Method, device and apparatus for splitting asphalt mixture components

Abstract

This application discloses a method, apparatus, and equipment for segmenting asphalt mixture components, relating to the field of asphalt pavement construction control technology. The method includes: acquiring an initial two-dimensional scan image of an asphalt mixture cross-section; preprocessing the initial two-dimensional scan image, including contrast enhancement and light field correction; dividing the preprocessed image into components using a dual-threshold segmentation method; performing morphological post-processing on the initial segmented image to obtain separated component segmented images; calculating the difference between the image area ratio and the actual volume ratio of each component segmented image; and calculating the relative error between the image sieve distribution and the actual sieve distribution of the aggregate components. When the difference or relative error exceeds a preset threshold range, adjusting the image processing parameters and re-performing the segmentation operation to obtain a target segmented image. The difference and relative error of the target segmented image are both within the preset threshold range, thus solving the problem of insufficient segmentation accuracy of asphalt mixture components.

Description

Technical Field

[0001] This application relates to the field of asphalt pavement construction control technology, and in particular to methods, apparatus and equipment for separating asphalt mixture components. Background Technology

[0002] Asphalt mixtures are multiphase materials composed of coarse aggregates, asphalt mortar, and voids, with a complex internal structure. The composition and spatial distribution of this internal structure significantly influence the macroscopic properties of asphalt mixtures. In-depth research into the distribution patterns of the internal microstructure of asphalt mixtures is crucial for optimizing asphalt mixture design methods and improving their macroscopic properties. Traditional methods typically utilize camera or computed tomography (CT) techniques to identify the microstructure of asphalt mixtures.

[0003] However, due to problems such as small differences in grayscale and blurred boundaries between the components of asphalt mixture, the segmentation accuracy of asphalt mixture components is insufficient.

[0004] Therefore, how to improve the segmentation accuracy of asphalt mixture components is an urgent problem to be solved. Summary of the Invention

[0005] The main objective of this application is to provide a method, apparatus, and equipment for segmenting asphalt mixture components, aiming to solve the technical problem of insufficient segmentation accuracy of asphalt mixture components.

[0006] To achieve the above objectives, this application proposes a method for segmenting asphalt mixture components, the method comprising:

[0007] Obtain an initial two-dimensional scan image of the cross-section of the asphalt mixture;

[0008] The initial two-dimensional scanned image is preprocessed to obtain a preprocessed image, the preprocessing including contrast enhancement and light field correction;

[0009] The preprocessed image is divided into components using a dual-threshold segmentation method to obtain an initial segmented image.

[0010] Morphological post-processing is performed on the initial segmented image to obtain segmented images of each separated component;

[0011] Based on the segmented images of each component, calculate the difference between the image area ratio and the actual volume ratio of each segmented image.

[0012] Based on the segmented images of each component, the relative error between the image sieve distribution and the actual sieve distribution of the aggregate components is calculated;

[0013] When the difference value or relative error exceeds the preset threshold range, the image processing parameters are adjusted and the segmentation operation is re-executed to obtain the target segmented image, wherein the difference value and relative error of the target segmented image are both within the preset threshold range.

[0014] In one embodiment, the step of preprocessing the initial two-dimensional scanned image to obtain a preprocessed image includes:

[0015] Based on the RGB color model, the initial two-dimensional scanned image is linearly enhanced by a preset multiple to obtain a contrast-enhanced image;

[0016] The contrast-enhanced image is subjected to light field correction based on a grayscale mask algorithm to obtain a preprocessed image.

[0017] In one embodiment, the step of linearly enhancing the initial two-dimensional scanned image by a preset multiple based on the RGB color model to obtain a contrast-enhanced image includes:

[0018] Based on the RGB color model, the initial two-dimensional scanned image is subjected to RGB channel separation to obtain the green channel;

[0019] The green channel is linearly enhanced by a preset multiple to obtain an enhanced green channel;

[0020] The enhanced green channel is then merged with the original red and blue channels to obtain a contrast-enhanced image.

[0021] In one embodiment, the step of performing light field correction on the contrast-enhanced image based on a grayscale mask algorithm to obtain a preprocessed image includes:

[0022] An initial mask is created in the contrast-enhanced image, and pixel regions with gray values ​​below the background threshold outside the mask are identified. The initial mask is used to cover the target region.

[0023] The background noise region is determined based on the grayscale threshold of the initial mask and the background grayscale threshold;

[0024] The background noise region is smoothed to obtain an intermediate mask, and the smoothing process is used to eliminate edge artifacts.

[0025] The effective image region is extracted based on the intermediate mask, and the invalid region is assigned an empty value to obtain the preprocessed image.

[0026] In one embodiment, the step of performing morphological post-processing on the initial segmented image to obtain separate segmented images of each component includes:

[0027] Calculate the foreground-to-background distance transformation matrix in the initial segmented image;

[0028] Add a minimum constraint marker to the distance transformation matrix; the minimum constraint marker is used to suppress non-true local minima.

[0029] Based on the minimum constraint marker, the distance transformation matrix is ​​modified to generate a controlled distance transformation graph;

[0030] Perform a watershed transformation on the controlled distance transformation map to generate a particle segmentation boundary label matrix;

[0031] Based on the particle segmentation boundary label matrix, the segmentation boundary is assigned as the background, and the segmented images of each component are output.

[0032] In one embodiment, the step of calculating the difference between the image area ratio and the actual volume ratio of each component segmented image based on the segmented images of each component includes:

[0033] Obtain the number of pixels in each segmented image to get the number of pixels in each component;

[0034] Calculate the proportion of the number of pixels in each component to the total number of pixels in the image to obtain the area ratio of each component in the image;

[0035] Calculate the actual volume percentage of each component based on the actual mass and density of each component in the asphalt mixture;

[0036] The difference value is calculated based on the area ratio of each component image and the actual volume ratio of each component.

[0037] In one embodiment, the step of calculating the relative error between the image sieve distribution and the actual sieve distribution of aggregate components based on the segmented images of each component includes:

[0038] Based on the physical length and pixel distance of the scale in each component segmented image, calculate the conversion coefficient between pixel size and actual size;

[0039] Calculate the minimum area threshold corresponding to the sieve aperture size based on the sieve aperture size and the conversion coefficient;

[0040] Based on the minimum area threshold, aggregate particles are screened step by step, and the equivalent diameter of each aggregate particle is calculated.

[0041] Based on the equivalent diameter, the cumulative throughput of each sieve aperture size is calculated to obtain the image sieve classification.

[0042] Calculate the relative error between the actual 3D sieve gradation and the image sieve gradation.

[0043] In one embodiment, adjusting the image processing parameters includes:

[0044] Modify at least one of the following parameters: green channel enhancement coefficient, morphological operation structuring element size, and watershed minimum constraint parameter.

[0045] Furthermore, to achieve the above objectives, this application also proposes a device for separating asphalt mixture components, the device comprising:

[0046] The image acquisition module is used to acquire an initial two-dimensional scan image of the cross-section of the asphalt mixture;

[0047] The image preprocessing module is used to preprocess the initial two-dimensional scan image to obtain a preprocessed image. The preprocessing includes contrast enhancement and light field correction.

[0048] The initial segmentation module is used to divide the preprocessed image into components using a dual-threshold segmentation method to obtain an initial segmented image.

[0049] The morphological processing module is used to perform morphological post-processing on the initial segmented image to obtain segmented images of each separated component.

[0050] The difference calculation module is used to calculate the difference between the image area ratio and the actual volume ratio of each component segmented image based on the segmented images of each component.

[0051] The error calculation module is used to calculate the relative error between the image sieve distribution and the actual sieve distribution of the aggregate components based on the segmented images of each component.

[0052] The parameter adjustment module is used to adjust the image processing parameters and re-execute the segmentation operation when the difference value or the relative error exceeds the preset threshold range, so as to obtain the target segmented image, wherein the difference value and relative error of the target segmented image are both within the preset threshold range.

[0053] In addition, to achieve the above objectives, this application also proposes a device for segmenting asphalt mixture components, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the asphalt mixture component segmentation method as described above.

[0054] One or more technical solutions proposed in this application have at least the following technical effects:

[0055] Initial two-dimensional scan images of asphalt mixture cross-sections were acquired. These images were then preprocessed to obtain preprocessed images, including contrast enhancement and light field correction. Contrast enhancement and light field correction eliminated illumination interference while enhancing the color difference between components and improving the clarity of segmentation boundaries. A dual-threshold segmentation method was used to divide the preprocessed images into components, obtaining initial segmented images, avoiding the missegmentation of gray-level overlapping areas in traditional single-threshold methods. Morphological post-processing was then performed on the initial segmented images to obtain separated component segmented images, eliminating noise and pores, accurately separating adhered particles, and improving the accuracy of microstructure characterization. Furthermore, based on the segmented images of each component, the difference between the image area ratio and the actual volume ratio of each component segmented image was calculated. Additionally, based on the segmented images of each component, the relative error between the image sieve gradation and the actual sieve gradation of the aggregate components was calculated. The difference reflects the reliability of the overall component proportions, while the relative error reflects the degree of reproduction of the aggregate gradation. When the difference value or relative error exceeds the preset threshold range, the image processing parameters are adjusted and the segmentation operation is re-executed to obtain the target segmented image. The difference value and relative error of the target segmented image are both within the preset threshold range. By utilizing the dual verification mechanism of difference value and relative error, the processing parameters are dynamically optimized to form a closed-loop feedback, ensuring that the segmentation accuracy of asphalt mixture components meets the requirements and providing reliable data support for mixture performance optimization. Attached Figure Description

[0056] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0057] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 This is a schematic flowchart of the first embodiment of the method for separating asphalt mixture components according to this application;

[0059] Figure 2 This is a flowchart illustrating the second embodiment of the method for separating asphalt mixture components according to this application;

[0060] Figure 3 This is a schematic diagram illustrating the separation effect before and after green filtering in an embodiment of this application;

[0061] Figure 4 This is a schematic diagram of grayscale processing before green filtering in an embodiment of this application;

[0062] Figure 5This is a schematic diagram of grayscale processing after green filtering in an embodiment of this application;

[0063] Figure 6 This is a schematic diagram showing the relationship between the preset multiple α (for 1.6) and the threshold spacing in an embodiment of this application.

[0064] Figure 7 This is a schematic diagram illustrating the background removal effect of an embodiment of this application;

[0065] Figure 8 This is a schematic diagram comparing grayscale images before and after background removal in an embodiment of this application;

[0066] Figure 9 This is a schematic diagram comparing the light field distribution thermal images before and after background removal in an embodiment of this application;

[0067] Figure 10 This is a schematic diagram of the three-component image after dual-threshold segmentation according to an embodiment of this application;

[0068] Figure 11 This is a flowchart illustrating the third embodiment of the method for separating asphalt mixture components according to this application;

[0069] Figure 12 This is a schematic diagram showing the relationship between the original gradation and the calculated two-dimensional gradation corresponding to Table 3 in the embodiments of this application;

[0070] Figure 13 This is a schematic diagram showing the relationship between the updated preset multiple α (for 2.05) and the threshold spacing in an embodiment of this application.

[0071] Figure 14 This is a schematic diagram illustrating the changes in the segmented image during image processing according to an embodiment of this application;

[0072] Figure 15 This is a schematic diagram of the modular structure of the asphalt mixture component segmentation device according to an embodiment of this application;

[0073] Figure 16 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the asphalt mixture component segmentation method in the embodiments of this application.

[0074] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0075] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0076] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0077] Due to the small grayscale differences and blurred, adherent boundaries among the components of asphalt mixtures, traditional image processing methods produce poor segmentation results, failing to meet the requirements for refined structural feature extraction and structural feature parameter calculation. Therefore, improving the segmentation accuracy of asphalt mixture components is a pressing issue that needs to be addressed.

[0078] Based on this, embodiments of this application provide a method for segmenting asphalt mixture components, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the method for segmenting asphalt mixture components according to this application.

[0079] In this embodiment, the method for segmenting the asphalt mixture components includes steps S10 to S70:

[0080] Step S10: Obtain an initial two-dimensional scan image of the asphalt mixture cross-section.

[0081] It should be noted that the initial two-dimensional scanned image can be a digital image of a cross-section of the asphalt mixture obtained through a high-precision scanner. For example, the asphalt mixture can be a standard asphalt mixture specimen, which can be prepared according to one or more methods such as the Marshall design method, the Superior Performing Asphalt Pavement (Superpave) design system, or the Course Aggregate Void Filling method (CAVF method), etc., without limitation here. The standard asphalt mixture gradation type can be continuously graded asphalt mixture (Asphalt Concrete, AC), stone mastic asphalt mixture (SMA), cast asphalt mixture (Gussasphalt, GT), etc., and the maximum nominal particle size can be common specifications such as 13mm and 16mm. For example, the standard asphalt mixture specimen is formed as a rotary compacted AC-13 asphalt mixture specimen (cylindrical) with a void ratio of 4%, with a specimen diameter of 150mm and a height of 115mm ± 5mm. After standard asphalt mixture specimens are formed, they can be cut. Cutting involves removing the areas with large porosity variations at the top and bottom of the cylindrical specimen, retaining only the central area with uniform porosity, and cutting it in half horizontally to obtain a more stable analytical cross-section. In practice, for example, for standard volume design specimens formed according to the Superpave system, 25mm can be removed from each of the top and bottom ends of the specimen, and considering blade wear of approximately 5mm, approximately 60mm of the central area can be retained and cut in half horizontally. The surface of the cut specimen is then treated, including rinsing to remove dust and impurities, and then polishing the surface sequentially with 400, 600, 800, 1000, and 3000 grit sandpaper to improve image clarity. After surface treatment, the images are scanned. The specific steps are as follows: Before scanning, the specimen is shielded from light using an internally blackened cover to block interference from external natural light sources and ensure image acquisition quality. Subsequently, a high-precision scanner is used to scan the asphalt mixture cross-section; four initial two-dimensional scan images can be obtained from one specimen. For example, the recommended resolution for the initial 2D scan image can be 1200 dpi, or 0.0211 mm / pixel.

[0082] Step S20: Preprocess the initial two-dimensional scan image to obtain a preprocessed image. The preprocessing includes contrast enhancement and light field correction.

[0083] It should be noted that preprocessing is used to enhance the sharpness of the initial 2D scanned image. Preprocessing includes contrast enhancement and light field correction. Contrast enhancement can widen the difference between the target area (such as aggregate or asphalt-covered areas) and the background by adjusting the grayscale distribution of the image. Light field correction can be understood as correcting the gradient phenomenon of bright centers and dark edges in the image caused by uneven illumination.

[0084] Step S30: The preprocessed image is divided into components using a dual threshold segmentation method to obtain an initial segmented image.

[0085] It should be noted that the dual-threshold segmentation method can be understood as using two thresholds (such as T1 and T2) to distinguish three types of regions in an image. These three types of regions can be divided into high grayscale (corresponding to aggregate regions), medium grayscale (corresponding to asphalt mortar regions), and low grayscale (corresponding to void regions). The initial segmented image can be a binary image to facilitate subsequent image analysis.

[0086] Step S40: Perform morphological post-processing on the initial segmented image to obtain segmented images of each separated component.

[0087] It should be noted that morphological post-processing can be understood as optimizing the boundaries of the initial segmented image to separate adhering particles. Specifically, this can be achieved by reducing the aggregate boundaries, segmenting overlapping aggregates, and restoring the original size of the aggregates.

[0088] Step S50: Based on the segmented images of each component, calculate the difference between the image area ratio and the actual volume ratio of each segmented image.

[0089] It should be noted that the image area ratio can be the proportion of pixels of each component (aggregate, asphalt, and voids) to the total number of pixels in the segmented image. The actual volume ratio can be the true volume ratio measured through physical experiments (such as extraction method or volumetric method). The difference value can be used to verify the overall distribution accuracy of the three components of aggregate, mortar, and voids.

[0090] Step S60: Based on the segmented images of each component, calculate the relative error between the image sieve distribution and the actual sieve distribution of the aggregate components.

[0091] It should be noted that sieving and grading can be understood as the mass distribution of aggregates with different particle sizes, and relative error can be understood as the deviation of the passing rate between the image-recognized grading and the actual sieving and grading.

[0092] Step S70: When the difference value or relative error exceeds the preset threshold range, adjust the image processing parameters and re-execute the segmentation operation to obtain the target segmented image. The difference value and relative error of the target segmented image are both within the preset threshold range.

[0093] It should be noted that the preset threshold range can be set based on engineering experience. Adjusting image processing parameters includes modifying at least one of the following: green channel enhancement coefficient, morphological operation structuring element size, and watershed minimum constraint parameter. The green channel enhancement coefficient is used to adjust the sharpness of the aggregate boundary, the morphological operation structuring element size can adjust the smoothness of the target segmented image, and the watershed minimum constraint parameter can constrain the number of segmented particles.

[0094] In this embodiment, after acquiring the initial two-dimensional scan image of the asphalt mixture cross-section, image quality is improved through contrast enhancement and light field correction. Combined with dual-threshold segmentation and morphological post-processing, high-precision separation of aggregates, mortar, and voids in the asphalt mixture is achieved. Furthermore, a closed-loop feedback system is formed by utilizing a dual verification mechanism of difference values ​​and relative errors. By dynamically adjusting the green channel gain, morphological kernel size, and watershed constraint parameters, the reliability and adaptability of the segmentation results are significantly improved. This not only meets the accuracy requirements of quantitative analysis of microstructure in road engineering but also provides reliable data support for mixture performance optimization.

[0095] Reference Figure 2 , Figure 2 This is a flowchart illustrating the second embodiment of the method for segmenting asphalt mixture components according to this application. Based on the above... Figure 1 The first embodiment shown presents a second embodiment of the method for segmenting asphalt mixture components according to this application.

[0096] In the second embodiment, step S20 includes:

[0097] Step S201: Based on the RGB color model, the initial two-dimensional scanned image is linearly enhanced by a preset multiple to obtain a contrast-enhanced image.

[0098] It should be noted that step S201 includes: based on the RGB color model, separating the RGB channels of the initial two-dimensional scanned image to obtain the green channel; linearly enhancing the green channel by a preset multiple to obtain the enhanced green channel; merging the enhanced green channel with the original red and blue channels to obtain a contrast-enhanced image.

[0099] It should be noted that the RGB (Red, Green, Blue) color model is the most basic additive color model in digital image processing, representing any color by mixing the three primary colors of red, green, and blue in different proportions. RGB channel separation can decompose a color image into three independent color component matrices: red (R), green (G), and blue (B), with each matrix storing the pixel values ​​of the corresponding channel.

[0100] For example, the extracted green channel image The overall pixel enhancement operation is performed according to the preset multiple α. The pixel values ​​in the G channel are proportionally amplified through a linear gain function, thereby enhancing the green band information in the image and improving the recognizability of targets with high green response features in the image. The enhanced G channel image is then fused with the R and B channels in the original image at the pixel level to construct a new enhanced RGB image, which highlights the green band information, effectively improves the edge clarity and contrast of the target components in the image, and generates a contrast-enhanced image suitable for subsequent segmentation processing. This is specifically expressed by formulas (1) to (5).

[0101]

[0102] In formulas (1) to (5), , , These are the color distribution functions for the three RGB channels separated from the image. This is the color distribution function of the G channel after channel expansion. Let be the overall RGB color distribution function of the original image. This is the overall RGB color distribution function of the image after channel enlargement. The separation effect before and after green filtering is shown below. Figure 4 As shown. For clarity of illustration, Figure 3 Including Figure 4 (Before green filter) and Figure 5 (After green filtering). Table 1 shows the threshold spacing as a function of the G channel expansion factor. α change.

[0103] Table 1

[0104]

[0105] As shown in Table 1, the fitting calculation α for the AC-13 gradation image was set to 1.6. The relationship curve between the preset multiple α and the threshold spacing is as follows. Figure 6 As shown.

[0106] Step S202: Perform light field correction on the contrast-enhanced image based on the grayscale mask algorithm to obtain the preprocessed image.

[0107] It should be noted that an initial mask is created in the contrast-enhanced image, and pixel regions with gray values ​​lower than the background threshold outside the mask are identified. The initial mask is used to cover the target region. The background noise region is determined based on the gray value threshold of the initial mask and the background gray value threshold. The background noise region is smoothed to obtain an intermediate mask. The smoothing process is used to eliminate edge artifacts. The effective image region is extracted based on the intermediate mask, and the invalid region is assigned an empty value to obtain the preprocessed image.

[0108] For example, the contrast-enhanced image is read into MATLAB software for preprocessing. The interactive mask drawing function `imellipse` is called to select the target region (ROI). A custom-shaped mask is created manually or automatically to cover the effective internal void area of ​​the asphalt mixture specimen slice image. The `createmask` function is used to convert this region into a logical matrix for subsequent background removal. For the outer region of the mask, a background grayscale threshold `T` is further used to identify and remove background noise areas with uneven grayscale distribution caused by uneven lighting, based on a set grayscale threshold `T` (which can be set to 55). Its logical judgment condition can be expressed as formula (6).

[0109]

[0110] Furthermore, the mask outside the mask and satisfying The pixel combination is used to construct the target composite background mask that needs to be removed. , Formula (7) must be satisfied.

[0111]

[0112] It should be noted that, considering the possibility of edge artifacts in the background area, the target area is removed by calling the strel and imidilate functions. Smoothing is performed to enhance the edge noise removal effect.

[0113] Furthermore, construct the final effective mask. It is used to accurately extract effective image information within the region of interest, and its logical expression is shown in formula (8).

[0114]

[0115] A new image is constructed based on this mask. The invalid region is set to NaN to achieve image standardization after light field correction and background removal, as shown in formula (9).

[0116]

[0117] In the above formula, The gray-level distribution function of the original image. This is a grayscale distribution function for the image after background noise removal using a custom multi-threshold method. Figure 7 A visualization of the effect after removing the background. Figure 8 This is a comparison diagram of grayscale images before and after background removal. Figure 9This is a schematic diagram comparing the light field distribution heatmaps before and after background removal.

[0118] In this embodiment, the contrast of the aggregate edge is significantly improved by selective enhancement of the green channel (solving the problem of gray-level overlap between components), and the effective area is accurately separated from the background by mask-guided noise removal: firstly, the noise pixels are located based on the spatial position (inside and outside the mask) and gray-level threshold (background threshold T), and then the boundary artifacts are eliminated by morphological smoothing. Finally, a pre-processed image with a clean background and balanced illumination is generated, which provides high-fidelity input for subsequent dual-threshold segmentation and fundamentally reduces the segmentation error caused by uneven illumination or background interference.

[0119] In one implementation, step S30 can be based on the maximum inter-class variance criterion, using a dual-threshold approach to divide the image grayscale distribution into background, intermediate, and foreground regions, corresponding to the three components of voids, asphalt mortar, and aggregate, respectively. By traversing combinations T1 and T2, the threshold pair that maximizes the inter-class variance is selected to achieve fast and accurate segmentation of the multi-components, resulting in an initial segmented image (the three-component image after dual-threshold segmentation). The separated three-component image is shown below. Figure 10 As shown.

[0120] In this embodiment, the separation degree between components is ensured by maximizing the inter-class variance, which completely solves the problem of misjudgment of adhesion at the interface between asphalt and aggregate in the traditional single threshold method. It can be adapted to different asphalt types and complex lighting environments, improves the segmentation stability, and enhances the versatility of the technology.

[0121] Reference Figure 11 , Figure 11 This is a flowchart illustrating the third embodiment of the method for segmenting asphalt mixture components in this application, based on the above. Figure 2 The second embodiment shown presents a third embodiment of the method for segmenting asphalt mixture components according to this application.

[0122] In the third embodiment, step S40 includes:

[0123] Step S401: Calculate the distance transformation matrix from foreground to background in the initial segmented image.

[0124] It should be noted that the distance transformation matrix is ​​a matrix generated by calculating the Euclidean distance from each foreground pixel to the nearest background pixel in the initial segmented image.

[0125] Step S402: Add a minimum value constraint marker to the distance transformation matrix. The minimum value constraint marker is used to suppress non-true local minima.

[0126] It should be noted that the minimum constraint label can be understood as a label map that prevents the watershed from being over-segmented by suppressing spurious local minimum points in the distance transformation matrix.

[0127] Step S403: Modify the distance transformation matrix based on the minimum constraint marker to generate a controlled distance transformation map.

[0128] It should be noted that the controlled distance transformation map can be understood as a modified distance map obtained by fusing the minimum value marker map with the original distance transformation matrix.

[0129] Step S404: Perform watershed transformation on the controlled distance transformation map to generate a particle segmentation boundary label matrix.

[0130] It should be noted that the watershed transformation treats the controlled distance map as a terrain surface and simulates the water injection process from the seed point (minimum value), with the water level confluence forming the dividing boundary.

[0131] Step S405: Based on the particle segmentation boundary label matrix, assign the segmentation boundary as the background and output the segmented images of each component.

[0132] It should be noted that boundary pixels with a value of 0 in the label matrix can be set as background values ​​(e.g., 0), while the original component labels are retained in non-boundary areas, and the separated component segmentation images are output.

[0133] For example, the `bwareaopen` function can be called on the initial segmented image to bidirectionally filter out white holes and black specks in the image based on the minimum recognition accuracy. It should be noted that the minimum recognition accuracy, i.e., the minimum accuracy that can be accurately identified in subsequent gradation verification, can be 635 pixels, with an equivalent diameter of approximately 0.6 mm. Then, the `imopen` function, in conjunction with a circular structuring element with a radius of 6 pixels, performs erosion and dilation operations on the image to further eliminate aggregate edge adhesion and unevenness. It should be noted that the circular structuring element with a radius of 6 pixels avoids excessive dilation that would result in a completely white image, and also avoids excessive erosion that would reduce the effective area of ​​the image, while ensuring the reasonableness of gradation errors during subsequent gradation verification. Furthermore, the `bwdist` function is used to calculate the negative distance transform of the image, combined with the `imextendedmin` function for minimum value constraint marking (the constraint parameter `x` can be set to 30, and flexibly adjusted according to the accuracy of the image verification results), and then the `imimposemin` function modifies the distance transform image. Finally, the `watershed` function is called to complete precise particle segmentation based on the watershed, extracting the final boundary line and assigning it as the background, achieving accurate particle separation.

[0134] In this embodiment, the spatial distribution of particles is quantified by distance transformation, and the seed point of the watershed is accurately located by combining minimum value constraint markers (suppressing false minimums caused by image noise or texture). A controlled distance map is generated to ensure that the watershed transformation only spreads from the effective center point. The precise separation of cohesive particles is achieved by using watershed boundary generation and background assignment, solving the problem of incomplete segmentation of dense aggregates by traditional morphological operations. This improves the statistical accuracy of microstructural parameters (such as aggregate particle size distribution and number of contact points), laying a technical foundation for the mechanical property analysis of asphalt mixtures.

[0135] In one implementation, based on the above embodiments, a specific calculation method is proposed for the difference value and relative error.

[0136] Specifically, the step of calculating the difference between the image area ratio and the actual volume ratio of each component segmented image based on the segmented images of each component includes: obtaining the number of pixels in each component segmented image to obtain the number of pixels in each component; calculating the proportion of the number of pixels in each component to the total number of pixels in the image to obtain the image area ratio of each component; calculating the actual volume ratio of each component based on the actual mass and density of each component of the asphalt mixture; and calculating the difference value based on the image area ratio and the actual volume ratio of each component.

[0137] For example, MATLAB software can be used to count the number of pixels in each segmented component region to obtain the total number of pixels for each part. Three typical regions include: aggregate region, asphalt mortar region, and void region. The number of pixels for each type of region... It can be calculated using formula (10).

[0138]

[0139] In formula (10), Let be the total number of pixels in the i-th class of components; For position The binary image value at that location; W , H The width and height of the image are segmented for each component. The number of pixels in each region is divided by the total number of pixels in the entire image. It is possible to calculate the area ratio of each component in the image. , which is represented by formula (11).

[0140]

[0141] Calculate the actual volume of each component based on the actual mass and density of the specimen. Volume of each component. The calculation is as shown in formula (12).

[0142]

[0143] In formula (12), Indicates the first i Actual mass (g) of the component class; This indicates the density of the component (g / cm3); This represents the actual volume (cm3) of the component. Furthermore, the formula for calculating the actual volume distribution ratio of each component is expressed as formula (13).

[0144]

[0145] The calculated difference values ​​are shown in Table 2.

[0146] Table 2

[0147]

[0148] Based on the results in Table 2, the recognition accuracy of image processing is verified by the following formula (14).

[0149]

[0150] In formula (14), the difference value It can be used to evaluate the differences in volume distribution of the i-th component (including aggregate, asphalt, and voids) using image recognition methods.

[0151] Specifically, the step of calculating the relative error between the image sieve distribution and the actual sieve distribution of aggregate components based on the segmented images of each component includes: calculating the conversion coefficient between pixel size and actual size based on the physical length and pixel distance of the scale in the segmented images of each component; calculating the minimum area threshold corresponding to the sieve aperture size according to the sieve aperture size and the conversion coefficient; screening aggregate particles step by step according to the minimum area threshold and calculating the equivalent diameter of each aggregate particle; calculating the cumulative pass rate of each sieve aperture size according to the equivalent diameter to obtain the image sieve distribution; and calculating the relative error between the three-dimensional actual sieve distribution and the image sieve distribution.

[0152] For example, a ruler object with a known physical length can be selected on each component segmented image, and the pixel length conversion index can be calculated. , as in formula (15).

[0153]

[0154] In formula (15), The physical length of the scale. This represents the pixel distance of the scale in the image. This conversion index is used to convert pixel units to actual millimeter units. The binarized image is filtered step-by-step according to an area threshold, based on the sieve aperture size. The corresponding minimum area threshold is calculated as shown in formula (16).

[0155]

[0156] It should be noted that particles smaller than the current sieve aperture size can be removed, and effective areas larger than that size can be extracted, and the sieving of all sieve apertures can be completed in a loop. Based on the selected minimum sieve aperture particle size, the particle area on the sieve is extracted, the area of ​​all connected regions is calculated, and converted into the actual area. The equivalent diameter of each particle can be calculated using the equivalent circle diameter formula, which is formula (17).

[0157]

[0158] In formula (17), L For the equivalent diameter, π Pi (π) is the mathematical constant for a circle. Based on the equivalent diameter... L With sieve aperture size The cumulative throughput of different sieve openings is calculated according to formulas (18) to (19).

[0159]

[0160] It should be noted that formula (18) is for the largest sieve aperture, and formula (19) is a calculation formula for the remaining sieve apertures. Among them, C can be defined as the total area of ​​all particles larger than 1.18 mm. It should be noted that 1.18 mm is one of the commonly used mortar boundary thresholds. Taking AC-13 as an example, when the minimum particle size is 0.6 mm, the relative error of the smaller sieve aperture is extremely large, while the relative error of different sieve apertures with a minimum particle size of 1.18 mm can be controlled within 10%. This indicates that the two-dimensional image is inaccurate in identifying fine aggregates smaller than 1.18 mm, which may be due to disturbances during cutting, noise in the image itself, and errors in image segmentation. Therefore, it is necessary to define the boundary of the inaccurate minimum particle size. The corresponding experimental data is shown in Table 3, and the relationship between the original gradation and the calculated two-dimensional gradation corresponding to Table 3 is shown in the figure. Figure 12 .

[0161] Table 3

[0162]

[0163] Furthermore, the recognition accuracy of image processing is verified by calculating the relative error formula using formula (20).

[0164]

[0165] Formula (20) is the formula for the relative error in calculating the gradation sieve aperture, where, For the first iThe relative error between the sieve aperture passing rate obtained from the two-dimensional image recognition of individual sieve apertures and the sieve aperture passing rate obtained from the three-dimensional actual gradation calculation. For the first two-dimensional image recognition i Passing rate of each sieve aperture; The first gradation obtained from the three-dimensional actual gradation calculation i The pass rate of each sieve hole.

[0166] In this embodiment, verification is performed in two ways: first, by comparing the volume distribution ratio calculated from the actual mass density; and second, by comparing it with sieve gradation data from traditional physical experiments. When the component difference value and the relative error of the gradation sieve aperture do not meet the threshold requirements, the image processing parameters (including the magnification factor of the G channel value, the size of the expansion and corrosion elements, and the watershed segmentation distance x, etc.) are adjusted in a timely manner until the error meets the requirements, thus obtaining an accurately segmented image. The image recognition difference value is less than 5%, and the relative error of gradation recognition is 10%.

[0167] To make the above embodiments clearer, the following experimental data are provided.

[0168] Specifically, Table 4 shows the calculation results of α calculated by fitting the SMA-13 ​​gradation image.

[0169] Table 4

[0170]

[0171] The relationship curve between the preset multiplier α and the threshold spacing is updated as follows: Figure 13 As shown. Figure 14 This diagram illustrates the changes in the segmented image during image processing in the above embodiments and implementation methods.

[0172] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the method of dividing the asphalt mixture components of this application. Any simple modifications based on this technical concept are within the scope of protection of this application.

[0173] This application also provides a device for separating asphalt mixture components, please refer to... Figure 15 The asphalt mixture component segmentation device includes:

[0174] The image acquisition module is used to acquire an initial two-dimensional scan image of the cross-section of the asphalt mixture;

[0175] The image preprocessing module is used to preprocess the initial two-dimensional scan image to obtain a preprocessed image. The preprocessing includes contrast enhancement and light field correction.

[0176] The initial segmentation module is used to divide the preprocessed image into components using a dual-threshold segmentation method to obtain an initial segmented image.

[0177] The morphological processing module is used to perform morphological post-processing on the initial segmented image to obtain segmented images of each separated component.

[0178] The difference calculation module is used to calculate the difference between the image area ratio and the actual volume ratio of each component segmented image based on the segmented images of each component.

[0179] The error calculation module is used to calculate the relative error between the image sieve distribution and the actual sieve distribution of the aggregate components based on the segmented images of each component.

[0180] The parameter adjustment module is used to adjust the image processing parameters and re-execute the segmentation operation when the difference value or the relative error exceeds a preset threshold range, thereby obtaining a target segmented image. The difference value and relative error of the target segmented image are both within the preset threshold range. The asphalt mixture component segmentation device provided in this application, employing the asphalt mixture component segmentation method in the above embodiments, can solve the technical problem of insufficient segmentation accuracy of asphalt mixture components. Compared with the prior art, the beneficial effects of the asphalt mixture component segmentation device provided in this application are the same as those of the asphalt mixture component segmentation method provided in the above embodiments, and other technical features in the asphalt mixture component segmentation device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0181] This application provides a device 2000 for dividing asphalt mixture components. The device 2000 includes: at least one processor 2001; and a memory 203 communicatively connected to the at least one processor 2001. The memory 2003 stores instructions that can be executed by the at least one processor 2001. The instructions are executed by the at least one processor 2001 to enable the at least one processor 2001 to perform the asphalt mixture component dividing method in the above embodiment 1.

[0182] The following is for reference. Figure 16 This diagram illustrates a structural schematic of an asphalt mixture component segmentation device 2000 suitable for implementing embodiments of this application. The asphalt mixture component segmentation device 2000 in this application embodiment may include a processor 2001 and a memory 2003. The processor 2001 and the memory 2003 are connected, for example, via a bus 2002. Optionally, the asphalt mixture component segmentation device 2000 may also include a transceiver 2004. It should be noted that in practical applications, the transceiver 2004 is not limited to one type, and the structure of this asphalt mixture component segmentation device 2000 does not constitute a limitation on the embodiments of this application.

[0183] Processor 2001 may be a CPU, a general-purpose processor, a DSP, an ASIC, an FPGA, or other programmable logic device, transistor logic device, hardware component, or any combination thereof. It may implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 2001 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0184] Bus 2002 may include a pathway for transmitting information between the aforementioned components. Bus 2002 may be a PCI bus or an EISA bus, etc. Bus 2002 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 16 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0185] The memory 2003 may be ROM or other type of static storage device capable of storing static information and instructions, RAM or other type of dynamic storage device capable of storing information and instructions, or EEPROM, CD-ROM or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0186] The memory 2003 is used to store the application code that executes the scheme of this application, and its execution is controlled by the processor 2001. The processor 2001 is used to execute the application code stored in the memory 2003.

[0187] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0188] The above description is only a partial embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for segmenting asphalt mixture components, characterized in that, The method for segmenting the asphalt mixture components includes: Obtain an initial two-dimensional scan image of the cross-section of the asphalt mixture; The initial two-dimensional scanned image is preprocessed to obtain a preprocessed image, the preprocessing including contrast enhancement and light field correction; Based on the maximum inter-class variance criterion, the image grayscale distribution is divided into background, middle region and foreground region by using dual thresholds. The preprocessed image is divided into three components: voids, asphalt mortar and aggregate, respectively. By traversing the first threshold combination and the second threshold combination, the threshold pair that maximizes the inter-class variance is selected to obtain the initial segmented image. Morphological post-processing is performed on the initial segmented image to obtain segmented images of each separated component; Based on the segmented images of each component, calculate the difference between the image area ratio and the actual volume ratio of each segmented image. Based on the segmented images of each component, the relative error between the image sieve distribution and the actual sieve distribution of the aggregate components is calculated; When the difference value or relative error exceeds the preset threshold range, the image processing parameters are adjusted and the segmentation operation is re-executed to obtain the target segmented image, wherein the difference value and relative error of the target segmented image are both within the preset threshold range. The step of calculating the relative error between the image sieve distribution and the actual sieve distribution of aggregate components based on the segmented images of each component includes: Based on the physical length and pixel distance of the scale in each component segmented image, calculate the conversion coefficient between pixel size and actual size; Calculate the minimum area threshold corresponding to the sieve aperture size based on the sieve aperture size and the conversion coefficient; Based on the minimum area threshold, aggregate particles are screened step by step, and the equivalent diameter of each aggregate particle is calculated. Based on the equivalent diameter, the cumulative throughput of each sieve aperture size is calculated to obtain the image sieve classification. Calculate the relative error between the actual three-dimensional sieve distribution and the image sieve distribution; The conversion coefficient between the pixel size and the actual size is expressed as follows: ； in, This is the conversion factor between pixel size and actual size. The physical length of the scale. The pixel distance of the ruler in the image; The minimum area threshold and equivalent diameter are respectively expressed as: ； ； in, Minimum area threshold, For the first i Each sieve aperture size, L For the equivalent diameter, π Pi A for L The corresponding pixel area; The cumulative throughput for the largest sieve aperture is expressed as: ； The recursive calculation formula for the remaining sieve apertures is expressed as follows: ； The relative error is expressed as: ； in, For the first i The relative error between the sieve aperture passing rate obtained from the two-dimensional image recognition of individual sieve apertures and the sieve aperture passing rate obtained from the three-dimensional actual gradation calculation. For two-dimensional image recognition Pass rate; The first gradation obtained from the three-dimensional actual gradation calculation i Passing rate of each sieve aperture; For the first i- One sieve aperture size, where C is the total area of ​​all particles larger than 1.18 mm.

2. The method as described in claim 1, characterized in that, The step of preprocessing the initial two-dimensional scanned image to obtain a preprocessed image includes: Based on the RGB color model, the initial two-dimensional scanned image is linearly enhanced by a preset multiple to obtain a contrast-enhanced image; The contrast-enhanced image is subjected to light field correction based on a grayscale mask algorithm to obtain a preprocessed image.

3. The method as described in claim 2, characterized in that, The step of linearly enhancing the initial two-dimensional scanned image by a preset multiple based on the RGB color model to obtain a contrast-enhanced image includes: Based on the RGB color model, the initial two-dimensional scanned image is subjected to RGB channel separation to obtain the green channel; The green channel is linearly enhanced by a preset multiple to obtain an enhanced green channel; The enhanced green channel is then merged with the original red and blue channels to obtain a contrast-enhanced image.

4. The method as described in claim 2, characterized in that, The step of performing light field correction on the contrast-enhanced image based on a grayscale mask algorithm to obtain a preprocessed image includes: An initial mask is created in the contrast-enhanced image, and pixel regions with gray values ​​below the background threshold outside the mask are identified. The initial mask is used to cover the target region. The background noise region is determined based on the grayscale threshold of the initial mask and the background grayscale threshold; The background noise region is smoothed to obtain an intermediate mask, and the smoothing process is used to eliminate edge artifacts. The effective image region is extracted based on the intermediate mask, and the invalid region is assigned an empty value to obtain the preprocessed image.

5. The method as described in claim 1, characterized in that, The step of performing morphological post-processing on the initial segmented image to obtain separate segmented images of each component includes: Calculate the foreground-to-background distance transformation matrix in the initial segmented image; Add a minimum constraint marker to the distance transformation matrix; the minimum constraint marker is used to suppress non-true local minima. Based on the minimum constraint marker, the distance transformation matrix is ​​modified to generate a controlled distance transformation graph; Perform a watershed transformation on the controlled distance transformation map to generate a particle segmentation boundary label matrix; Based on the particle segmentation boundary label matrix, the segmentation boundary is assigned as the background, and the segmented images of each component are output.

6. The method as described in claim 1, characterized in that, The step of calculating the difference between the image area ratio and the actual volume ratio of each component segmented image based on each component segmented image includes: Obtain the number of pixels in each segmented image to get the number of pixels in each component; Calculate the proportion of the number of pixels in each component to the total number of pixels in the image to obtain the area ratio of each component in the image; Calculate the actual volume percentage of each component based on the actual mass and density of each component in the asphalt mixture; The difference value is calculated based on the area ratio of each component image and the actual volume ratio of each component.

7. The method according to any one of claims 1 to 6, characterized in that, The adjustment of image processing parameters includes: Modify at least one of the following parameters: green channel enhancement coefficient, morphological operation structuring element size, and watershed minimum constraint parameter.

8. A device for dividing asphalt mixture components, characterized in that, The asphalt mixture component segmentation device includes: The image acquisition module is used to acquire an initial two-dimensional scan image of the cross-section of the asphalt mixture; The image preprocessing module is used to preprocess the initial two-dimensional scan image to obtain a preprocessed image. The preprocessing includes contrast enhancement and light field correction. The initial segmentation module is used to divide the image grayscale distribution based on the maximum inter-class variance criterion and using dual thresholds to divide the preprocessed image into background, middle region and foreground region, corresponding to the three components of voids, asphalt mortar and aggregate, respectively. By traversing the first threshold combination and the second threshold combination, the threshold pair that maximizes the inter-class variance is selected to obtain the initial segmented image. The morphological processing module is used to perform morphological post-processing on the initial segmented image to obtain segmented images of each separated component. The difference calculation module is used to calculate the difference between the image area ratio and the actual volume ratio of each component segmented image based on the segmented images of each component. The error calculation module is used to calculate the relative error between the image sieve distribution and the actual sieve distribution of the aggregate components based on the segmented images of each component. The parameter adjustment module is used to adjust the image processing parameters and re-execute the segmentation operation to obtain the target segmented image when the difference value or the relative error exceeds the preset threshold range. The difference value and relative error of the target segmented image are both within the preset threshold range. The error calculation module is also used to calculate the conversion coefficient between pixel size and actual size based on the physical length and pixel distance of the scale in each component segmented image; Calculate the minimum area threshold corresponding to the sieve aperture size based on the sieve aperture size and the conversion coefficient; Based on the minimum area threshold, aggregate particles are screened step by step, and the equivalent diameter of each aggregate particle is calculated. Based on the equivalent diameter, the cumulative throughput of each sieve aperture size is calculated to obtain the image sieve classification. Calculate the relative error between the actual three-dimensional sieve distribution and the image sieve distribution; The conversion coefficient between the pixel size and the actual size is expressed as follows: ； in, This is the conversion factor between pixel size and actual size. The physical length of the scale. The pixel distance of the ruler in the image; The minimum area threshold and equivalent diameter are respectively expressed as: ； ； in, Minimum area threshold, For the first i Each sieve aperture size, L For the equivalent diameter, π Pi A for L The corresponding pixel area; The cumulative throughput for the largest sieve aperture is expressed as: ； The recursive calculation formula for the remaining sieve apertures is expressed as follows: ； The relative error is expressed as: ； in, For the first i The relative error between the sieve aperture passing rate obtained from the two-dimensional image recognition of individual sieve apertures and the sieve aperture passing rate obtained from the three-dimensional actual gradation calculation. For the first two-dimensional image recognition i Passing rate of each sieve aperture; The first gradation obtained from the three-dimensional actual gradation calculation i Passing rate of each sieve aperture; For the first i- One sieve aperture size, where C is the total area of ​​all particles larger than 1.18 mm.

9. A device for separating asphalt mixture components, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the method for segmenting asphalt mixture components as claimed in any one of claims 1 to 7.

Images