A Physically Enhanced Asphalt Concrete Optical Image Acquisition and Segmentation Method

Abstract

A physical color enhancement-based optical image acquisition and segmentation method for asphalt concrete addresses problems in existing methods, such as unstable image quality due to uneven ambient lighting, difficulty in distinguishing the colors of aggregates and asphalt mortar, lack of pore geometric information, and insufficient generalization ability of segmentation algorithms. This method involves preparing a physical light shield to isolate ambient light, utilizing the difference in hydrophilicity and hydrophobicity between aggregates and asphalt mortar after washing and drying to enhance color contrast, and selectively filling pores with a white powder colorant to impart white characteristics to the pore phase. Two optical image acquisitions are performed. The aggregate phase image is grayscaled using the maximum channel value principle followed by threshold segmentation, while the pore phase image is grayscaled using the average channel value principle followed by threshold segmentation. Finally, a bilinear interpolation algorithm is used to unify pixel specifications, correct pixel values, and perform Boolean operations to obtain a complete cross-sectional composite image containing the aggregate phase, pore phase, and asphalt mortar phase.

Description

Technical Field

[0001] This application relates to the field of optical image acquisition and processing technology, and in particular to a method for acquiring and segmenting optical images of asphalt concrete based on physical color enhancement. Background Technology

[0002] Optical image acquisition and segmentation of asphalt concrete cross-sections is a common method for obtaining geometric morphological information on various phases such as aggregates and pores in a cross-section, and it is widely used in microstructure analysis and pavement performance evaluation. During acquisition, optical equipment such as digital cameras are typically used to photograph the cross-section, and then digital image processing algorithms are used to extract the geometric features of each phase.

[0003] However, the above methods have the following problems in engineering applications: First, the image acquisition process lacks effective isolation from ambient light, and uneven distribution of natural light leads to local overexposure or underexposure of the image. The color consistency of different batches of acquisition results is poor, resulting in large areas of white pixel clusters or loss of fine aggregate information in the aggregate phase segmentation results. Second, the aggregate and asphalt mortar have similar colors in the cross-sectional image, and the aggregate exhibits multiple colors in the same cross-section due to differences in internal mineral composition. The pixel feature values ​​of the three phases (aggregate phase, asphalt mortar phase, and porous phase) overlap significantly, making it difficult to accurately distinguish them using a unified algorithm. Third, the pore space is filled with air, and the geometric information of the pore area is replaced by the image of other components at the bottom of the pore during optical imaging, resulting in the loss of pore geometric information. Fourth, existing image segmentation methods require specially designed morphological processing algorithms for images from different sources, and it is difficult to completely avoid manual intervention such as outlining the aggregate contour and filling in the pore position. The processing accuracy of cross-sectional images from different sources is difficult to meet engineering requirements. Summary of the Invention

[0004] The purpose of this invention is to provide a method for optical image acquisition and segmentation of asphalt concrete based on physical color enhancement, in order to solve the following problems existing in the prior art: the lack of effective isolation of ambient light during the image acquisition process, uneven distribution of natural light leading to local overexposure or underexposure of the image, and poor color consistency of acquisition results from different batches; the similarity in color between aggregates and asphalt mortar, and the multiple colors of aggregates due to differences in mineral composition, resulting in a large overlap of pixel feature values ​​of the three phases, making accurate segmentation difficult; the pore space being filled with transparent air medium, causing the pore geometric information to be replaced by images of other components at the bottom of the pores during optical imaging, resulting in the loss of pore geometric information; and existing image segmentation methods require specially designed algorithms for images from different sources, making it difficult to avoid human intervention and resulting in insufficient generalization ability.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A method for optical image acquisition and segmentation of asphalt concrete based on physical color enhancement includes the following steps: (1) Section preparation: Physically cut the asphalt concrete specimen to obtain a physical section; (2) Light shield preparation: Prepare a physical light shield, which is equipped with a white light strip and an acquisition window; (3) Aggregate phase contrast enhancement: Use a cleaning agent and a soft brush to deeply clean the physical section, and place the section in a ventilated place to dry after cleaning; (4) Aggregate phase image acquisition: Place the physical section processed in step (3) in the physical light shield, turn on the white light strip, and use an optical acquisition device to acquire the first optical image of the section through the acquisition window to obtain the image for the aggregate phase. (5) Filling and color enhancement of pore phase: After uniformly spreading white powder colorant on the surface of the cross section obtained in step (3), tilt the cross section and use a brush to lightly brush away excess powder on the surface along the inclined direction. The natural concave structure of the pores is used to achieve selective filling of the powder; (6) Pore phase image acquisition: Place the physical cross section processed in step (5) inside the physical light shield, and perform a second optical image acquisition under the same camera position, angle and light shield setting conditions as in step (4) to obtain a color image for extracting the geometric features of the pore phase; (7) Image segmentation: Grayscale the color image obtained in step (4) using the channel maximum value principle, and then using a threshold d Agg Binarization segmentation is performed on ∈ (0, 255) to obtain the aggregate phase extraction result; the color image obtained in step (6) is grayscaled using the channel mean principle, and then the threshold d is used. Void Binarization segmentation was performed on ∈ (0, 255) to obtain the pore phase extraction result; (8) Image synthesis: Bilinear interpolation algorithm was used to unify the pixel specifications of the aggregate phase extraction result and the pore phase extraction result; the pore phase pixel value in the pore phase extraction result was corrected to 0 and the non-pore phase pixel value was corrected to 62, the aggregate phase pixel value in the aggregate phase extraction result was corrected to 255 and the non-aggregate phase pixel value was corrected to 62; the two corrected images were subjected to pixel superposition Boolean operation, and the pixels with a pixel value greater than or equal to 255 were classified as aggregate phase, the pixels with a pixel value equal to 62 were classified as pore phase, and the pixels with a pixel value equal to 124 were classified as asphalt mortar phase, so as to obtain a complete cross-sectional synthesized image containing aggregate phase, pore phase and asphalt mortar phase.

[0006] Preferably, the cleaning agent in step (3) is transparent, colorless water, and the temperature during the drying process does not exceed 45 degrees Celsius.

[0007] Preferably, the white powder colorant in step (5) is CaSO4·2H2O, and the powder particle size is 1 micrometer to 100 micrometers; the white powder colorant may also be selected from one or more non-toxic and harmless white powder solids among CaCO3, Na2SO4, MgO, and Al(OH)3.

[0008] Preferably, the color image obtained by optical image acquisition in steps (4) and (6) is a 3-channel 8-Bit RGB color image with a pixel specification of 1500 pixels × 3000 pixels.

[0009] Preferably, the threshold d in step (7) Agg The value of d is 90. Void The value is 220.

[0010] Preferably, the method for unifying pixel specifications in step (8) is a bilinear interpolation algorithm. When the number of pixels in the unified image is greater than that before unification, a pixel expansion method is used, and when the number of pixels in the unified image is less than that before unification, a pixel reduction method is used.

[0011] Meanwhile, this invention discloses a physical light shield for optical image acquisition of asphalt concrete cross-sections, comprising: a light shield housing, enclosed by light-shielding material, used to isolate ambient light; white light strips, disposed around the inside of the light shield housing with their emitting surfaces facing the bottom of the housing, the emitting surfaces of the white light strips being parallel to the bottom surface of the housing; an acquisition window, located at the top of the housing, used to regulate the shooting position and angle of the optical acquisition equipment; a platform, disposed at the bottom of the housing, used to support the cross-section to be acquired, the platform having a center mark; and a portable power supply, electrically connected to the white light strips, used to provide independent power to the white light strips; wherein, all surfaces of the components inside the light shield housing are matte to prevent specular reflection of light from causing overexposure of the cross-section image.

[0012] Preferably, the external dimensions of the light-shielding box are 45cm×45cm×60cm, the dimensions of the collection window are 4cm×4cm, and the dimensions of the shelf are 45cm×45cm×1cm.

[0013] Preferably, there are four white light strips, which are evenly distributed on the same horizontal line around the inside of the light-shielding box. Each white light strip measures 44cm × 0.8cm × 0.8cm. The bottom surface of the white light strip is the light-emitting surface, while the top, front, back, left, and right sides are made of opaque material. The bottom of the white light strip is 40cm above the bottom inner side of the light-shielding box.

[0014] The beneficial effects of this invention are: (1) By preparing a special physical light shield, the light shield material is used to physically isolate the interference of external ambient light. Combined with four white light strips evenly distributed around the inside of the light shield box, stable and uniform supplementary light is provided, so that the uniformity of illumination of the asphalt concrete section is effectively guaranteed. The results of different batches of collection show a high degree of color consistency, eliminating the problem of local overexposure and underexposure caused by uneven natural illumination, and effectively avoiding the appearance of large white pixel groups and the loss of fine aggregate information in the aggregate phase segmentation results. (2) By taking advantage of the difference in the physical and chemical properties of water molecules between aggregates (hydrophilic) and asphalt mortar (hydrophobic), a transparent aqueous solution is used in conjunction with a soft brush to perform deep cleaning of the cross-section and then air-dry it (temperature not exceeding 45 degrees Celsius). This significantly enhances the color contrast between the aggregate phase and the asphalt mortar phase, effectively solving the imaging interference problems of "different materials presenting similar colors" and "the same materials presenting different colors". The problem of aggregate phase segmentation is simplified to a binary classification problem, and the aggregate phase can be extracted quickly and accurately directly through threshold segmentation. (3) Based on the idea of ​​material substitution, white powder CaSO4·2H2O (particle size from 1 micrometer to 100 micrometers) is used as a colorant. The physical structure of the natural depression of the pores is used to achieve selective filling of the powder, giving the pore phase a distinct white feature. This fundamentally solves the core problem of the loss of pore geometric information caused by the pore space being filled by transparent air medium. At the same time, it avoids the powder adhering to the surface of the aggregate phase and the asphalt mortar phase, and simplifies the problem of pore phase segmentation into a binary classification problem. (4) By decomposing the three-class classification problem into two independent binary classification problems, the channel maximum value principle is used for grayscale conversion in combination with the threshold d. Agg Extracting aggregate phases and channel mean values, grayscale matching threshold d Void The porous phase is extracted, and then pixel value correction (the pixel value of the porous phase is corrected to 0, the pixel value of the aggregate phase is corrected to 255, and the initial value of the indeterminate phase is set to 62) and Boolean operation is performed to synthesize a complete three-phase cross-sectional information image. This significantly reduces the complexity of the image segmentation algorithm and effectively avoids the interference of subjective factors caused by human intervention. After verification on multiple asphalt concrete cross-sections with different gradations, aggregate colors and materials, this method has good versatility for cross-sections from different sources and can be quickly deployed to engineering sites to meet engineering accuracy requirements. Attached Figure Description

[0015] Figure 1 A flowchart illustrating the physical enhancement-based optical image acquisition and segmentation method for asphalt concrete.

[0016] Figure 2 This is a schematic diagram of a physical light shield.

[0017] Figure 3 This is a schematic diagram of the structure of a white light strip.

[0018] Figure 4 This is a schematic diagram illustrating the working state of cross-sectional optical image acquisition using a physical light shield.

[0019] Figure 5 This is a schematic diagram showing the setup for different combinations of light sources to be turned on within a physical light shield.

[0020] Figure 6 This is a schematic diagram comparing optical images of the cross-section under different lighting conditions after aggregate phase contrast enhancement treatment.

[0021] Figure 7 This is a schematic diagram comparing optical images of a cross-section under different lighting conditions after being treated with porous phase filling and color enhancement.

[0022] Figure 8 This is a schematic diagram comparing the extraction results of aggregate phase and pore phase from cross-sectional optical images under different lighting conditions.

[0023] Figure 9 The flowchart of the image synthesis method is as follows: (a) original cross-section image (grayscale result); (b) pore phase extraction result (stage I); (c) aggregate phase extraction result (stage I); (d) pore phase extraction result (stage II); (e) aggregate phase extraction result (stage II); (f) pore phase extraction result (stage III); (g) aggregate phase extraction result (stage III); (h) fusion of pore-aggregate phase results.

[0024] Figure 10 A schematic diagram illustrating the principle of the method for unifying image pixel specifications: (a) before pixel specification unification; (b) after pixel specification unification.

[0025] Figure 11 A comparative diagram of the composite images of the pore phase extraction results and the aggregate phase extraction results at different processing stages: (a) Pixel specification unification not completed; (b) Pixel specification unification completed.

[0026] Figure 12 A schematic diagram of the method for correcting pixel information of each component of an image: (a) pore phase; (b) aggregate phase.

[0027] Figure 13 A schematic diagram illustrating the entire process of optical image acquisition and segmentation of asphalt concrete based on physical color enhancement.

[0028] Figure 14 This is a schematic diagram showing the optical image acquisition effect for different asphalt concrete cross sections.

[0029] Figure 15 This is a schematic diagram of the priority determination logic for each material phase during image synthesis. Detailed Implementation

[0030] The specific embodiments of the present invention will be described in detail with reference to the accompanying drawings.

[0031] The present invention will be further described in detail below with reference to the embodiments. Unless otherwise specified, all raw materials used in the following embodiments are commercially available products. The white powder colorant CaSO4·2H2O (particle size 1 micrometer to 100 micrometers) was purchased from Sinopharm Chemical Reagent Co., Ltd., with a purity of not less than 99%. The optical image acquisition device uses a digital camera with manual focus function, and the image acquisition result is a 3-channel 8-bit RGB color image. The image processing algorithm is implemented based on the Python programming language, and the image processing library used is OpenCV.

[0032] Example 1 This embodiment provides a complete implementation process for an optical image acquisition and segmentation method for asphalt concrete based on physical color enhancement, including steps such as physical mask preparation, aggregate phase contrast enhancement, aggregate phase image acquisition, pore phase filling color enhancement, pore phase image acquisition, image segmentation, and image synthesis.

[0033] Step 1: Section Preparation. Following the requirements of the "Test Procedures for Asphalt and Asphalt Concrete in Highway Engineering" and the "Technical Specifications for Construction of Asphalt Pavement on Highways," determine parameters such as aggregate gradation, void ratio, and asphalt-aggregate ratio. Prepare hexahedral asphalt concrete specimens measuring 150mm × 300mm × 50mm using a roller mill. Use a concrete cutter to horizontally cut the specimen from its waist, obtaining a smooth physical section of asphalt concrete measuring 150mm × 300mm. Check the smoothness of the section; the surface should be free of pits (except for original pores) caused by knife marks and without obvious wavy undulations.

[0034] Step 2: Preparation of the physical light shield. (Reference) Figure 2 A physical light shield was prepared, with external dimensions of 45cm × 45cm × 60cm. The light shield enclosure was constructed of light-shielding material to physically isolate external ambient light interference. Four white light strips were evenly installed around the inside of the light shield enclosure, with the four white light strips positioned on the same horizontal line. Each white light strip measured 44cm × 0.8cm × 0.8cm.

[0035] refer to Figure 3 The bottom surface of the white light strip (a rectangular area of ​​44cm × 0.8cm) is the light-emitting surface, facing the bottom of the light-shielding box and parallel to the bottom surface of the box. The top, front, back, left, and right sides are all made of opaque material. The bottom of the white light strip is 40cm above the inner bottom of the light-shielding box. All surfaces of the components inside the light-shielding box are matte to prevent specular reflection of light from causing overexposure of the cross-sectional image.

[0036] A 4cm x 4cm acquisition window is provided on the top of the light-shielding enclosure to standardize the shooting position and angle of the optical acquisition equipment. A 45cm x 45cm x 1cm platform is provided at the bottom of the light-shielding enclosure, with a center mark to support the cross-section to be acquired and ensure that the center of the cross-section is aligned with the center of the acquisition window. The white light strip is powered by a portable power supply, achieving a stable power supply independent of an external power source.

[0037] Step 3: Enhance the contrast of aggregates. Utilizing the difference in the physicochemical properties of water molecules between aggregates (hydrophilic) and asphalt mortar (hydrophobic), transparent, colorless water is used as a cleaning agent. A soft brush is then used to thoroughly clean and scrub the area where the physical cross-section of the asphalt concrete was sampled, ensuring that the cross-section surface is free of dust and loose gravel.

[0038] After cleaning, lay the cross-section flat in a ventilated area to air dry naturally for approximately 40 minutes (at an ambient temperature of 32 degrees Celsius). The temperature should not exceed 45 degrees Celsius during the entire drying process to prevent the asphalt mortar from softening and flowing under high temperatures. The dried cross-section surface should be dry and free of water stains. This treatment significantly enhances the color contrast between the aggregate phase and the asphalt mortar phase, effectively resolving the imaging interference problems of "different materials displaying similar colors" and "the same material displaying different colors."

[0039] Step 4: Aggregate Phase Image Acquisition. (Reference) Figure 4 Place the physical light shield horizontally, and place the asphalt concrete section treated in step three on the platform, aligning the geometric center of the section with the center mark, and adjusting the inclination of the section to 0 degrees. Turn on all four white light strips inside the light shield, with the emitting surface of each light strip pointing directly downwards.

[0040] Insert the lens of the optical acquisition device into the acquisition window, with an insertion distance not exceeding 1 cm, and align the focus center with the center of the platform. Once focusing is complete and the entire cross-sectional image is clearly visible, complete the first optical image acquisition, obtaining a 3-channel 8-bit RGB color image for extracting the geometric features of the aggregate phase. The image pixel specifications are 1500 pixels × 3000 pixels. Record the camera position, angle, and light shield settings at this time for reproduction in step six.

[0041] Step 5: Filling and color enhancement of porous phases. Based on the clean and dry cross-section obtained in Step 3, pour 5g of white powdered CaSO4·2H2O (powder particle size 1 to 100 micrometers) onto the cross-section surface and spread the powder evenly on the cross-section surface using a glass rod.

[0042] Incline the cross-section to 15 degrees and gently brush away excess CaSO4·2H2O powder from top to bottom along the slope. This filling process avoids rigid friction with the asphalt concrete cross-section to prevent powder from adhering to the aggregate and asphalt mortar phase surfaces. Utilizing the natural depressions in the pores, the powder selectively fills the porous phase, giving it a distinct white characteristic and fundamentally solving the core problem of missing pore geometric information.

[0043] Step Six: Pore Phase Image Acquisition. (Reference) Figure 4 Place the physical light shield horizontally, and place the asphalt concrete section treated in step five on the platform, aligning the geometric center of the section with the center mark, and adjusting the inclination of the section to 0 degrees. Turn on all four white light strips inside the light shield, with the emitting surface of each light strip pointing directly downwards.

[0044] Insert the lens of the optical acquisition device into the acquisition window, with an insertion distance not exceeding 1 cm. Align the focus center with the center of the stage. The camera position, angle, and lens hood settings should be exactly the same as in step four. Once focusing is complete and the entire cross-sectional image is clearly visible, complete the second optical image acquisition to obtain a 3-channel 8-bit RGB color image for extracting the geometric features of the porous phase. The image pixel specifications are 1500 pixels × 3000 pixels.

[0045] Step 7: Image Segmentation. For the aggregate phase image obtained in Step 4, grayscale conversion is performed using the maximum channel value principle, that is, the maximum value among the R, G, and B channels is taken as the grayscale value for each pixel: η = Max(R, G, B). After grayscale conversion, a binary classification threshold d is defined. Agg ∈ (0, 255), when gray value η≤d Agg When the pixel is determined to be a non-aggregate phase (pixel value set to 0); when the grayscale value η > d Agg When this occurs, the pixel is identified as the aggregate phase (pixel value set to 1), and the aggregate phase extraction result is obtained. In this embodiment, d Agg The value is 90.

[0046] For the porous phase image obtained in step six, grayscale conversion is performed using the channel mean principle, that is, the arithmetic mean of the R, G, and B channels is taken as the grayscale value for each pixel: η = (R + G + B) / 3. After grayscale conversion, a binary classification threshold d is defined. Void ∈ (0, 255), when gray value η≤d Void When the pixel is in the non-porous phase (pixel value set to 0); when the grayscale value η > d Void When this occurs, the pixel is identified as a porous phase (pixel value set to 1), and the porous phase extraction result is obtained. In this embodiment, d Void The value is 220. Both of the above segmentation results are 1-bit binary images.

[0047] Step 8: Image Compositing. (Reference) Figure 9 To address the issue of inconsistent pixel specifications between aggregate phase extraction results and pore phase extraction results caused by system errors in optical acquisition equipment, a bilinear interpolation algorithm is first used to unify the pixel specifications of the two images, achieving two-dimensional spatial alignment.

[0048] refer to Figure 10 In the bilinear interpolation algorithm, the number of pixels in the image after pixel specification unification is k² times the number of pixels in the image before unification, where k is the ratio of the number of pixels on one side of the image before and after pixel specification unification, i.e., k = (q+1) / 2. The pixel value at each position in the new image after unification is calculated by the weighted average of the four surrounding known pixels: η enh (i+u, j+v) = (1-u / q)(1-v / q)η(i, j) + (1-u / q)(v / q)η(i, j+1) + (u / q)(1-v / q)η(i+1, j) + (u / q)(v / q)η(i+1, j+1), where i and j are the coordinates of known pixels, and u and v are the coordinate offsets of the new pixel relative to the four surrounding known pixels. When the number of pixels in the unified image is greater than before unification, a pixel expansion method is used; when the number of pixels in the unified image is less than before unification, a pixel reduction method is used (these are the inverse operations of the pixel expansion method).

[0049] refer to Figure 12 After unifying the pixel specifications, the component pixel information of the two images was corrected. For the image extracted from the porous phase, the pixel values ​​of the porous phase were uniformly corrected to 0, and the pixel values ​​of the non-porous phase were uniformly corrected to 62, resulting in the corrected porous phase image. For the image extracted from the aggregate phase, the pixel values ​​of the aggregate phase were uniformly corrected to 255, and the pixel values ​​of the non-aggregate phase were uniformly corrected to 62, resulting in the corrected aggregate phase image.

[0050] refer to Figure 15 The above-mentioned correction rule is designed based on the following: aggregates are essentially dense rock particles, and the aggregate phase does not contain pores or asphalt mortar; pores are the remaining spaces between aggregate particles after compaction that are not filled by asphalt mortar, and the pore phase is surrounded by the asphalt mortar phase. Therefore, the pixel discrimination priority during image synthesis is: aggregate > pore > asphalt mortar. 62 is approximately taken from 255 / 4 as the initial pixel value for the indeterminate phase, facilitating subsequent determination of the material phase based on pixel value.

[0051] Perform a Boolean operation to overlay pixels onto the two corrected images: θ 合成 (x, y) = θ 孔隙 (x, y) + θ 集料(x, y). After superposition, there are four cases: the pixel value of the porous phase 0 is superimposed with the pixel value of the aggregate phase 255 to get 255 (aggregate phase); the pixel value of the porous phase 0 is superimposed with the pixel value of the non-aggregate phase 62 to get 62 (porous phase); the pixel value of the non-porous phase 62 is superimposed with the pixel value of the aggregate phase 255 to get 317 (automatically truncated to 255 when it exceeds the 8-bit range, and classified as aggregate phase); the pixel value of the non-porous phase 62 is superimposed with the pixel value of the non-aggregate phase 62 to get 124 (asphalt mortar phase).

[0052] Ultimately, pixels with a value greater than or equal to 255 in the composite image were classified as aggregate phase, pixels with a value equal to 62 as porous phase, and pixels with a value equal to 124 as asphalt mortar phase, resulting in a complete cross-sectional composite image containing aggregate, porous, and asphalt mortar phases. (Reference) Figure 13 The actual processing results of the complete process in this embodiment show that the method accurately and efficiently purifies the geometric information of aggregate phase and pore phase in optical images of asphalt concrete cross sections.

[0053] Example 2 This embodiment is used to verify the versatility of the method of the present invention for asphalt concrete sections with different gradations, aggregate colors and materials.

[0054] refer to Figure 14 Based on the requirements of the "Test Procedures for Asphalt and Asphalt Concrete in Highway Engineering" and the "Technical Specifications for Construction of Asphalt Pavement in Highways", five new and independent dense-graded asphalt concrete rutting slab sections (referred to as specimens 1 to 5) were prepared. Each specimen differed significantly from the specimen used in Example 1 in terms of gradation, aggregate color, and aggregate material.

[0055] It should be noted that in the method proposed in this invention, the color of the aggregate phase itself is uncontrollable, while the color of the filling material in the porous phase is controllable. Therefore, the image processing of the aggregate phase, whose color is uncontrollable, is more random. Thus, to test the applicability of the proposed method to aggregate phase segmentation under extreme conditions, the aggregates used in the re-prepared rut slabs had a darker average color (closer to asphalt mortar) and a smaller average particle size. By increasing the compaction degree, the interlocking degree between aggregates was increased, and the spacing was smaller, thereby increasing the difficulty of processing the optical image of the aggregate phase. Unfortunately, because the asphalt mixture is too compact under high pressure, with a porosity of essentially zero, it is impossible to simultaneously test the processing results of the aggregate phase.

[0056] The five cross-sections were prepared sequentially according to steps one through eight of Example 1, including cross-section preparation, physical light shield preparation, aggregate phase contrast enhancement, aggregate phase image acquisition, and image segmentation processing. The physical light shield used a device with the same specifications as in Example 1 (external dimensions 45cm × 45cm × 60cm, four white light strips evenly distributed on the same horizontal line, the bottom of the light strips 40cm above the inner bottom of the light shield, and an acquisition window size of 4cm × 4cm).

[0057] The cleaning agent used is clear, colorless water. A soft-bristled brush is used to thoroughly clean each surface. The drying process is carried out at a temperature not exceeding 45 degrees Celsius. The image acquisition uses an image pixel specification of 1500 pixels × 3000 pixels.

[0058] Regarding image segmentation threshold setting, for cross-sections with different aggregate colors and materials, d Agg The value is set to 90, consistent with Example 1, so there is no need to adjust the threshold parameter separately for different source sections.

[0059] refer to Figure 14 The results show that for five asphalt concrete cross sections with different gradations, aggregate colors, and aggregate materials, the method of the present invention can stably and effectively achieve accurate purification (extraction) of the geometric information of aggregate phase and asphalt mortar phase in the cross section optical image. This proves that the present invention has good versatility for asphalt concrete cross sections from different sources, can be quickly deployed and applied to engineering sites, and meets engineering accuracy requirements.

[0060] Comparative Example 1 This comparative example is used to verify the necessity of a physical light shield for image acquisition quality. (Reference) Figure 6 and Figure 8 Without using a physical light shield, optical images of the asphalt concrete cross-section after aggregate phase contrast enhancement treatment were acquired in natural strong light and natural weak light environments, respectively. The remaining operations were the same as in Example 1.

[0061] The results show that under strong natural light conditions, uneven illumination distribution leads to localized overexposure in the images. After image segmentation, large areas of white pixels appear in the aggregate phase extraction results, indicating adhesion between individual aggregates. Under weak natural light conditions, images are underexposed, resulting in significant loss of fine aggregate information in the aggregate phase extraction results after image segmentation, and an abnormally increased proportion of asphalt mortar phase. Therefore, without using a physical light shield, the overexposure and underexposure problems caused by uneven natural illumination severely affect image segmentation accuracy and fail to meet engineering accuracy requirements.

[0062] Comparative Example 2 This comparative example is used to verify the necessity of pore phase filling and color enhancement processing for extracting pore phase geometric information. Without performing the pore phase filling and color enhancement processing in step five, a second optical image is directly acquired from the cross-section after aggregate phase contrast enhancement processing, and grayscale is performed using the channel mean principle, with a threshold d. Void =220 is used for binarization segmentation, and the remaining operations are the same as in Example 1.

[0063] The results showed that because the pore space was filled with a transparent air medium, the geometric information of the pore region was replaced by the images of other components at the bottom of the pores during optical imaging. This resulted in severe distortion of the pore phase extraction results, with a significant loss of pore geometric information, making it impossible to accurately reflect the true geometric morphology of the cross-sectional pores. A comparison between Example 1 and Comparative Example 2 demonstrates that using white powdered CaSO4·2H2O for pore phase filling and color enhancement is a key step in achieving accurate extraction of pore geometric information. This treatment imparts a distinct white characteristic to the pore phase, fundamentally solving the core problem of missing pore geometric information, with unexpectedly positive results.

[0064] Test Example 1 This test case comprehensively evaluates the image acquisition and segmentation effects of Example 1, Example 2 (samples 1 to 5) and Comparative Example 1 and Comparative Example 2. The results are shown in Table 1.

[0065] Table 1 Comparison of image acquisition and segmentation effects in various embodiments and comparative examples As can be seen from Table 1, the extraction results of aggregate phase and pore phase of the cross section in Example 1 are accurate and complete. The extraction results of aggregate phase of the five cross sections from different sources in Example 2 are all accurate and complete, proving that the method of the present invention has good versatility. Comparative Examples 1 and 2 respectively demonstrate the necessity of physical light shield and pore phase filling color enhancement processing for ensuring the quality of image acquisition and segmentation.

[0066] It is understood that the white powder colorant is not limited to CaSO4·2H2O, but can also be one or more non-toxic and harmless white powder solids selected from CaCO3, Na2SO4, MgO, and Al(OH)3. The powder particle size is limited to between 1 micrometer and 100 micrometers, and all of them can achieve the selective filling and color enhancement effect of the porous phase.

[0067] It is understood that the cleaning agent is not limited to pure water, but can also be other transparent and colorless aqueous solutions, as long as the water quality meets the requirement of being basically transparent and colorless. The temperature during the drying process must not exceed 45 degrees Celsius.

[0068] Obviously, the optical image acquisition results are not limited to 3-channel 8-bit RGB color images, but may also include single-channel 8-bit grayscale images; the image pixel specifications are not limited to 1500 pixels × 3000 pixels, but may also include 1k, 2k, 4k and other arbitrary size ratio types to meet different engineering accuracy requirements.

[0069] It is understood that the asphalt concrete specimens are not limited to hexahedrons of 150mm×300mm×50mm, but can be specimens of any size including planar cross-sections, including thin plates, cylinders, polyhedra and other arbitrary shapes, and can be obtained from laboratory on-site preparation, sampling of newly built road pavement and cutting of existing road structures, etc.

Claims

1. A method for acquiring and segmenting optical images of asphalt concrete based on physical color enhancement, characterized in that, Includes the following steps: (1) Cross-section preparation: Physically cut the asphalt concrete specimen to obtain a physical cross-section; (2) Light shield preparation: Prepare a physical light shield, which is equipped with a white light strip and a collection window; (3) Aggregate phase contrast enhancement: Use a cleaning agent and a soft brush to deeply clean the physical cross-section, and place the cross-section in a ventilated place to dry after cleaning; (4) Aggregate phase image acquisition: Place the physical cross-section processed in step (3) in the physical light shield, turn on the white light strip, and use an optical acquisition device to perform the first optical image acquisition of the cross-section through the collection window to obtain a color image for the extraction of aggregate phase geometric features; (5) Pore phase filling and color enhancement: Use white powder for coloring. After the agent is evenly spread on the cross-section surface obtained in step (3), the cross-section is tilted and the excess powder on the surface is removed by brushing along the inclined direction. The powder is selectively filled by utilizing the natural concave structure of the pores; (6) Porous phase image acquisition: The physical cross-section processed in step (5) is placed in the physical light shield. Under the same camera position, angle and light shield setting conditions as in step (4), a second optical image acquisition is performed to obtain a color image for extracting the geometric features of the pore phase; (7) Image segmentation: The color image obtained in step (4) is grayscaled using the channel maximum value principle, and then binarized and segmented using the threshold dAgg∈(0,255) to obtain the aggregate phase extraction result; The color image obtained in step (6) is converted to grayscale using the channel mean principle, and then binarized and segmented using the threshold dVoid∈(0,255) to obtain the pore phase extraction result; (8) Image synthesis: The pixel specifications of the aggregate phase extraction results and the pore phase extraction results are unified by bilinear interpolation algorithm; the pore phase pixel value in the pore phase extraction results is corrected to 0 and the non-pore phase pixel value is corrected to 62; the aggregate phase pixel value in the aggregate phase extraction results is corrected to 255 and the non-aggregate phase pixel value is corrected to 62. A Boolean operation is performed on the two corrected images to overlay pixels. After overlay, pixels with a value greater than or equal to 255 are classified as aggregate phase, pixels with a value equal to 62 are classified as porous phase, and pixels with a value equal to 124 are classified as asphalt mortar phase, resulting in a complete cross-sectional composite image containing aggregate phase, porous phase, and asphalt mortar phase.

2. The method for acquiring and segmenting optical images of asphalt concrete based on physical color enhancement according to claim 1, characterized in that, The cleaning agent mentioned in step (3) is clear, colorless water, and the temperature during the drying process does not exceed 45 degrees Celsius.

3. The method for acquiring and segmenting optical images of asphalt concrete based on physical color enhancement according to claim 1, characterized in that, The white powder colorant mentioned in step (5) is CaSO4·2H2O, and the powder particle size of the white powder colorant is 1 micrometer to 100 micrometers.

4. The method for acquiring and segmenting optical images of asphalt concrete based on physical color enhancement according to claim 1, characterized in that, The white powder colorant mentioned in step (5) is selected from one or more of CaCO3, Na2SO4, MgO, and Al(OH)3, and the powder particle size of the white powder colorant is from 1 micrometer to 100 micrometers.

5. The method for acquiring and segmenting optical images of asphalt concrete based on physical color enhancement according to claim 1, characterized in that, The color images obtained by optical image acquisition in steps (4) and (6) are 3-channel 8-bit RGB color images with a pixel specification of 1500 pixels × 3000 pixels.

6. The method for acquiring and segmenting optical images of asphalt concrete based on physical color enhancement according to claim 1, characterized in that, The external dimensions of the physical light shield in step (2) are 45cm×45cm×60cm, and the dimensions of the acquisition window are 4cm×4cm.

7. The method for acquiring and segmenting optical images of asphalt concrete based on physical color enhancement according to claim 1, characterized in that, The physical light shield in step (2) has four white light strips inside. The four white light strips are evenly distributed on the same horizontal line. The size of each white light strip is 44cm×0.8cm×0.8cm. The bottom surface of the light strip is the light-emitting surface, and the other surfaces do not emit light and are not transparent. The height of the bottom of the white light strip from the bottom inner side of the light shield is 40cm.

8. The method for acquiring and segmenting optical images of asphalt concrete based on physical color enhancement according to claim 1, characterized in that, In step (7), the threshold dAgg is 90 and the threshold dVoid is 220.

9. The method for acquiring and segmenting optical images of asphalt concrete based on physical color enhancement according to claim 1, characterized in that, The method for unifying pixel specifications in step (8) is a bilinear interpolation algorithm.

Images