A method for analyzing surface roughness of fine aggregate based on SEM image processing

Abstract

The application is suitable for the technical field of highway engineering, and provides a fine aggregate surface roughness analysis method based on SEM image processing, which comprises the following steps: obtaining a multi-scale SEM image of fine aggregate after standardization processing; generating a target gray image based on the multi-scale SEM image; wherein the target gray image is a gray image of an effective morphology characterization area reserved after eliminating edge interference areas and surface contamination areas; determining adhesion contact type fluctuation characteristics, mechanical interlocking type steep change characteristics and adsorption stable texture characteristics based on the target gray image; calculating roughness indexes based on the adhesion contact type fluctuation characteristics, the mechanical interlocking type steep change characteristics and the adsorption stable texture characteristics; and generating fine aggregate surface roughness analysis results according to the roughness indexes; the application provides accurate data basis for asphalt mixture design, effectively improves the evaluation accuracy of aggregate and asphalt adhesion performance, and thus effectively prevents early diseases of the pavement.

Description

Technical Field

[0001] This invention relates to the field of highway engineering technology, and in particular to a method for analyzing the surface roughness of fine aggregates based on SEM image processing. Background Technology

[0002] Asphalt pavement, due to its comfort and ease of maintenance, has become the primary structure of highways. However, with increasing service life, it commonly exhibits defects such as cracks and rutting. The main causes are concentrated in construction quality, environmental factors, and load effects, with construction quality being the most critical. Raw material control is paramount. In some areas, early defects in newly constructed pavements are caused by a shortage of high-quality aggregates, as the aggregates used in early construction had relatively superior performance. However, with prolonged mining, the output of high-quality aggregates has decreased. Although most aggregates meet specifications, their index values ​​have declined to varying degrees. Visual inspection and XRD lithology analysis both reveal significant fluctuations and variability in the quality of aggregates produced in recent years. This leads to reduced adhesion and compatibility between aggregates and asphalt, resulting in early defects in asphalt pavements. Existing technologies offer limited testing on the adhesion between asphalt and aggregates, mainly including low-temperature bonding tests and adhesion tests between asphalt and coarse aggregates. These tests primarily provide subjective evaluations of asphalt adhesion from a macroscopic perspective. Currently, there are no quantitative analytical indicators for the adhesion between asphalt and aggregates, making it impossible to reflect the true adhesion performance.

[0003] Therefore, a method for analyzing the surface roughness of fine aggregates based on SEM image processing is proposed. Summary of the Invention

[0004] This invention provides a method for analyzing the surface roughness of fine aggregates based on SEM image processing, which addresses the problem that there are no quantitative analytical indicators for the adhesion between asphalt and aggregates, thus failing to reflect the true adhesion performance.

[0005] This invention provides a method for analyzing the surface roughness of fine aggregates based on SEM image processing, comprising: Obtain multi-scale SEM images of the standardized fine aggregate; A target grayscale image is generated based on the multi-scale SEM image; wherein, the target grayscale image is a grayscale image of the effective morphological characterization region retained after removing edge interference regions and surface contamination regions; Based on the target grayscale image, determine the adhesive contact type undulation feature, the mechanical interlocking type steep change feature, and the adsorption stable type texture feature; Roughness index is calculated based on the adhesive contact type undulation feature, mechanical interlocking type steep change feature and adsorption stable type texture feature; The surface roughness analysis results of the fine aggregate are generated based on the roughness index.

[0006] Furthermore, generating the target grayscale image based on the multi-scale SEM image includes: The edge gradient magnitude and edge gradient direction dispersion of each pixel in a multi-scale SEM image are extracted using a multi-scale edge detection operator. Based on the edge gradient magnitude and edge gradient direction dispersion, each pixel is clustered and grouped to obtain several edge feature groups; The target group located outside the outline of the aggregate particles and discontinuous with the main texture in the plurality of edge feature groups is marked as an edge interference region; wherein, the edge gradient magnitude of the pixels in the target group is lower than a first threshold, and / or the edge gradient direction dispersion of the pixels in the target group is higher than a second threshold.

[0007] Furthermore, the step of generating a target grayscale image based on the multi-scale SEM image further includes: A surface contamination texture feature library is constructed, which includes gray-level co-occurrence matrix features of typical contaminants in SEM images; The multi-scale SEM image is divided into blocks, and the contrast, energy, and homogeneity index of each image block are extracted as the current texture features. Calculate the similarity between the current texture feature and the pollutant texture features in the feature library, and mark image patches with similarity exceeding a preset threshold as surface contamination areas.

[0008] Furthermore, the step of generating a target grayscale image based on the multi-scale SEM image further includes: The edge interference region and the surface contamination region are merged into an invalid region mask; Remove the pixels covered by the invalid region mask from the multi-scale SEM image of the fine aggregate after the standardization process; The empty areas after pixel removal are filled with gray values ​​of adjacent effective texture areas to obtain continuous effective morphological representation areas.

[0009] Furthermore, the adhesive contact-type undulation features are used to quantify the contribution of surface unevenness to the asphalt contact area, including: fluctuation amplitude The calculation formula is: Fluctuation density The calculation formula is: in: Characterizing the average height difference between surface peaks and valleys, To effectively characterize the number of peak-valley pairs detected within the region, For the first The grayscale value of each surface peak. For adjacent The grayscale value of each surface valley; Characterizes the number of concave and convex structures per unit area. This represents the total pixel area of ​​the effective topographic representation region.

[0010] Furthermore, the mechanically interlocking abrupt change feature is used to quantify the effect of the angle on the mechanical interlocking of asphalt, including: Sharpness of edges The calculation formula is: Percentage of areas with abrupt changes The calculation formula is: in: The peak value representing the local gray-level gradient. For effective morphological characterization region, For position The gradient magnitude of a pixel; Characterizes the proportion of pixels with significant grayscale abrupt changes. This represents the number of pixels whose gradient magnitude exceeds a preset sharpness threshold. The total number of pixels in the effective topographic representation region.

[0011] Furthermore, the gradient magnitude in the sharpness of the angle is calculated using the Sobel operator, including: in, for The result of directional gradient convolution, for Result of directional gradient convolution; The sharpness threshold of the percentage of the abrupt change region. Set as: in: The mean of the gradient magnitude within the effective region. This represents the standard deviation of the gradient magnitude within the effective region.

[0012] Furthermore, the adsorption-stabilized texture feature is used to quantify the influence of microtexture on the uniformity of asphalt adsorption, including: Texture complexity The calculation formula is: Microtexture distribution uniformity The calculation formula is: in: Characterizes the degree of disorder in microtexture. The number of gray levels in the image. For the gray-level co-occurrence matrix at the offset The probability value at that location; Characterizes the stability of asphalt adhesion. The number of character blocks that constitute the effective region. For the first Local texture entropy of each segmented region This represents the mean of the local texture entropy across all segmented regions.

[0013] Furthermore, the calculation of roughness indices based on the adhesive contact-type undulation features, mechanically interlocking-type steep change features, and adsorption-stable texture features includes: in: This refers to the roughness index value; For the sub-index values ​​of adhesion contact-type undulation characteristics, , and These are the fluctuation amplitudes. and undulation density The normalized value, and These are the corresponding weight coefficients. ; For mechanically interlocked abrupt change characteristics, the sub-index value is... , and Angular sharpness and the proportion of abruptly changing areas The normalized value, and For the corresponding weighting coefficients, ; This refers to the sub-index value of adsorption-stable texture features. , and Texture complexity and uniformity of microtexture distribution The normalized value, and For the corresponding weighting coefficients, ; , , The weighting coefficients for the sub-indicator values ​​satisfy the following conditions: .

[0014] Furthermore, generating the surface roughness analysis results of the fine aggregate based on the roughness index includes: The roughness index value is mapped to five levels of fine aggregate surface roughness, including extremely smooth surface, relatively smooth surface, medium rough surface, relatively rough surface and extremely rough surface. The mapping relationship satisfies the monotonically increasing property: the larger the roughness index value, the higher the surface roughness.

[0015] As can be seen from the above technical solutions, the present invention has the following advantages: This invention, based on multi-scale SEM images, uses image processing techniques to remove edge interference areas and surface contamination areas to obtain a target grayscale image, ensuring the integrity of the effective morphology representation area. From the target grayscale image, it extracts adhesion contact-type undulation features, mechanical interlocking-type steep change features, and adsorption stability-type texture features related to asphalt adhesion, avoiding the limitations of traditional single-indicator models. Based on these three types of features, a comprehensive roughness index is calculated. The resulting roughness index reflects the influence of the microstructure of fine aggregate surfaces on asphalt adhesion performance. Finally, the index is mapped to a five-level roughness analysis result, providing accurate data for asphalt mixture design, effectively improving the accuracy of aggregate-asphalt adhesion performance evaluation, and thus effectively preventing early pavement distress. Attached Figure Description

[0016] Figure 1 This is a schematic flowchart of an embodiment of a method for analyzing the surface roughness of fine aggregates based on SEM image processing according to the present invention; Figure 2 This is a schematic flowchart of another embodiment of the fine aggregate surface roughness analysis method based on SEM image processing in this invention; Figure 3 This is a schematic flowchart of another embodiment of the fine aggregate surface roughness analysis method based on SEM image processing in this invention; Figure 4 This is a schematic flowchart of another embodiment of the fine aggregate surface roughness analysis method based on SEM image processing in this invention; Figure 5 This is a schematic diagram of the untreated diabase and limestone used in this invention; Figure 6 This is a schematic diagram of diabase and limestone after leveling and cleaning in this invention; Figure 7 This is a schematic diagram of SEM scans of diabase and limestone after gold plating in this invention. Figure 8 This is a schematic diagram of diabase and limestone representing the effective morphological region in this invention. Detailed Implementation

[0017] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “corresponding to,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0018] Example 1 The implementation method in this embodiment can be implemented in a system, on a server, or on a terminal; no specific limitation is made. The following section will describe the fine aggregate surface roughness analysis method based on SEM image processing in this application from the perspective of system implementation. Please refer to... Figures 1 to 4 The method provided in this application includes the following steps: S1. Obtain multi-scale SEM images of the standardized fine aggregate; In this embodiment, the fine aggregate particles to be detected are pre-treated, including surface cleaning to remove surface dust contaminants; gold film is sprayed to improve conductivity and reduce charge accumulation during SEM imaging; a field emission scanning electron microscope is used, with a fixed accelerating voltage and working distance, and the magnification is adjusted according to the gradient, acquiring at least three sets of image sequences at different locations for each particle to ensure surface representativeness. Next, the original SEM images are histogram-defined to unify all images to the same grayscale distribution, eliminating brightness differences; pixel resolution is calibrated using a standard scale sample to ensure accurate size conversion at different magnifications; and multi-scale images of the same region are aligned using SIFT feature point matching to establish a spatial coordinate mapping relationship.

[0019] The standardization process described above can resolve grayscale deviations and scale distortions in SEM images caused by differences in device parameters and environment, ensuring the objectivity of subsequent analysis. It generates standardized image data that conforms to the input of edge detection and texture analysis algorithms, avoiding misjudgments of features due to differences in image quality.

[0020] S2. Generate a target grayscale image based on multi-scale SEM images; wherein, the target grayscale image is a grayscale image of the effective morphological characterization region retained after removing edge interference regions and surface contamination regions; In practical applications, due to the presence of charge buildup, sample preparation scratches, and surface contaminants in the original SEM images, coupled with the limitations of grayscale algorithms in identifying interference areas, direct grayscale conversion can lead to two types of problems: non-aggregate surface structures are misidentified as valid textures, and contaminant areas distort the true surface morphology. Therefore, edge interference areas and surface contamination areas are removed during the grayscale conversion process to extract the valid areas.

[0021] In this embodiment, the removal of edge interference regions is achieved through the following steps: S211. Extract the edge gradient magnitude and edge gradient direction dispersion of each pixel in a multi-scale SEM image using a multi-scale edge detection operator; S212. Cluster and group each pixel based on the edge gradient magnitude and edge gradient direction dispersion to obtain several edge feature groups; S213. Mark the target group located outside the outline of the aggregate particles and discontinuous with the main texture in a number of edge feature groups as edge interference regions; wherein the edge gradient magnitude of the pixels in the target group is lower than the first threshold, and / or the edge gradient direction dispersion of the pixels in the target group is higher than the second threshold.

[0022] Specifically, in the scenario of evaluating the roughness of fine aggregates, the actual aggregate profile has a high gradient magnitude and low directional dispersion. Therefore, the two dimensions of edge gradient magnitude and edge gradient directional dispersion are used to jointly separate the two types of interference.

[0023] First, a three-scale Sobel operator (3×3, 5×5, 7×7) is used to perform convolution operations on the SEM image to calculate the gradient magnitude and orientation angle of each pixel. Then, each pixel in the multi-scale SEM image is represented as a feature vector. (Normalized to the [0,1] interval); Clustering using the DBSCAN algorithm, setting the neighborhood radius. , minimum number of points ,generate Each edge feature group is a set of spatially continuous pixels with similar features.

[0024] Valid group: High gradient magnitude ( And low directional dispersion ( ), corresponding to the true outline of the aggregate; Interference group: low gradient amplitude ( or high directional dispersion ), such as scratches and noise.

[0025] After determining the edge feature groups, groups that simultaneously meet the following three conditions are marked as edge interference regions: 1. Located outside the contour: the distance from the centroid of the group to the geometric center of the aggregate is greater than 0.8 times the equivalent radius of the particle; 2. Discontinuous with the main texture: the Hausdorff distance between the group boundary and the main texture region is greater than 5 pixels; 3. Meets one of the threshold conditions: the first condition is that the edge gradient amplitude is lower than the first threshold, indicating that the edge intensity is too weak and is an artifact; the second condition is that the edge gradient direction dispersion is higher than the second threshold, indicating that the edge direction is messy and is noise interference. It should be noted that "and / or" here means that the determination is triggered when either threshold condition is met.

[0026] In this embodiment, the removal of surface contamination areas is achieved through the following steps: S221. Construct a surface contamination texture feature library, which includes the gray-level co-occurrence matrix features of typical contaminants in SEM images; S222. Perform block processing on the multi-scale SEM image, and extract the contrast, energy and homogeneity index of each image block as the current texture feature; S223. Calculate the similarity between the current texture feature and the pollutant texture features in the feature library, and mark image blocks with similarity exceeding a preset threshold as surface contamination areas.

[0027] This embodiment achieves contaminated area identification by constructing a surface contamination texture feature library. This library is designed for typical contaminants on fine aggregate surfaces, including dust (particle size <5μm), grease (rut residue), salt crystals (de-icing agent residue), and organic impurities (plant debris). Specifically, texture features of contaminants are collected in high-magnification SEM images (×5000). A gray-level co-occurrence matrix is ​​calculated for each type of contaminant image, and three texture indices—contrast, energy, and homogeneity—are extracted. Contrast quantifies the intensity of local gray-level changes, reflecting the difference in contrast between the contaminant and the aggregate matrix; energy describes texture uniformity, with contaminant energy values ​​significantly higher than the natural texture of the aggregate; and homogeneity characterizes structural regularity, with crystalline contaminants exhibiting high homogeneity. The establishment of this feature library provides a benchmark for contamination discrimination and solves the problem of statistical separability between contaminants and the bulk texture of fine aggregate surfaces.

[0028] In the block processing stage, the SEM image is divided into 32×32 pixel sub-blocks (covering the smallest contaminant size). Contrast, energy, and homogeneity indicators are extracted for each block, and feature vectors are constructed. The Euclidean distance between the feature vector of each image block and the contaminant vector in the feature library is calculated. When the distance between the feature vector and the contaminant vector of a certain type is less than a preset threshold, it indicates that the texture features of the image block are highly similar to those of the contaminant. At this time, the image block is marked as a surface contamination area, and fragmented misjudged areas are optimized through morphological closing operations.

[0029] Based on the above-described determination of edge interference regions and pollution regions, the process of generating effective regions is as follows: S231. Merge the edge interference area and the surface contamination area into an invalid area mask; S232. Remove the pixels covered by the invalid region mask from the multi-scale SEM image of the standardized fine aggregate; S233. Fill the empty areas after removing pixels with the gray values ​​of adjacent effective texture areas to obtain continuous effective morphological representation areas.

[0030] The edge interference region determined in S213 is spatially superimposed with the surface contamination region marked in S223 to generate a unified invalid region mask. Based on this mask, all covered pixels are removed from the standardized multi-scale SEM image to ensure that the invalid region is completely removed. Finally, for the empty regions generated by the removal, the adaptive weighted average of the pixel gray values ​​in the adjacent effective texture region (within a range of 5 pixels outside the boundary) is used to fill them (the weight decreases with distance), thereby eliminating image holes and reconstructing the continuity of surface morphology. The final output is a seamless target grayscale image that contains only the aggregate body structure information, i.e., a continuous effective morphology representation region.

[0031] S3. Determine the adhesive contact type undulation feature, the mechanical interlocking type steep change feature, and the adsorption stable type texture feature based on the target grayscale image; To quantify the contribution of fine aggregate surface roughness to asphalt adhesion performance, complementary features were extracted from three physical action dimensions based on the target grayscale image containing only the effective morphological characterization region generated by S2.

[0032] I. Adhesive contact-type undulation characteristics are used to quantify the contribution of surface unevenness to the asphalt contact area, including: fluctuation amplitude The calculation formula is: Fluctuation density The calculation formula is: in: Characterizing the average height difference between surface peaks and valleys, To effectively characterize the number of peak-valley pairs detected within the region, For the first The grayscale value of each surface peak. For adjacent The grayscale value of each surface valley; Characterizes the number of concave and convex structures per unit area. This represents the total pixel area of ​​the effective topographic representation region.

[0033] II. Mechanically interlocking abrupt change characteristics are used to quantify the effect of sharp angles on the mechanical interlocking of asphalt, including: Sharpness of edges The calculation formula is: Percentage of areas with abrupt changes The calculation formula is: in: The peak value representing the local gray-level gradient. For effective morphological characterization region, For position The gradient magnitude of a pixel; Characterizes the proportion of pixels with significant grayscale abrupt changes. This represents the number of pixels whose gradient magnitude exceeds a preset sharpness threshold. The total number of pixels in the effective topographic representation region.

[0034] The gradient magnitude in the corner sharpness is calculated using the Sobel operator, including: in, for The result of directional gradient convolution, for Result of directional gradient convolution; The sharpness threshold of the percentage of the abrupt change region. Set as: in: The mean of the gradient magnitude within the effective region. This represents the standard deviation of the gradient magnitude within the effective region.

[0035] III. Adsorption-stabilized texture features are used to quantify the influence of microtexture on the uniformity of asphalt adsorption, including: Texture complexity The calculation formula is: Microtexture distribution uniformity The calculation formula is: in: Characterizes the degree of disorder in microtexture. The number of gray levels in the image. For the gray-level co-occurrence matrix at the offset The probability value at that location; Characterizes the stability of asphalt adhesion. The number of character blocks that constitute the effective region. For the first Local texture entropy of each segmented region This represents the mean of the local texture entropy across all segmented regions.

[0036] S4. Calculate roughness index based on adhesive contact type undulation characteristics, mechanical interlocking type steep change characteristics and adsorption stable type texture characteristics; The above three types of characteristics are uniformly quantified into a comprehensive roughness index. First, the sub-index values ​​of each sub-index, namely the sub-index values ​​of the adhesive contact type undulation characteristics, are calculated separately. Sub-index values ​​of mechanically interlocked abrupt change characteristics Sub-index values ​​of adsorption-stable texture features Then, the sub-indicator values ​​are globally merged, and the calculation formula is as follows: in: This refers to the roughness index value; , and These are the fluctuation amplitudes. and undulation density The normalized value, and These are the corresponding weight coefficients. ; , and Angular sharpness and the proportion of abruptly changing areas The normalized value, and For the corresponding weighting coefficients, ; , and Texture complexity and uniformity of microtexture distribution The normalized value, and For the corresponding weighting coefficients, ; , , The weighting coefficients for the sub-indicator values ​​satisfy the following conditions: .

[0037] S5. Generate the surface roughness analysis results of fine aggregates based on the roughness index.

[0038] Roughness index value calculated based on S4 They are divided into five levels with clear engineering significance through a pre-defined monotonically increasing mapping relationship: Extremely smooth surface ( The surface lacks significant undulations and sharp edges, resulting in extremely poor asphalt adhesion and a high risk of peeling. Smoother surfaces ): The presence of weak textures in certain areas makes the asphalt film prone to shear slip. Medium roughness surface ( It combines a basic concave-convex structure with moderately sharp edges; Rougher surfaces ( Dense peaks and valleys and sharp edges significantly enhance mechanical interlocking. Extremely rough surface ( Highly complex microtextures form strong adsorption and anchoring.

[0039] The grading results are directly output as a roughness analysis report containing grade labels and risk warnings, such as the analysis results for rougher surfaces; its monotonically increasing characteristic ( (The larger the value, the higher the roughness) ensures the physical consistency and engineering interpretability of the evaluation results.

[0040] The above implementation method is verified below with a specific experimental process: 1. All aggregates used are derived from two types of lithological aggregates: diabase and limestone, both native to Guangxi. Figure 5 As shown, Figure 5 The left image shows diabase. Figure 5 The image on the right shows limestone; the particle size is 2.36mm, the surface of the aggregate shows no signs of artificial treatment, the texture is uniform, there are no impurities visible to the naked eye, and the quantity is 3 particles; 2. Take any one side of the selected 3 aggregates as the bottom surface and use a file to level the bottom surface of the aggregates; 3. For example Figure 6 As shown, Figure 6 The left image shows diabase. Figure 6 The right image shows limestone. After leveling the bottom surface, rinse the three aggregates with clean water and place them in a clean container. Then place them together in a constant temperature oven, adjust the oven temperature to 180℃, heat for more than 4 hours, and then take them out to cool for later use. 4. Place the 3 cleaned aggregates into the SEM plating equipment, with the bottom of the aggregates facing the base of the plating equipment; 5. For example Figure 7 As shown, Figure 7 The left image shows diabase. Figure 7The right image shows limestone. Since the asphalt film thickness of dense-graded rubber asphalt mixtures is generally 6-8 μm, a SEM scan magnification of 3000× was selected, with an image unit size of 10 μm. When using SEM to observe the surface structure of the aggregates, the SEM instrument was turned on to scan three gold-plated aggregates. The scan magnification was set to 3000×, and the scan unit size was 10 μm. Three non-overlapping scan points were selected for each of the three aggregates, for a total of nine aggregate points scanned. Additionally, a ×200 (100 μm field of view) macroscopic contour image and a ×5000 (2 μm field of view) microscopic texture image were acquired to form a complete multi-scale dataset. 6. Perform the following on the 3000× image: Edge interference removal: Scratches on the limestone surface are identified as interference due to their low gradient amplitude; Contaminated area identification: Dust clumps on the diabase surface with energy values ​​>0.85 are marked as contaminated; Invalid area repair: Generate continuous and effective topographic regions, such as... Figure 8 As shown, Figure 8 The left image shows diabase. Figure 8 The image on the right shows limestone.

[0041] 7. Following the steps outlined above, the two types of lithological aggregates were processed. The processed data is shown in the table below: Table 1. Quantification results of fine aggregate surface roughness based on multiple feature indices The data in the table shows that limestone has significantly higher undulation density (34.5%↑213.6%), a higher percentage of abrupt change zones (45.9%↑124%), and higher uniformity (0.85%↑67%) than diabase. Limestone roughness index Comprehensive roughness index of diabase According to the roughness index classification, limestone was rated as "extremely rough surface," while diabase was rated as "medium rough surface," indicating that the limestone surface is rougher. This classification result shows that limestone has a better roughness than diabase, providing a reliable and comprehensive data basis for a comprehensive quantitative study of the interfacial adhesion between asphalt and aggregates. Furthermore, from the perspective of aggregate micromorphology, limestone mainly exhibits a near-ellipsoidal structure, while diabase is predominantly lamellar. The latter results in a smaller effective contact area with asphalt, thus weakening the asphalt-aggregate interface adhesion, further indicating that the interfacial adhesion between asphalt and limestone aggregates is superior.

[0042] It is understood that those skilled in the art can combine various implementation methods in the above embodiments under the guidance of the above examples to obtain technical solutions with multiple implementation methods.

[0043] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for analyzing surface roughness of fine aggregate based on SEM image processing, characterized by, include: Obtain multi-scale SEM images of the standardized fine aggregate; A target grayscale image is generated based on the multi-scale SEM image; wherein, the target grayscale image is a grayscale image of the effective morphological characterization region retained after removing edge interference regions and surface contamination regions; Based on the target grayscale image, determine the adhesive contact type undulation feature, the mechanical interlocking type steep change feature, and the adsorption stable type texture feature; The adhesive contact-type undulation features are used to quantify the contribution of surface unevenness to the asphalt contact area, including: amplitude of the undulation The calculation formula is: Fluctuation density The calculation formula is: in: Characterizing the average height difference between surface peaks and valleys, To effectively characterize the number of peak-valley pairs detected within the region, For the first The grayscale value of each surface peak. For adjacent The grayscale value of each surface valley; Characterizes the number of concave and convex structures per unit area. The total pixel area of ​​the effective topographic representation region; The mechanically interlocking abrupt change feature is used to quantify the effect of the angle on the mechanical interlocking of asphalt, including: Sharpness of edges The calculation formula is: Percentage of areas with abrupt changes The calculation formula is: in: The peak value representing the local gray-level gradient. For effective morphological characterization region, For position The gradient magnitude of a pixel; Characterizes the proportion of pixels with significant grayscale abrupt changes. This represents the number of pixels whose gradient magnitude exceeds a preset sharpness threshold. The total number of pixels in the effective topographic representation region; The adsorption-stabilized texture features are used to quantify the influence of microtexture on the uniformity of asphalt adsorption, including: Texture complexity The calculation formula is: Microtexture distribution uniformity The calculation formula is: in: Characterizing the disorder of microtextures, The number of gray levels in the image. For the gray-level co-occurrence matrix at the offset The probability value at that location; Characterizes the stability of asphalt adhesion. The number of character blocks that constitute the effective region. For the first Local texture entropy of each segmented region The mean of the local texture entropy across all segmented regions; Roughness index is calculated based on the adhesive contact type undulation feature, mechanical interlocking type steep change feature and adsorption stable type texture feature; The surface roughness analysis results of the fine aggregate are generated based on the roughness index.

2. The method for analyzing the surface roughness of fine aggregates based on SEM image processing according to claim 1, characterized in that, The process of generating a target grayscale image based on the multi-scale SEM image includes: The edge gradient magnitude and edge gradient direction dispersion of each pixel in a multi-scale SEM image are extracted using a multi-scale edge detection operator. Based on the edge gradient magnitude and edge gradient direction dispersion, each pixel is clustered and grouped to obtain several edge feature groups; The target group located outside the outline of the aggregate particles and discontinuous with the main texture in the plurality of edge feature groups is marked as an edge interference region; wherein, the edge gradient magnitude of the pixels in the target group is lower than a first threshold, and / or the edge gradient direction dispersion of the pixels in the target group is higher than a second threshold.

3. The method for analyzing the surface roughness of fine aggregates based on SEM image processing according to claim 1, characterized in that, The process of generating a target grayscale image based on the multi-scale SEM image further includes: A surface contamination texture feature library is constructed, which includes gray-level co-occurrence matrix features of typical contaminants in SEM images; The multi-scale SEM image is divided into blocks, and the contrast, energy, and homogeneity index of each image block are extracted as the current texture features. Calculate the similarity between the current texture feature and the pollutant texture features in the feature library, and mark image patches with similarity exceeding a preset threshold as surface contamination areas.

4. The method for analyzing the surface roughness of fine aggregates based on SEM image processing according to any one of claims 1-3, characterized in that, The process of generating a target grayscale image based on the multi-scale SEM image further includes: The edge interference region and the surface contamination region are merged into an invalid region mask; Remove the pixels covered by the invalid region mask from the multi-scale SEM image of the fine aggregate after the standardization process; The empty areas after pixel removal are filled with gray values ​​of adjacent effective texture areas to obtain continuous effective morphological representation areas.

5. The method for analyzing the surface roughness of fine aggregates based on SEM image processing according to claim 1, characterized in that, The gradient magnitude in the sharpness of the edges is calculated using the Sobel operator, including: in, for The result of directional gradient convolution, for Result of directional gradient convolution; The sharpness threshold of the percentage of the abrupt change region. Set as: in: The mean of the gradient magnitude within the effective region. This represents the standard deviation of the gradient magnitude within the effective region.

6. The method for analyzing the surface roughness of fine aggregates based on SEM image processing according to claim 1, characterized in that, The roughness index is calculated based on the adhesive contact type undulation characteristics, mechanical interlocking type steep change characteristics, and adsorption stable type texture characteristics, including: in: This refers to the roughness index value; For the sub-index values ​​of adhesion contact-type undulation characteristics, , and These are the fluctuation amplitudes. and undulation density The normalized value, and These are the corresponding weight coefficients. ; For mechanically interlocked abrupt change characteristics, the sub-index value is... , and Angular sharpness and the proportion of abruptly changing areas The normalized value, and For the corresponding weighting coefficients, ; This refers to the sub-index value of adsorption-stable texture features. , and Texture complexity and uniformity of microtexture distribution The normalized value, and For the corresponding weighting coefficients, ; , , The weighting coefficients for the sub-indicator values ​​satisfy the following conditions: .

7. The method for analyzing the surface roughness of fine aggregates based on SEM image processing according to claim 6, characterized in that, The step of generating fine aggregate surface roughness analysis results based on the roughness index includes: The roughness index value is mapped to five levels of fine aggregate surface roughness, including extremely smooth surface, relatively smooth surface, medium rough surface, relatively rough surface and extremely rough surface. The mapping relationship satisfies the monotonically increasing property: the larger the roughness index value, the higher the surface roughness.

Images