to obtain a grayscale image;

[0018] Step 112. Perform 9-neighbor mean convolution denoising on the grayscale image to obtain a denoised image;

[0019] Step 113. Perform a fifth-order grayscale transformation on the denoised image to obtain an edge-enhanced image;

[0020] Step 114. Binarize the edge enhancement image to obtain a binarized image.

[0021] As a preferred embodiment of the present invention, in step 12, the processing step for each binarized image includes:

[0022] Step 121. Extract the contour of each coarse aggregate particle based on the binarized image;

[0023] Step 122. Calculate the number of contact points for each coarse aggregate particle;

[0024] Step 123. Calculate the total number of contact points of all coarse aggregate particles, and divide the total number of contact points by the number of coarse aggregate particles to obtain the average number of contact points;

[0025] Step 124. Calculate the total area of ​​all coarse aggregate particles I, calculate the total area of ​​all coarse aggregate particles with a contact point number greater than or equal to the threshold number II, and divide the total area II by the total area I to obtain the skeleton ratio.

[0026] Step 125. Divide the binarized image of the extracted coarse aggregate particle outlines into multiple sub-regions, and calculate the uniformity index using the following formula:

[0027]

[0028] in, As a uniformity index, The number of sub-regions The total area of ​​all coarse aggregate particles in the i-th sub-region is 3. for The average area of ​​each sub-region.

[0029] As a preferred embodiment of the present invention, step 122, the calculation of the number of contact points for each coarse aggregate particle, includes:

[0030] Step 1221. Determine the center position of the coarse aggregate particles;

[0031] Step 1222. Search for other coarse aggregate particles within a preset radius threshold range, using the center position as the center.

[0032] Step 1223. When the minimum boundary distance between a coarse aggregate particle and another coarse aggregate particle is less than the boundary distance threshold, the corresponding other coarse aggregate particle is taken as a contact point of the coarse aggregate particle.

[0033] Step 1224. Count the number of contact points of coarse aggregate particles.

[0034] As a preferred embodiment of the present invention, step 13, the calculation steps of the Marshall stability and dynamic stability of each binarized image, include:

[0035] The Marshall stability is calculated using the following formula:

[0036]

[0037] in, For Marshall stability, The average number of contact points. For skeleton ratio, The parameters a1, a2, a3, and a4 are obtained by fitting historical data to represent the uniformity index.

[0038] The dynamic stability is calculated using the following formula:

[0039]

[0040] in, For dynamic stability, The average number of contact points. For skeleton ratio, The parameters b1, b2, b3, and b4 are obtained by fitting historical data to represent the uniformity index.

[0041] As a preferred embodiment of the present invention, step 2, which involves coloring the pre-standard component by masking the coarse aggregate area and the asphalt adhesive area, specifically includes:

[0042] Step 21. Apply the asphalt adhesive mask to the pre-standard part to expose the coarse aggregate area of ​​the pre-standard part. Prepare dark gray to black epoxy ink and spray it onto the surface of the coarse aggregate area of ​​the pre-standard part using a spray gun.

[0043] Step 22. Apply the coarse aggregate area mask to the pre-standard part to expose the asphalt adhesive area of ​​the pre-standard part. Prepare light gray-beige epoxy ink and spray it onto the surface of the asphalt adhesive area of ​​the pre-standard part using a spray gun.

[0044] As a preferred embodiment of the present invention, step 3, which involves performing contrast verification and error analysis based on the pre-standard part after coloring, specifically includes:

[0045] Step 31. Based on the pre-standard part after coloring, obtain the average gray value of the coarse aggregate area and the average gray value of the asphalt adhesive area, calculate the gray value difference between the average gray value of the coarse aggregate area and the average gray value of the asphalt adhesive area, and determine whether the gray value difference is greater than the gray value difference threshold. When the gray value difference is greater than the gray value difference threshold, the contrast verification is passed; otherwise, the contrast verification is not passed.

[0046] Step 32. Based on the image of the pre-standard part after coloring, obtain the average number of contact points, skeleton ratio, and uniformity index of the pre-standard part. Calculate the first deviation value based on the average number of contact points of the pre-standard part and the average number of contact points of the original standard part. Calculate the second deviation value based on the skeleton ratio of the pre-standard part and the skeleton ratio of the original standard part. Calculate the third deviation value based on the uniformity index of the pre-standard part and the uniformity index of the original standard part. Determine whether the absolute value of the first deviation value is less than or equal to the average contact point deviation threshold, the absolute value of the second deviation value is less than or equal to the skeleton ratio deviation threshold, and the absolute value of the third deviation value is less than or equal to the uniformity index deviation threshold. If the absolute value of the first deviation value is less than or equal to the average contact point deviation threshold, the absolute value of the second deviation value is less than or equal to the skeleton ratio deviation threshold, and the absolute value of the third deviation value is less than or equal to the uniformity index deviation threshold, the error analysis passes; otherwise, the error analysis fails.

[0047] In a second aspect of the present invention, an apparatus for manufacturing SMA-13 ​​standard parts based on image recognition is provided, comprising:

[0048] The original standard part drawing acquisition module is configured to acquire the base image of the SMA-13 ​​mixture specimen; determine the original standard part through the base image; and acquire the original standard part drawing based on the binarized image of the original standard part.

[0049] The pre-standard part partition coloring module is configured to obtain a pre-standard part by laser engraving based on the original standard part drawing; cut a PET film based on the original standard part drawing to obtain a mask for the coarse aggregate area and a mask for the asphalt adhesive area; and perform partition coloring on the pre-standard part using the mask for the coarse aggregate area and the mask for the asphalt adhesive area.

[0050] The final standard part acquisition module is configured to acquire the pre-standard part after coloring, and perform contrast verification and error analysis based on the pre-standard part after coloring; when both contrast verification and error analysis pass, the pre-standard part is used as the final standard part.

[0051] In summary, the present invention has the following beneficial effects:

[0052] 1. This invention, through the establishment of unified, high-precision physical standard parts and the application of image recognition algorithms, achieves precise quantitative analysis of the aggregate distribution and particle contact state of SMA-13 ​​mixture. The laser engraving accuracy in the standard part manufacturing process reaches ±0.01mm, and combined with a dedicated ink color scheme, ensures clear grayscale contrast during image acquisition, effectively improving the accuracy and comparability of the recognition results. It solves the problem of inconsistent calibration in traditional standard parts, providing a unified and reliable benchmark for digital image recognition, and significantly improving the comparability and credibility of the detection results.

[0053] 2. This invention employs a pre-standard part trial production and error feedback adjustment mechanism. Through a step-by-step, parameterized processing procedure, it significantly reduces the complexity and human error inherent in traditional standard part preparation. Combined with processes such as PET film separation, zoned spraying, and multi-layer curing, it ensures the stability of the standard part's texture and color, thereby improving production efficiency and the finished product qualification rate.

[0054] 3. This invention explicitly sets feature points smaller than 1 mm as the threshold for determining aggregate contact points. Combined with optimized image recognition technology, it achieves automatic positioning and accurate counting of contact points. This scheme significantly improves recognition accuracy compared to traditional methods, providing reliable technical support for the quantitative analysis of particle contact states.

[0055] 4. The process described in this invention is compatible with various digital image acquisition devices and analysis software, and is suitable for both laboratory and field testing scenarios. By adjusting the threshold and parameters, this method can also be extended to the fabrication and structural analysis of standard components for other asphalt mixture types, demonstrating good engineering applicability and technological extension value.

[0056] 5. By establishing clear criteria for determining contact points, methods for calculating structural parameters, and specifications for manufacturing standard parts, this invention promotes the transformation of asphalt mixture microstructure testing from experience-based judgment to digitalization and standardization, providing technical support for the improvement and intelligent upgrading of the industry's quality assessment system.

[0057] Further or more detailed beneficial effects will be described in conjunction with specific embodiments in the detailed implementation. Attached Figure Description

[0058] Figure 1 A flowchart illustrating the manufacturing method of the SMA-13 ​​standard part based on image recognition according to an embodiment of the present invention is shown;

[0059] Figure 2 A block diagram of an apparatus for manufacturing SMA-13 ​​standard parts based on image recognition according to an embodiment of the present invention is shown. Detailed Implementation

[0060] Embodiments of the present invention will now be described in more detail with reference to the accompanying drawings. While some embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the invention. It should be understood that the accompanying drawings and embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the invention.

[0061] In the description of embodiments of the present invention, the term "comprising" and similar terms should be understood as open-ended inclusion, i.e., "including but not limited to". The term "based on" should be understood as "at least partially based on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc., may refer to different or the same objects. Other explicit and implicit definitions may also be included below.

[0062] Figure 1 A flowchart illustrating a method for manufacturing an SMA-13 ​​standard part based on image recognition, according to an embodiment of the present invention, is shown. The method includes:

[0063] Step 1. Obtain the basic image of the SMA-13 ​​mixture specimen; determine the original standard part based on the basic image; obtain the drawing of the original standard part based on the binarized image of the original standard part;

[0064] Step 2. Obtain a pre-standard part by laser engraving based on the original standard part drawing; cut PET film based on the original standard part drawing to obtain a mask for the coarse aggregate area and a mask for the asphalt adhesive area; color the pre-standard part by dividing it into sections using the mask for the coarse aggregate area and the mask for the asphalt adhesive area;

[0065] Step 3. Obtain the pre-standard part after coloring, and perform contrast verification and error analysis based on the pre-standard part after coloring image; when both contrast verification and error analysis pass, use the pre-standard part as the final standard part; when either contrast verification or error analysis fails, return to step 2 and adjust the operation parameters of step 2 based on the error analysis results.

[0066] In step 1 of this embodiment, obtaining the basic image of the SMA-13 ​​mixture specimen specifically includes:

[0067] Multiple sets of SMA-13 ​​mixture specimens were prepared, and the cured SMA-13 ​​mixture specimens were cut transversely with a cutting machine to obtain a flat internal cross-section. The internal cross-section of each SMA-13 ​​mixture specimen was photographed to obtain basic images. Specifically, based on the determined SMA-13 ​​gradation median ±0.3% range for a specific project, 10 key sieve aperture parameters were determined, and the mass percentage of each grade was calculated. Multiple sets of SMA-13 ​​mixture specimens were prepared according to the "Test Procedures for Asphalt and Asphalt Mixtures in Highway Engineering" (JTG3410-2025). The cured specimens were cut transversely with a cutting machine to obtain a flat internal cross-section. Digital images of the cross-section were acquired using a high-resolution industrial camera, obtaining no fewer than 50 sets of basic images.

[0068] In step 1 of this embodiment, determining the original standard part through the basic image specifically includes:

[0069] Step 11. Binarize the base images to obtain a binary image corresponding to each base image. The processing steps for each base image include: Step 111. Converting the base image to grayscale to obtain a grayscale image; Step 112. Performing 9-neighborhood mean convolution denoising on the grayscale image to obtain a denoised image; Step 113. Performing a fifth-order grayscale transform on the denoised image to obtain an edge-enhanced image; Step 114. Binarizing the edge-enhanced image to obtain a binary image. In the binary image, white represents coarse aggregate particles, and black represents fine aggregate particles, asphalt mastic, and voids.

[0070] Step 12. Obtain the average number of contact points, skeleton ratio, and uniformity index corresponding to each binarized image. The processing steps for each binarized image include:

[0071] Step 121. Extract the contour of each coarse aggregate particle based on the binarized image. The contour of each coarse aggregate particle can be extracted using the findContours function of OpenCV. Each coarse aggregate particle has a contour array, which stores the x and y coordinates of all pixels on the contour boundary of the coarse aggregate particle.

[0072] Step 122. Calculate the number of contact points for each coarse aggregate particle. The calculation steps for the number of contact points for each coarse aggregate particle include: Step 1221. Determine the center position of the coarse aggregate particle; Step 1222. Search for other coarse aggregate particles within a preset radius threshold range, using the center position as the center; Step 1223. When the minimum boundary distance between a coarse aggregate particle and another coarse aggregate particle is less than the boundary distance threshold, the corresponding other coarse aggregate particle is considered as one contact point of the coarse aggregate particle; Step 1224. Count the number of contact points for each coarse aggregate particle.

[0073] Taking coarse aggregate particle No. 1 as an example, first obtain the center position of coarse aggregate particle No. 1, and then search for other coarse aggregate particles within a preset radius threshold range (e.g., 2mm). Assuming that coarse aggregate particles No. 5, No. 8, and No. 10 appear within the search range (if a coarse aggregate particle is partially found within the search range, it is considered to be another coarse aggregate particle within the preset radius threshold range), then coarse aggregate particles No. 5, No. 8, and No. 10 are considered other coarse aggregate particles within the preset radius threshold range. Then, obtain the minimum boundary distance between coarse aggregate particle #5 and coarse aggregate particle #1 (let's assume it's s1), the minimum boundary distance between coarse aggregate particle #8 and coarse aggregate particle #1 (let's assume it's s2), and the minimum boundary distance between coarse aggregate particle #10 and coarse aggregate particle #1 (let's assume it's s3). Assuming s1 is greater than the boundary distance threshold (e.g., 1 mm), and s2 and s3 are less than the boundary distance threshold, then coarse aggregate particles #8 and #10 are the contact points of coarse aggregate particle #1. Finally, the number of contact points for coarse aggregate particle #1 is determined to be 2. Repeating steps 1221 to 1224 can determine the number of contact points for each coarse aggregate particle.

[0074] Step 123. Calculate the total number of contact points for all coarse aggregate particles, and divide the total number of contact points by the number of coarse aggregate particles to obtain the average number of contact points. Adding up the total number of contact points for all coarse aggregate particles gives the total number of contact points, and dividing the total number of contact points by the number of coarse aggregate particles gives the average number of contact points.

[0075] Step 124. Calculate the total area of ​​all coarse aggregate particles, and calculate the total area of ​​all coarse aggregate particles with a contact point count greater than or equal to a threshold value, then divide the total area of ​​particle 1 by the total area of ​​particle 2 to obtain the skeleton ratio. In this embodiment, the threshold value can be 3. Assume there are coarse aggregate particles No. 1 (3 contact points), No. 2 (2 contact points), and No. 3 (5 contact points). Add the areas of coarse aggregate particles No. 1, No. 2, and No. 3 to obtain the total area of ​​particle 1. Add the areas of coarse aggregate particles No. 3 and No. 4 to obtain the total area of ​​particle 1. Finally, divide the total area of ​​particle 1 by the total area of ​​particle 2 to obtain the skeleton ratio.

[0076] Step 125. Divide the binarized image of the extracted coarse aggregate particle outlines into multiple sub-regions, and calculate the uniformity index using the following formula:

[0077]

[0078] in, As a uniformity index, The number of sub-regions The total area of ​​all coarse aggregate particles in the i-th sub-region is 3. for The average area of ​​each sub-region.

[0079] Assuming the image is divided into 8 sub-regions in a fan shape, the total area of ​​all coarse aggregate particles in sub-region 1 is 540 square millimeters, in sub-region 2 it is 510 square millimeters, in sub-region 3 it is 560 square millimeters, in sub-region 4 it is 490 square millimeters, in sub-region 5 it is 530 square millimeters, in sub-region 6 it is 550 square millimeters, in sub-region 7 it is 500 square millimeters, and in sub-region 8 it is 570 square millimeters. Therefore, the average area of ​​the 8 sub-regions is 531.25 square millimeters, and the final uniformity index is 29 square millimeters.

[0080] Step 13. Calculate the Marshall stability and dynamic stability corresponding to each binarized image based on the average number of contact points, skeleton ratio, and uniformity index. The calculation steps for the Marshall stability and dynamic stability of each binarized image include:

[0081] The Marshall stability is calculated using the following formula:

[0082]

[0083] in, For Marshall stability, The average number of contact points. For skeleton ratio, As a uniformity index, parameters a1, a2, a3, and a4 are obtained by fitting historical data. For example, the Marshall stability calculation formula can be:

[0084]

[0085] The dynamic stability is calculated using the following formula:

[0086]

[0087] in, For dynamic stability, The average number of contact points. For skeleton ratio, As a uniformity index, parameters b1, b2, b3, and b4 are obtained by fitting historical data. For example, the dynamic stability calculation formula can be:

[0088]

[0089] Step 14. Select the binarized image with the best overall performance based on the Marshall stability and dynamic stability of each binarized image, and determine the SMA-13 ​​mixture specimen corresponding to the selected binarized image as the original standard specimen.

[0090] Assuming that the 23rd binarized image has the best overall performance based on Marshall stability and dynamic stability, then the SMA-13 ​​mixture specimen corresponding to the 23rd binarized image is determined as the original standard specimen.

[0091] In step 1 of this embodiment, obtaining the original standard part drawing based on the binarized image of the original standard part specifically involves: converting the obtained binarized image into a vector graphic that can be read by AutoCAD using professional vectorization processing software.

[0092] In step 2 of this embodiment, obtaining the pre-standard part based on the original standard part drawing via laser marking specifically involves: inputting the converted vector image into a laser marking machine, setting the core process parameters of the laser marking machine as follows: resolution 600 dpi, line width 0.1 mm, scanning speed 150 mm / s, laser power 20 W, pulse frequency 50 kHz, focal length 160 mm, while controlling the equipment positioning accuracy ≤ ±0.01 mm and repeatability ≤ ±0.005 mm. A metal block with a diameter of 101.6 mm and a thickness of 20 mm is obtained, and the pre-standard part is obtained by laser marking on the surface of the metal block.

[0093] In step 2 of this embodiment, the process of coloring the pre-standard part by masking the coarse aggregate area and the asphalt adhesive area specifically includes:

[0094] Step 21. Apply the asphalt adhesive area mask to the pre-standard part to expose the coarse aggregate area. Prepare a dark gray to black epoxy ink and spray it onto the surface of the coarse aggregate area of ​​the pre-standard part using a spray gun. Specifically: Apply the asphalt adhesive area mask to expose the coarse aggregate area. Mix carbon black ink and iron black ink in a 4:1 ratio to create a dark gray to black ink. Use epoxy weather-resistant ink. Place the spray gun 10cm away from the surface of the metal block, set the spraying pressure to 0.3MPa, and spray one thin layer. After spraying, allow it to dry, ensuring that the ink fills the shallow grooves without covering the texture boundaries. Then remove the asphalt adhesive area mask.

[0095] Step 22. Apply the coarse aggregate area mask to the pre-standard part to expose the asphalt adhesive area. Prepare a light gray-beige epoxy ink and spray it onto the surface of the asphalt adhesive area of ​​the pre-standard part using a spray gun. Specifically: Apply the coarse aggregate area mask to expose the asphalt adhesive area. Mix titanium dioxide ink, zinc barium white ink, and ochre petroleum ink in a ratio of 8:1:1 to prepare a light gray-beige ink. Select epoxy weather-resistant ink as the ink type. Place the spray gun 10cm away from the surface of the metal block, set the spraying pressure to 0.3MPa, and spray one thin layer. After spraying, let it stand and dry to ensure that the ink fills the shallow grooves without covering the texture boundaries. Then remove the coarse aggregate area mask.

[0096] In step 3 of this embodiment, the contrast verification and error analysis based on the pre-standard part after coloring specifically includes:

[0097] Step 31. Based on the image of the pre-standard part after coloring, obtain the average gray value of the coarse aggregate area and the average gray value of the asphalt adhesive area. Calculate the gray value difference between the two areas and determine if the gray value difference is greater than the gray value difference threshold. If the gray value difference is greater than the threshold, the contrast verification passes; otherwise, the contrast verification fails. After all areas are sprayed, cure at 100℃ for 60 minutes, then spray a matte clear varnish to eliminate reflection. Obtain the image of the pre-standard part after coloring and use ImageJ to obtain the average gray value of the coarse aggregate area (assumed to be 37.6) and the average gray value of the asphalt adhesive area (assumed to be 164.3). The calculated gray value difference is 126.7. Assuming the gray value difference threshold is 50, since the gray value difference is greater than the threshold, the contrast verification passes.

[0098] Step 32. Based on the image of the pre-standard part after coloring, obtain the average number of contact points, skeleton ratio, and uniformity index of the pre-standard part. Calculate the first deviation value based on the average number of contact points of the pre-standard part and the average number of contact points of the original standard part. Calculate the second deviation value based on the skeleton ratio of the pre-standard part and the skeleton ratio of the original standard part. Calculate the third deviation value based on the uniformity index of the pre-standard part and the uniformity index of the original standard part. Determine whether the absolute value of the first deviation value is less than or equal to the average contact point deviation threshold, the absolute value of the second deviation value is less than or equal to the skeleton ratio deviation threshold, and the absolute value of the third deviation value is less than or equal to the uniformity index deviation threshold. If the absolute value of the first deviation value is less than or equal to the average contact point deviation threshold, the absolute value of the second deviation value is less than or equal to the skeleton ratio deviation threshold, and the absolute value of the third deviation value is less than or equal to the uniformity index deviation threshold, the error analysis passes; otherwise, the error analysis fails.

[0099] The average number of contact points, skeleton ratio, and uniformity index of the pre-standard part are obtained from the image after coloring. The specific acquisition steps are the same as steps 11 and 12. It is assumed that the average number of contact points of the pre-standard part is 1355, the skeleton ratio is 48.3%, and the uniformity index is 0.138 square millimeters; the average number of contact points of the original standard part is 1391, the skeleton ratio is 46.7%, and the uniformity index is 0.128 square millimeters. The calculated first deviation value is -2.59%, the second deviation value is 3.43%, and the third deviation value is 7.81%. It is assumed that the deviation thresholds for the average number of contact points are 5%, the skeleton ratio deviation threshold is 3%, and the uniformity index deviation threshold is 2%. The absolute value of the first deviation value is less than the average number of contact points deviation threshold, which meets the requirements. However, the absolute values ​​of the second and third deviation values ​​are greater than the skeleton ratio deviation threshold and the uniformity index deviation threshold, respectively, which do not meet the requirements. Therefore, the error analysis fails.

[0100] When the error analysis fails, a comprehensive analysis of the average number of contact points, skeleton ratio, and uniformity index of the pre-standard parts reveals the following:

[0101] The low average number of contact points indicates a loss of outline and blurred edges in the aggregate contact area. The corresponding process defects are an excessively wide laser-etched line and poor adhesion between the PET film and the metal block, leading to pigment diffusion during spraying and causing the contact points to "merge." The solution is to reduce the width of the etched lines and improve the adhesion accuracy between the PET film and the metal block. Specific adjustments to the operating parameters include: 1. Adjusting the line width from 0.1mm to 0.05mm; 2. Roughening the surface of the metal block with sandblasting to enhance the adhesion friction of the PET board; 3. Using high-temperature resistant tape to fix the PET film, ensuring no air bubbles or curling edges, and eliminating pigment diffusion caused by gaps.

[0102] The error of a high skeleton ratio stems from an overestimation of the equivalent diameter of the coarse aggregate. The corresponding process defects are excessive pigment coating thickness leading to "thickening" of the aggregate edges and an excessively high printing scaling ratio. It is necessary to control the pigment coating thickness and calibrate the printing scaling ratio. Specific operational parameters to be adjusted include: 1. Using air-assisted fine spraying, reducing the spraying pressure from 0.3MPa to 0.15MPa; 2. Reducing the amount of each spray, replacing "single thick spray" with "multiple thin sprays"; 3. Based on the deviation of the high skeleton ratio, setting the scaling factor for the re-recorded image to the reverse scaling factor to offset the problem of an excessively large aggregate area, thus reducing the skeleton ratio deviation at its source. The formula for calculating the reverse scaling factor is... Therefore, when reburning, the image scaling ratio is set to 98.3%.

[0103] The large uniformity error originates from localized deviations in aggregate distribution density. The corresponding process defects are localized pixel shifts during recording and uneven pigment deposition during spraying, leading to distortion of the aggregate outline in certain areas. An optimization plan is needed to adjust the printing resolution and improve spray uniformity. Specific operational parameters to be adjusted include: 1. Increasing the recording resolution from 600dpi to 1200dpi to reduce pixel size, restore the edges and fine contact areas of the original sample aggregate, and avoid contact point loss due to pixel blurring; 2. Using uniform-speed horizontal spraying, controlling the speed at 5~8cm / s to avoid localized accumulation during manual spraying.

[0104] Then repeat steps 2 and 3:

[0105] When obtaining a pre-standard part based on the original standard part drawing using laser engraving, the converted vector image is input into the laser marking machine, the resolution is adjusted to 1200dpi, the line width is 0.05mm, the image scaling ratio is set to 98.3%, and the engraving is performed on the surface of the metal block. After engraving, the surface of the metal block is roughened by sandblasting.

[0106] When coloring pre-standard parts by masking the coarse aggregate area and the asphalt adhesive area, high-temperature resistant tape is used to attach and fix the PET masking of the asphalt adhesive area, exposing the coarse aggregate area. A spray gun with dark gray to near-black ink is placed 10cm away from the surface of the metal block. Air-assisted fine spraying is used, with a spraying pressure set to 0.15MPa. A uniform, horizontal spraying speed of 5-8cm / s is employed, using a "multi-layer thin spraying" process (curing for 15 minutes after each thin spray), for a total of two layers. The surface is then cured at 60 degrees Celsius for 20 minutes. Afterward, the asphalt adhesive area masking is removed, and high-temperature resistant tape is used to attach and fix the PET masking of the coarse aggregate area, exposing the asphalt adhesive area. The ink is then replaced with a light gray-beige, and the remaining spraying settings are the same as for the coarse aggregate area.

[0107] After all areas were sprayed, the coating was cured at 100℃ for 60 minutes. Then, a matte clear coat was applied to eliminate reflections. Images of the pre-standard part after coloring were acquired, and the average grayscale value of the coarse aggregate area (assumed to be 43.1) and the average grayscale value of the asphalt adhesive area (assumed to be 172.6) were obtained using ImageJ. The calculated grayscale difference was 129.5. Assuming a grayscale difference threshold of 50, since the grayscale difference value is greater than the threshold, the contrast verification was deemed successful.

[0108] The average number of contact points, skeleton ratio, and uniformity index of the pre-standard part are obtained from the image after coloring. The specific acquisition steps are the same as steps 11 and 12. Assume the second pre-standard part has an average of 1386 contact points, a skeleton ratio of 46.9%, and a uniformity index of 0.129 mm²; the original standard part has an average of 1391 contact points, a skeleton ratio of 46.7%, and a uniformity index of 0.128 mm². The calculated first deviation value is -0.36%, the second deviation value is 0.43%, and the third deviation value is 0.78%. The deviation thresholds for the average number of contact points are 5%, the skeleton ratio deviation threshold is 3%, and the uniformity index deviation threshold is 2%. The absolute values ​​of the first, second, and third deviations are all less than the average number of contact points deviation threshold, the skeleton ratio deviation threshold, and the uniformity index deviation threshold, respectively. All three deviations meet the requirements, so the error analysis is passed. The second pre-standard part is used as the final standard part.

[0109] Figure 2 A block diagram of an apparatus for manufacturing SMA-13 ​​standard parts based on image recognition, according to an embodiment of the present invention, is shown. The apparatus includes:

[0110] The original standard part drawing acquisition module 402 is configured to acquire the basic image of the SMA-13 ​​mixture specimen; determine the original standard part through the basic image; and acquire the original standard part drawing based on the binarized image of the original standard part.

[0111] The pre-standard part partition coloring module 404 is configured to obtain a pre-standard part by laser engraving based on the original standard part drawing; to cut a PET film based on the original standard part drawing to obtain a mask for the coarse aggregate area and a mask for the asphalt adhesive area; and to perform partition coloring on the pre-standard part through the mask for the coarse aggregate area and the mask for the asphalt adhesive area.

[0112] The final standard part acquisition module 406 is configured to acquire the colorized image of the pre-standard part, and perform contrast verification and error analysis based on the colorized image of the pre-standard part; when both contrast verification and error analysis pass, the pre-standard part is used as the final standard part.

[0113] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. A method for manufacturing an SMA-13 standard part based on image recognition, characterized in that, include: Step 1. Obtain the basic image of the SMA-13 ​​mixture specimen; The original standard part is determined by the base image; Obtain the original standard part drawing based on the binarized image of the original standard part; Step 2. Obtain a pre-standard part by laser engraving based on the original standard part drawing; Cut PET film based on the original standard part drawings to obtain the coarse aggregate area mask and the asphalt adhesive area mask; use the coarse aggregate area mask and the asphalt adhesive area mask to color the pre-standard part in sections; Step 3. Obtain the pre-standard part after coloring, and perform contrast verification and error analysis based on the pre-standard part after coloring image; when both contrast verification and error analysis pass, use the pre-standard part as the final standard part; when either contrast verification or error analysis fails, return to step 2 and adjust the operation parameters of step 2 based on the error analysis results.

2. The method of claim 1, wherein, In step 1, obtaining the basic image of the SMA-13 ​​mixture specimen specifically includes: Multiple sets of SMA-13 ​​mixture specimens were prepared, and the cured SMA-13 ​​mixture specimens were cut transversely with a cutting machine to obtain a flat internal cross-section. The internal cross-section of each SMA-13 ​​mixture specimen was photographed with a camera to obtain a basic image.

3. The method according to claim 1, characterized in that, In step 1, determining the original standard part through the basic image specifically includes: Step 11. Binarize the base images to obtain a binarized image corresponding to each of the base images; Step 12. Based on the binarized images, obtain the average number of contact points, skeleton ratio, and uniformity index corresponding to each binarized image. Step 13. Calculate the Marshall stability and dynamic stability corresponding to each of the binarized images based on the average number of contact points, skeleton ratio, and uniformity index. Step 14. Select the binarized image with the best overall performance based on the Marshall stability and dynamic stability of each binarized image, and determine the SMA-13 ​​mixture specimen corresponding to the selected binarized image as the original standard specimen.

4. The method according to claim 3, characterized in that, In step 11, the processing steps for each of the basic images include: Step 111. Perform grayscale processing on the base image to obtain a grayscale image; Step 112. Perform 9-neighbor mean convolution denoising on the grayscale image to obtain a denoised image; Step 113. Perform a fifth-order grayscale transformation on the denoised image to obtain an edge-enhanced image; Step 114. Binarize the edge enhancement image to obtain a binarized image.

5. The method according to claim 3, characterized in that, In step 12, the processing steps for each binarized image include: Step 121. Extract the contour of each coarse aggregate particle based on the binarized image; Step 122. Calculate the number of contact points for each coarse aggregate particle; Step 123. Calculate the total number of contact points of all coarse aggregate particles, and divide the total number of contact points by the number of coarse aggregate particles to obtain the average number of contact points; Step 124. Calculate the total area of ​​all coarse aggregate particles I, calculate the total area of ​​all coarse aggregate particles with a contact point number greater than or equal to the threshold number II, and divide the total area II by the total area I to obtain the skeleton ratio. Step 125. Divide the binarized image of the extracted coarse aggregate particle outlines into multiple sub-regions, and calculate the uniformity index using the following formula: ， in, As a uniformity index, The number of sub-regions The total area of ​​all coarse aggregate particles in the i-th sub-region is 3. for The average area of ​​each sub-region.

6. The method according to claim 5, characterized in that, In step 122, the calculation of the number of contact points for each coarse aggregate particle includes: Step 1221. Determine the center position of the coarse aggregate particles; Step 1222. Search for other coarse aggregate particles within a preset radius threshold range, using the center position as the center. Step 1223. When the minimum boundary distance between the coarse aggregate particle and another coarse aggregate particle is less than the boundary distance threshold, the corresponding other coarse aggregate particle is taken as a contact point of the coarse aggregate particle. Step 1224. Count the number of contact points of the coarse aggregate particles.

7. The method according to claim 3, characterized in that, In step 13, the calculation steps for the Marshall stability and dynamic stability of each binarized image include: The Marshall stability is calculated using the following formula: ， in, For Marshall stability, The average number of contact points. For skeleton ratio, The parameters a1, a2, a3, and a4 are obtained by fitting historical data to represent the uniformity index. The dynamic stability is calculated using the following formula: ， in, For dynamic stability, The average number of contact points. For skeleton ratio, The parameters b1, b2, b3, and b4 are obtained by fitting historical data to represent the uniformity index.

8. The method according to claim 1, characterized in that, In step 2, the pre-standard parts are colored in sections by masking the coarse aggregate area and the asphalt adhesive area, which specifically includes: Step 21. Apply the asphalt adhesive mask to the pre-standard part to expose the coarse aggregate area of ​​the pre-standard part. Prepare dark gray to black epoxy ink and spray it onto the surface of the coarse aggregate area of ​​the pre-standard part using a spray gun. Step 22. Apply the coarse aggregate area mask to the pre-standard part to expose the asphalt adhesive area of ​​the pre-standard part. Prepare light gray-beige epoxy ink and spray it onto the surface of the asphalt adhesive area of ​​the pre-standard part using a spray gun.

9. The method according to claim 1, characterized in that, Step 3, specifically, includes contrast verification and error analysis based on the pre-standard part's colored image, which includes: Step 31. Based on the image of the pre-standard part after coloring, obtain the average gray value of the coarse aggregate area and the average gray value of the asphalt adhesive area, calculate the gray value difference between the average gray value of the coarse aggregate area and the average gray value of the asphalt adhesive area, and determine whether the gray value difference is greater than the gray value difference threshold. When the gray value difference is greater than the gray value difference threshold, the contrast verification is passed; otherwise, the contrast verification is not passed. Step 32. Based on the image of the pre-standard part after coloring, obtain the average number of contact points, skeleton ratio, and uniformity index of the pre-standard part. Calculate a first deviation value based on the average number of contact points of the pre-standard part and the average number of contact points of the original standard part. Calculate a second deviation value based on the skeleton ratio of the pre-standard part and the skeleton ratio of the original standard part. Calculate a third deviation value based on the uniformity index of the pre-standard part and the uniformity index of the original standard part. Determine whether the absolute value of the first deviation value is less than or equal to the average contact point deviation threshold, whether the absolute value of the second deviation value is less than or equal to the skeleton ratio deviation threshold, and whether the absolute value of the third deviation value is less than or equal to the uniformity index deviation threshold. If the absolute value of the first deviation value is less than or equal to the average contact point deviation threshold, the absolute value of the second deviation value is less than or equal to the skeleton ratio deviation threshold, and the absolute value of the third deviation value is less than or equal to the uniformity index deviation threshold, the error analysis passes; otherwise, the error analysis fails.

10. A fabrication apparatus for SMA-13 ​​standard parts based on image recognition, characterized in that, include: The original standard parts drawing acquisition module is configured to acquire the basic image of the SMA-13 ​​mixture specimen; The original standard part is determined by the base image; Obtain the original standard part drawing based on the binarized image of the original standard part; The pre-standard part partitioning coloring module is configured to obtain pre-standard parts based on laser engraving of the original standard part drawings; Cut PET film based on the original standard part drawings to obtain the coarse aggregate area mask and the asphalt adhesive area mask; use the coarse aggregate area mask and the asphalt adhesive area mask to color the pre-standard part in sections; The final standard part acquisition module is configured to acquire the pre-standard part after coloring, and perform contrast verification and error analysis based on the pre-standard part after coloring; when both contrast verification and error analysis pass, the pre-standard part is used as the final standard part.

Images