Image sensor-based uhpc appearance quality detection system and method

Abstract

The application discloses a UHPC appearance quality detection system and method based on an image sensor, relates to the technical field of UHPC appearance quality detection, and acquires a basic image by suppressing reflection interference through a polarizing filter, extracts three-dimensional space features of recesses and holes by using structure light stripe distortion detection, realizes illumination balance in combination with adaptive light supplement, and enhances the contrast of cracks and holes through atomized water film, so as to construct a high-quality image data set; based on the image set, a crack index is calculated, a surface pore area proportion is analyzed to calculate a surface pore index, a color space is used to obtain a color difference index of defects and a reference area, threshold determination and strategy generation are performed on three types of defects; when any two types of indexes exceed the threshold, a defect linkage evaluation mechanism is started; when the three types of indexes exceed the threshold at the same time, a composite defect risk determination mechanism is triggered, high-risk component classification, factory freezing, process correction and strict re-inspection of the same batch are realized.

Description

Technical Field

[0001] This invention relates to the field of UHPC appearance quality inspection technology, specifically to a UHPC appearance quality inspection system and method based on an image sensor. Background Technology

[0002] Ultra-high performance concrete (UHPC), as a new generation of high-performance building materials, is widely used in engineering fields such as bridges, tunnels, and prefabricated components due to its high strength, high durability, and excellent impermeability. However, UHPC may still develop appearance defects such as surface cracks, surface porosity, pitting, and color differences during production, pouring, and curing. These defects not only affect the appearance quality of the components but may also become potential structural hazards. Therefore, how to efficiently and accurately conduct surface quality inspection of UHPC components is a key issue in ensuring their service safety and lifespan.

[0003] Existing inspection methods mainly include manual visual inspection, contact non-destructive testing, and image acquisition and analysis methods based on ordinary industrial cameras. Among them, manual inspection is inefficient and highly subjective; contact methods are cumbersome to operate and have insufficient coverage; while traditional image processing-based inspection methods are often limited by factors such as lighting and the complexity of surface defect types, making it difficult to achieve high-precision and high-stability defect identification.

[0004] At the image acquisition level, existing technologies face the following challenges: UHPC surfaces may have flat and smooth areas, and strong reflections or uneven lighting can easily cause image overexposure and shadows, thus interfering with defect identification; UHPC surfaces may be dense and smooth, or they may have local capillaries or rough textures, resulting in unstable extracted image features.

[0005] In terms of defect identification, existing technologies also have significant shortcomings: UHPC crack widths are often in the range of 0.05~0.1mm, making it difficult for existing industrial cameras or traditional algorithms to detect and identify them stably; defects such as cracks, surface pores, pitting, and color differences may overlap in grayscale and morphological features, and single threshold or edge detection methods are prone to confusion; the edges of surface pores and pitting often show a gradual transition with normal rough surfaces, leading to inaccurate segmentation and thus affecting quantitative defect analysis. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a UHPC appearance quality inspection system and method based on image sensors to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a UHPC appearance quality inspection method based on an image sensor, comprising the following steps: Step 1: Improve the surface image quality of UHPC components through multi-source image acquisition and enhancement; obtain basic images by suppressing reflection interference using polarizing filters; extract spatial features of depressions and holes through structured light stripe distortion detection; achieve balanced illumination enhancement by combining adaptive supplementary lighting; enhance the contrast of cracks and holes by using atomized water film; integrate multiple images to construct a high-quality image dataset. Step 2: Based on the base image after reflection suppression and the enhanced image with equalization in the high-quality image dataset, extract the gray-level uniformity parameter Gref of the reference area on the surface of the component from the base image, extract the abnormal gray-level distribution parameter Gabn of the crack area from the enhanced image, and calculate the crack index Cr by combining the crack width uw and depth ud with the crack threshold Tcr. This will help determine whether the UHPC component has a risk of crack defects. If so, appropriate strategies will be given. Step 3: Based on the spatial feature data of depressions and holes in the high-quality image dataset, combined with the image after wetting enhancement, extract the surface pore detection area Apore and the total detection area Atot, calculate the surface pore index Ph, and compare it with the surface pore area ratio threshold TPh to determine whether the surface pore distribution of the UHPC component is within the allowable range. If not, give the corresponding strategy. Step 4: Based on the illumination equalization enhanced image and the wet enhancement image in the high-quality image dataset, the lightness component L2*, red-green axis color component a2*, and yellow-blue axis color component b2* of the reference area are extracted and combined with the lightness component L1*, red-green color component a1*, and yellow-blue color component b1* of the defect area to calculate the color difference index Dc. The index Dc is then compared with the color difference judgment threshold TDc to determine whether there is a color difference defect on the surface of the UHPC component. If so, an appropriate strategy is given. Step 5: When any two types of defect indices exceed the threshold simultaneously, the defect linkage assessment mechanism is activated, and linkage strategies for structural review, process tracing, or local repair and re-inspection are generated based on the spatial overlap of defects. When all three types of indices exceed the threshold, the composite defect risk judgment mechanism is triggered, automatically determining that the component has serious composite defects, and triggering high-risk classification, factory freezing, process correction, and stricter re-inspection of the same batch.

[0008] Preferably, step one includes: S11. Image sensors are deployed on the surface of the UHPC component, and polarization filters are installed to perform surface imaging acquisition on the component, monitor and suppress reflection interference in the bright areas, and obtain the basic image after reflection suppression. S12. Project structured light stripes onto the surface of the component, collect stripe deformation data through an image sensor, and use distortion analysis to detect tiny pits and pores on the surface to obtain spatial feature data of the pits and pores. S13. During the acquisition process, in conjunction with the adaptive lighting system, the brightness and angle of the light source are automatically adjusted based on the image histogram entropy value feedback to enhance the overall brightness uniformity and obtain a balanced and enhanced image. S14. Atomize the surface of the component to form a thin water film. Improve the contrast between cracks and holes through the optical effect of the water film and obtain the image after wetting enhancement. S15. Integrate the base image after reflection suppression, the spatial feature data of depressions and holes, the image after illumination equalization enhancement, and the image after wetting enhancement to establish a high-quality image dataset.

[0009] Preferably, step two includes: S21. Based on the base image after reflection suppression and the illumination equalization enhancement image in the high-quality image dataset, extract the gray-level uniformity parameter Gref of the reference area on the surface of the component from the base image, extract the abnormal gray-level distribution parameter Gabn of the crack area from the enhancement image, and combine the crack width uw and depth ud.

[0010] Preferably, step two also includes: S22. Extract geometric feature parameters of the crack region based on the enhanced image, including crack width uw, estimated crack depth ud, and gray-level contrast anomaly Gabn. Combined with the gray-level uniformity parameter Gref of the reference region, after dimensionless processing, calculate and obtain the crack index Cr.

[0011] S23. By setting a crack threshold Tcr and comparing the crack index Cr with the crack threshold Tcr, the first evaluation results are obtained, including: When the crack index Cr < crack threshold Tcr, it means that the width and depth of the cracks on the surface of the UHPC component are within the allowable range, there is no risk of crack defects, the current detection status is maintained and monitoring continues. This indicates that the width or depth of cracks on the surface of the UHPC component exceeds the allowable range, posing a risk of crack defects. This triggers the first warning instruction and generates the first strategy: pixel-level boundary marking of the abnormal crack area is highlighted in the detection interface, and an audio-visual prompt device is activated to alert quality inspectors; a dynamic tracking subprocess is initiated to continuously collect and update the crack extension length and direction. If the crack propagation rate further increases, a crack risk level upgrade mechanism is triggered, generating an emergency response strategy of "stopping use and scrapping for review".

[0012] Preferably, step three includes: S31. Based on the spatial feature data of depressions and holes in the high-quality image dataset, combined with the wet-enhanced image; the spatial feature data is used to perform three-dimensional structural analysis of the hole morphology to determine the geometric distribution and boundary range of the hole; at the same time, the contrast between the hole edge and the background is enhanced by the wet-enhanced image, the surface pore detection area Apore is extracted, and the total detection area Atot is obtained from the overall image.

[0013] Preferably, step three also includes: S32. After dimensionless processing of the extracted surface pore detection area Apore and the total detection area Atot, the surface porosity index Ph is calculated.

[0014] S33. By setting a preset surface pore area ratio threshold TPh, and comparing and analyzing the surface porosity index Ph with the surface pore area ratio threshold TPh, the second evaluation result is obtained, including: When the surface porosity index Ph < the surface porosity area ratio threshold TPh, it indicates that the surface porosity distribution of the UHPC component is within the allowable range, the apparent compactness of the component meets the requirements, and continuous monitoring is required. When the surface porosity index Ph is greater than or equal to the surface porosity area ratio threshold TPh, it indicates that the surface porosity distribution of the UHPC component is outside the allowable range, and there are pitting or hole defects. It is determined that there is a risk of abnormal surface porosity, triggering a second warning instruction and generating a second strategy: clustering and marking the concentrated distribution area of ​​surface porosity in the detection interface and generating a surface porosity density heat map; at the same time, linking the automated quality inspection recording equipment to include the current area in the "requiring repair and review" list; if the surface porosity distribution spatially overlaps with the crack area, the defect linkage assessment mechanism is triggered, and step five is entered.

[0015] Preferably, step four includes: S41. Based on the illumination equalization enhanced image and the wet enhancement image in the high-quality image dataset; use CIELab color space conversion to extract color components from the defect area and the reference area respectively; extract the lightness component L2*, red-green axis color component a2* and yellow-blue axis color component b2* of the reference area in the illumination equalization enhanced image, and extract the lightness component L1*, red-green color component a1* and yellow-blue color component b1* of the defect area in the wet enhancement image.

[0016] Preferably, step four also includes: S42. By combining the extracted lightness component L2*, red-green axis color component a2*, and yellow-blue axis color component b2* of the reference area with the lightness component L1*, red-green color component a1*, and yellow-blue color component b1* of the defect area, and after dimensionless processing, the color difference index Dc is calculated.

[0017] S43. By using a preset color difference judgment threshold TDc, and comparing and analyzing the color difference index Dc with the color difference judgment threshold TDc, the third evaluation result is obtained, including: When the color difference index Dc < the color difference judgment threshold TDc, it means that there are no color difference defects on the surface of the UHPC component, and continuous monitoring is required. When the color difference index Dc ≥ the color difference judgment threshold TDc, it indicates that there is a color difference defect on the surface of the UHPC component, triggering the third early warning instruction and generating the third strategy: generate a pseudo-color enhanced display for the color difference area in the detection interface, and simultaneously generate a process abnormality prompt, prompting quality inspectors to check the curing process, casting process, UHPC workability control and raw material batch consistency; if the color difference area overlaps with the surface pore or crack area, the composite defect risk judgment mechanism is triggered, and proceed to step five.

[0018] Preferably, step five includes: S51. When any two or more of the crack index Cr, surface porosity index Ph, and color difference index Dc simultaneously exceed their respective thresholds, the defect linkage assessment mechanism is automatically activated, and the linkage assessment results include: If the crack area and the surface pore area overlap in space, it indicates that the defect poses a structural risk. A linkage strategy is generated: automatically create a three-dimensional spatial defect distribution map, give priority to the display of overlapping areas, and conduct structural performance verification and detection. If the crack area overlaps with the color difference area, it indicates that maintenance or material process issues may lead to crack expansion. A linkage strategy is generated: the process monitoring database is linked to retrieve maintenance curves and raw material batch information, and material batch traceability and maintenance process review are performed. If the surface pore area overlaps with the color difference area, it indicates that there is a coupling defect between surface density and raw material uniformity. A linkage strategy is generated: the current area is marked as a "composite appearance anomaly area" and a local compensation repair and secondary imaging re-inspection process is triggered. S52. When the crack index Cr, surface porosity index Ph, and color difference index Dc all exceed their respective thresholds, the system automatically triggers the composite defect risk assessment mechanism and obtains the risk assessment results, including: The UHPC component surface was found to have multiple superimposed defects, posing serious risks to both appearance and structure. A composite defect handling strategy was generated: the composite defect area was automatically classified as a "high-risk component" and simultaneously uploaded to the quality traceability database; an "stop use and scrap review" instruction was automatically generated, and the quality inspection scheduling system was linked to freeze the current UHPC component's factory release authority; the "process correction and feedback" sub-process was initiated, and the production process parameters were retrospectively reviewed based on historical inspection data to generate correction suggestions; and a re-inspection instruction was triggered, so that components in the same batch underwent stricter testing to prevent the spread of batch defects.

[0019] Preferably, the UHPC appearance quality inspection system based on an image sensor includes: The multi-source image acquisition and enhancement module is used to improve the surface image quality of UHPC components through multi-source image acquisition and enhancement methods; it uses polarization filters to suppress reflection interference and acquire basic images; it extracts the spatial features of depressions and holes through structured light stripe distortion detection; it achieves balanced illumination enhancement by combining adaptive supplementary lighting; it uses atomized water film to improve the contrast of cracks and holes; and it integrates multiple images to construct a high-quality image data set. The crack monitoring module is used to extract the gray-level uniformity parameter Gref of the reference area on the surface of the component from the base image after reflection suppression and the enhanced image with equal illumination from the high-quality image dataset, and extract the abnormal gray-level distribution parameter Gabn of the crack area from the enhanced image. Combined with the crack width uw and depth ud, the crack index Cr is calculated and compared with the crack threshold Tcr to determine whether there is a risk of crack defects in the UHPC component. If so, an appropriate strategy is given. The surface porosity monitoring module is used to extract the surface porosity detection area Apore and the total detection area Atot based on the spatial feature data of depressions and pores in the high-quality image dataset and the image after wetting enhancement. It calculates the surface porosity index Ph and compares it with the surface porosity area ratio threshold TPh to determine whether the surface porosity distribution of the UHPC component is within the allowable range. If not, it provides an appropriate strategy. The color difference monitoring module is used to extract the lightness component L2*, red-green axis color component a2*, and yellow-blue axis color component b2* of the reference area from the high-quality image dataset, and combine them with the lightness component L1*, red-green color component a1*, and yellow-blue color component b1* of the defect area to calculate the color difference index Dc. The index Dc is then compared with the color difference judgment threshold TDc to determine whether there is a color difference defect on the surface of the UHPC component. If a defect exists, an appropriate strategy is applied. The defect linkage and risk assessment module is used to activate the defect linkage assessment mechanism when any two types of defect indices exceed the threshold at the same time. Based on the spatial overlap of defects, it generates linkage strategies for structural review, process tracing, or local repair and re-inspection. When all three types of indices exceed the threshold, the composite defect risk assessment mechanism is triggered, which automatically determines that the component has serious composite defects and triggers high-risk classification, factory freeze, process correction, and stricter re-inspection of the same batch.

[0020] This invention provides a UHPC appearance quality inspection system and method based on an image sensor. It has the following beneficial effects: (1) The UHPC appearance quality inspection system and method based on image sensor effectively solves the problem of unstable imaging quality caused by smooth reflection and local blind area of ​​UHPC surface by deploying image sensor on the surface of UHPC component and combining it with polarization filter to suppress strong reflection, adaptive supplementary lighting to achieve illumination balance, using structured light stripe distortion analysis to detect micro holes, and atomized water film to enhance crack contrast, thus ensuring the clarity and stability of the acquired image.

[0021] (2) The UHPC appearance quality inspection system and method based on image sensor, based on multi-source high-quality images acquired by image sensor, constructs crack index Cr, surface porosity index Ph and color difference index Dc respectively, and performs quantitative calculation through multi-dimensional parameters such as crack, spatial hole area ratio and color component difference, avoiding the misjudgment and missed judgment problems that are easy to occur in traditional single threshold or edge detection methods, and improving the accuracy and robustness of defect identification.

[0022] (3) The UHPC appearance quality inspection system and method based on image sensor can automatically trigger the linkage evaluation mechanism according to the spatial overlap relationship of defects when two or more types of defect indices exceed the threshold at the same time, distinguish structural risks, material and process problems and appearance anomalies, generate targeted linkage disposal strategies, and realize the upgrade from single-point defect detection to defect coupling risk judgment.

[0023] (4) The UHPC appearance quality inspection system and method based on image sensor triggers a composite defect risk judgment mechanism when all three types of defect indices exceed the threshold. The system automatically classifies the components as high-risk components and links the quality traceability database, quality inspection scheduling system and process correction sub-process to perform factory freezing, batch traceability, parameter backtracking and stricter re-inspection, so as to realize closed-loop control of detection-judgment-disposal-traceability and ensure the quality and safety of UHPC components throughout their entire life cycle. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the steps of the UHPC appearance quality inspection method based on an image sensor according to the present invention; Figure 2 This is a block diagram and flowchart of the UHPC appearance quality inspection system based on an image sensor according to the present invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] Example 1 Please see Figure 1 This invention provides a method for inspecting the appearance quality of a UHPC based on an image sensor, comprising the following steps: Step 1: Improve the surface image quality of UHPC components through multi-source image acquisition and enhancement; obtain basic images by suppressing reflection interference using polarizing filters; extract spatial features of depressions and holes through structured light stripe distortion detection; achieve balanced illumination enhancement by combining adaptive supplementary lighting; enhance the contrast of cracks and holes by using atomized water film; integrate multiple images to construct a high-quality image dataset. Step 2: Based on the base image after reflection suppression and the enhanced image with equalization in the high-quality image dataset, extract the gray-level uniformity parameter Gref of the reference area on the surface of the component from the base image, extract the abnormal gray-level distribution parameter Gabn of the crack area from the enhanced image, and calculate the crack index Cr by combining the crack width uw and depth ud with the crack threshold Tcr. This will help determine whether the UHPC component has a risk of crack defects. If so, appropriate strategies will be given. Step 3: Based on the spatial feature data of depressions and holes in the high-quality image dataset, combined with the image after wetting enhancement, extract the surface pore detection area Apore and the total detection area Atot, calculate the surface pore index Ph, and compare it with the surface pore area ratio threshold TPh to determine whether the surface pore distribution of the UHPC component is within the allowable range. If not, give the corresponding strategy. Step 4: Based on the illumination equalization enhanced image and the wet enhancement image in the high-quality image dataset, the lightness component L2*, red-green axis color component a2*, and yellow-blue axis color component b2* of the reference area are extracted and combined with the lightness component L1*, red-green color component a1*, and yellow-blue color component b1* of the defect area to calculate the color difference index Dc. The index Dc is then compared with the color difference judgment threshold TDc to determine whether there is a color difference defect on the surface of the UHPC component. If so, an appropriate strategy is given. Step 5: When any two types of defect indices exceed the threshold simultaneously, the defect linkage assessment mechanism is activated, and linkage strategies for structural review, process tracing, or local repair and re-inspection are generated based on the spatial overlap of defects. When all three types of indices exceed the threshold, the composite defect risk judgment mechanism is triggered, automatically determining that the component has serious composite defects, and triggering high-risk classification, factory freezing, process correction, and stricter re-inspection of the same batch.

[0027] In this embodiment, by deploying image sensors on the surface of UHPC components and combining them with multi-source enhancement methods such as polarization filters, adaptive supplementary lighting, structured light, and atomized water film, the imaging quality of surface defects is significantly improved. Based on image features, crack index, surface porosity index, and color difference index are calculated, enabling multi-dimensional quantitative judgment. When multiple indices exceed the threshold, the system can automatically trigger defect linkage assessment and composite risk judgment based on the spatial distribution information obtained by the image sensors, thereby ensuring the accuracy of defect detection results and the closed-loop nature of subsequent handling.

[0028] Example 2 This embodiment is an explanation based on Embodiment 1. Please refer to it. Figure 1 Specifically, step one includes: S11. Image sensors are deployed on the surface of the UHPC component, and polarization filters are installed to perform surface imaging acquisition on the component, monitor and suppress reflection interference in the bright areas, and obtain the basic image after reflection suppression. S12. Project structured light stripes onto the surface of the component, collect stripe deformation data through an image sensor, and use distortion analysis to detect tiny pits and pores on the surface to obtain spatial feature data of the pits and pores. S13. During the acquisition process, in conjunction with the adaptive lighting system, the brightness and angle of the light source are automatically adjusted based on the image histogram entropy value feedback to enhance the overall brightness uniformity and obtain a balanced and enhanced image. S14. Atomize the surface of the component to form a thin water film. Improve the contrast between cracks and holes through the optical effect of the water film and obtain the image after wetting enhancement. S15. Integrate the base image after reflection suppression, the spatial feature data of depressions and holes, the image after illumination equalization enhancement, and the image after wetting enhancement to establish a high-quality image dataset.

[0029] In this embodiment, by deploying image sensors on the surface of UHPC components and combining them with multi-source enhancement methods such as polarization filters, adaptive supplementary lighting, structured light, and atomized water film, reflection interference can be effectively suppressed, brightness uniformity can be improved, and the contrast between cracks and surface pores can be enhanced during the acquisition process. This will establish a high-quality image data set and provide more accurate basic data support for subsequent quantitative analysis of defects.

[0030] Example 3 This embodiment is an explanation based on Embodiment 2. Please refer to it. Figure 1 Specifically, step two includes: S21. Based on the base image after reflection suppression and the illumination equalization enhancement image in the high-quality image dataset, extract the gray-level uniformity parameter Gref of the reference area on the surface of the component from the base image, extract the abnormal gray-level distribution parameter Gabn of the crack area from the enhancement image, and combine the crack width uw and depth ud.

[0031] In this embodiment, by extracting the gray-scale uniformity parameter Gref of the reference area on the component surface from the base image and extracting the abnormal gray-scale distribution parameter Gabn of the crack area from the enhanced image, and combining the crack width uw and depth ud for comprehensive modeling and analysis, it is possible to quantify the geometric size and gray-scale anomalies of the crack in multiple dimensions on the basis of eliminating reflection and illumination interference, thereby improving the accuracy and sensitivity of crack defect identification, avoiding misjudgment caused by relying on a single geometric feature, and thus more reliably reflecting the true crack risk level on the surface of the UHPC component.

[0032] Example 4 This embodiment is an explanation based on Embodiment 3. Please refer to it. Figure 1 Specifically, step two also includes: S22. Extract geometric feature parameters of the crack region based on the enhanced image, including crack width uw, crack depth estimate ud, and gray-level contrast anomaly Gabn. Combined with the gray-level uniformity parameter Gref of the reference region, after dimensionless processing, calculate the crack index Cr as follows:

[0033] In the formula, f represents the comprehensive quantization function of crack geometry and gray-scale anomaly; S23. By setting a crack threshold Tcr and comparing the crack index Cr with the crack threshold Tcr, the first evaluation results are obtained, including: When the crack index Cr < crack threshold Tcr, it means that the width and depth of the cracks on the surface of the UHPC component are within the allowable range, there is no risk of crack defects, the current detection status is maintained and monitoring continues. This indicates that the width or depth of cracks on the surface of the UHPC component exceeds the allowable range, posing a risk of crack defects. This triggers the first warning instruction and generates the first strategy: pixel-level boundary marking of the abnormal crack area is highlighted in the detection interface, and an audio-visual prompt device is activated to alert quality inspectors; a dynamic tracking subprocess is initiated to continuously collect and update the crack extension length and direction. If the crack propagation rate further increases, a crack risk level upgrade mechanism is triggered, generating an emergency response strategy of "stopping use and scrapping for review".

[0034] The method for obtaining the crack threshold Tcr is as follows: By statistically analyzing a large amount of crack detection data of UHPC components under different curing conditions and stress conditions, the normal and abnormal ranges of crack width and depth distribution are extracted. Combined with material mechanical properties, durability requirements and structural safety standards, a reasonable critical crack threshold is determined. With reference to relevant concrete crack control specifications, industry testing standards and experience data from quality inspection agencies, and combined with expert judgment, this threshold is formulated to accurately reflect the severity of crack defects and ensure the reliability of component appearance quality and structural performance.

[0035] In this embodiment, crack width, depth, and grayscale anomaly features are extracted based on enhanced images, and a crack index is constructed. This index is then compared with a preset threshold to achieve quantitative and dynamic assessment of surface crack defects in UHPC components. This not only enables early identification of crack risks but also automatically adjusts the risk level by tracking crack propagation trends in real time. This effectively improves the accuracy and timeliness of crack detection, ensuring the appearance quality and structural safety of the components.

[0036] Example 5 This embodiment is an explanation based on Embodiment 4. Please refer to it. Figure 1 Specifically, step three includes: S31. Based on the spatial feature data of depressions and holes in the high-quality image dataset, combined with the wet-enhanced image; the spatial feature data is used to perform three-dimensional structural analysis of the hole morphology to determine the geometric distribution and boundary range of the hole; at the same time, the contrast between the hole edge and the background is enhanced by the wet-enhanced image, the surface pore detection area Apore is extracted, and the total detection area Atot is obtained from the overall image.

[0037] In this embodiment, by combining the spatial feature data of depressions and holes with the wetted and enhanced image, the geometric distribution and boundary range of holes can be accurately analyzed at the three-dimensional structural level, and the contrast of hole edges can be enhanced at the two-dimensional image level. This enables the accurate extraction of the surface pore detection area Apore and the total detection area Atot, effectively improving the accuracy and reliability of surface pore defect identification of UHPC components.

[0038] Example 6 This embodiment is an explanation based on Embodiment 5. Please refer to it. Figure 1 Specifically, step three also includes: S32. After dimensionless processing of the extracted surface pore detection area Apore and the total detection area Atot, the surface porosity index Ph is calculated using the following formula:

[0039] S33. By setting a preset surface pore area ratio threshold TPh, and comparing and analyzing the surface porosity index Ph with the surface pore area ratio threshold TPh, the second evaluation result is obtained, including: When the surface porosity index Ph < the surface porosity area ratio threshold TPh, it indicates that the surface porosity distribution of the UHPC component is within the allowable range, the apparent compactness of the component meets the requirements, and continuous monitoring is required. When the surface porosity index Ph is greater than or equal to the surface porosity area ratio threshold TPh, it indicates that the surface porosity distribution of the UHPC component is outside the allowable range, and there are pitting or hole defects. It is determined that there is a risk of abnormal surface porosity, triggering a second warning instruction and generating a second strategy: clustering and marking the concentrated distribution area of ​​surface porosity in the detection interface and generating a surface porosity density heat map; at the same time, linking the automated quality inspection recording equipment to include the current area in the "requiring repair and review" list; if the surface porosity distribution spatially overlaps with the crack area, the defect linkage assessment mechanism is triggered, and step five is entered.

[0040] The method for obtaining the surface porosity area ratio threshold TPh is as follows: By detecting and statistically analyzing the surface porosity distribution of a large number of UHPC components under different mix proportions and casting process conditions, the ratio range of the surface porosity detection area to the total detection area is extracted. Combined with the material apparent density requirements and the durability standards of the service environment, a reasonable critical value for the surface porosity area ratio is determined. Referring to the concrete appearance quality assessment indicators in industry standards, the allowable range of surface porosity provided by equipment manufacturers, and the experience judgment of quality inspection experts, this threshold is formulated to accurately reflect the acceptable degree of apparent density of UHPC components, promptly identify the risk of surface pitting or void defects, and improve the appearance quality and long-term durability of components.

[0041] In this embodiment, by performing dimensionless calculations on the surface pore detection area Apore and the total detection area Atot, the surface pore index Ph is obtained and compared with the preset surface pore area ratio threshold TPh. This enables a quantitative assessment of the surface pore distribution of UHPC components. It can not only identify pitting or hole-like defects in a timely manner, but also automatically generate surface pore thermal maps and repair checklists under abnormal conditions. Furthermore, it enables spatial linkage judgment with crack detection results, thereby improving the accuracy of defect identification and the level of intelligent quality inspection and handling.

[0042] Example 7 This embodiment is an explanation based on Embodiment 6. Please refer to it. Figure 1 Specifically, step four includes: S41. Based on the illumination equalization enhanced image and the wet enhancement image in the high-quality image dataset; use CIELab color space conversion to extract color components from the defect area and the reference area respectively; extract the lightness component L2*, red-green axis color component a2* and yellow-blue axis color component b2* of the reference area in the illumination equalization enhanced image, and extract the lightness component L1*, red-green color component a1* and yellow-blue color component b1* of the defect area in the wet enhancement image.

[0043] In this embodiment, by extracting the CIELab color components L2*, a2*, and b2* of the reference region from the enhanced image under uniform illumination, and combining them with the corresponding components L1*, a1*, and b1* of the defect region in the enhanced wet image, the surface color information of the UHPC component is obtained in a refined manner by region. This effectively avoids the influence of uneven illumination and reflection interference on the detection results, thereby improving the accuracy and stability of color difference defect identification.

[0044] Example 8 This embodiment is an explanation based on Embodiment 7. Please refer to it. Figure 1 Specifically, step four also includes: S42. By combining the extracted lightness component L2*, red-green axis color component a2*, and yellow-blue axis color component b2* of the reference area with the lightness component L1*, red-green color component a1*, and yellow-blue color component b1* of the defect area, after dimensionless processing, the color difference index Dc is calculated using the following formula:

[0045] S43. By using a preset color difference judgment threshold TDc, and comparing and analyzing the color difference index Dc with the color difference judgment threshold TDc, the third evaluation result is obtained, including: When the color difference index Dc < the color difference judgment threshold TDc, it means that there are no color difference defects on the surface of the UHPC component, and continuous monitoring is required. When the color difference index Dc ≥ the color difference judgment threshold TDc, it indicates that there is a color difference defect on the surface of the UHPC component, triggering the third early warning instruction and generating the third strategy: generate a pseudo-color enhanced display for the color difference area in the detection interface, and simultaneously generate a process abnormality prompt, prompting quality inspectors to check the curing process, casting process, UHPC workability control and raw material batch consistency; if the color difference area overlaps with the surface pore or crack area, the composite defect risk judgment mechanism is triggered, and proceed to step five.

[0046] Color difference defects may be caused not only by curing process issues, such as insufficient covering conditions, improper temperature and humidity control, insufficient or excessive curing time, but also by the casting process, such as stratification, residual air bubbles, and uneven slurry distribution during molding. At the same time, the workability of UHPC also affects surface color difference. When the slump and fluidity are insufficient or excessive, it is easy to cause slurry separation or bleeding, thus causing local color differences. In addition, batch differences of raw materials are also a factor that cannot be ignored. Differences in cement, admixtures, and sand ratios from different sources or in different proportions can all cause uneven surface color.

[0047] The control strategies for color difference defects include: First, in terms of curing process, temperature and humidity control parameters need to be optimized to ensure sufficient coverage and avoid abnormal surface strength and color due to insufficient curing time or over-curing; Second, in terms of casting process, by improving the casting method and vibration process, stratification and air bubble residue can be reduced to ensure uniform slurry distribution; Third, in terms of UHPC workability, the slump and flowability range can be adjusted to avoid insufficient flow due to excessive viscosity or separation and bleeding of slurry due to excessive thinness, thus maintaining stable molding performance; Finally, in terms of raw material batch management, a quality traceability mechanism for cement, admixtures and sand ratio should be established to reduce color difference caused by batch differences, and real-time process parameter correction should be carried out in combination with AI detection feedback results to achieve dynamic control of the entire process from raw materials to molding to curing.

[0048] The method for obtaining the color difference judgment threshold TDc is as follows: By statistically analyzing the surface color data of a large number of UHPC components under different raw material batches, curing environments, and molding process conditions, the distribution range of the color difference index between defect areas and reference areas is extracted. Combined with material uniformity requirements and appearance quality standards, a reasonable color difference judgment threshold is determined. Referring to the relevant color difference evaluation standards of the CIELab color space, industry appearance consistency specifications, and manufacturers' process control parameters, and combined with the experience of quality inspection experts, this threshold is formulated to accurately reflect the color difference sensitivity of UHPC components, promptly identify material batch differences or process abnormalities, and ensure the visual consistency and process stability of components.

[0049] In this embodiment, the color difference index Dc is calculated and compared with a preset threshold TDc to achieve quantitative judgment of color difference defects on the surface of UHPC components. When an abnormality is detected, the color difference area can be highlighted in the interface in a pseudo-color manner, and a process abnormality prompt can be generated in conjunction with it. This helps quality inspectors to quickly locate the problem area and trace possible deviations in raw materials or curing processes, greatly improving the visualization and traceability efficiency of color difference defect detection.

[0050] Example 9 This embodiment is an explanation based on Embodiment 8. Please refer to it. Figure 1 Specifically, step five includes: S51. When any two or more of the crack index Cr, surface porosity index Ph, and color difference index Dc simultaneously exceed their respective thresholds, the defect linkage assessment mechanism is automatically activated, and the linkage assessment results include: If the crack area and the surface pore area overlap in space, it indicates that the defect poses a structural risk. A linkage strategy is generated: automatically create a three-dimensional spatial defect distribution map, give priority to the display of overlapping areas, and conduct structural performance verification and detection. If the crack area overlaps with the color difference area, it indicates that maintenance or material process issues may lead to crack expansion. A linkage strategy is generated: the process monitoring database is linked to retrieve maintenance curves and raw material batch information, and material batch traceability and maintenance process review are performed. If the surface pore area overlaps with the color difference area, it indicates that there is a coupling defect between surface density and raw material uniformity. A linkage strategy is generated: the current area is marked as a "composite appearance anomaly area" and a local compensation repair and secondary imaging re-inspection process is triggered. S52. When the crack index Cr, surface porosity index Ph, and color difference index Dc all exceed their respective thresholds, the system automatically triggers the composite defect risk assessment mechanism and obtains the risk assessment results, including: The UHPC component surface was found to have multiple superimposed defects, posing serious risks to both appearance and structure. A composite defect handling strategy was generated: the composite defect area was automatically classified as a "high-risk component" and simultaneously uploaded to the quality traceability database; an "stop use and scrap review" instruction was automatically generated, and the quality inspection scheduling system was linked to freeze the current UHPC component's factory release authority; the "process correction and feedback" sub-process was initiated, and the production process parameters were retrospectively reviewed based on historical inspection data to generate correction suggestions; and a re-inspection instruction was triggered, so that components in the same batch underwent stricter testing to prevent the spread of batch defects.

[0051] In this embodiment, by constructing a linkage assessment and composite defect risk judgment mechanism based on the crack index Cr, surface porosity index Ph, and color difference index Dc, spatial coupling detection and risk classification control of multiple types of defects in UHPC components are achieved. This not only automatically generates three-dimensional defect distribution maps and process traceability information, but also automatically triggers the "high-risk component" classification and factory freeze mechanism when multiple defects coexist. It also generates process correction and re-inspection instructions based on historical inspection data, thereby significantly improving the intelligence, traceability, and batch defect control capabilities of the quality inspection process.

[0052] Example 10 Please refer to Figure 2 The UHPC appearance quality inspection system based on image sensors includes: The multi-source image acquisition and enhancement module is used to improve the surface image quality of UHPC components through multi-source image acquisition and enhancement methods; it uses polarization filters to suppress reflection interference and acquire basic images; it extracts the spatial features of depressions and holes through structured light stripe distortion detection; it achieves balanced illumination enhancement by combining adaptive supplementary lighting; it uses atomized water film to improve the contrast of cracks and holes; and it integrates multiple images to construct a high-quality image data set. The crack monitoring module is used to extract the gray-level uniformity parameter Gref of the reference area on the surface of the component from the base image after reflection suppression and the enhanced image with equal illumination from the high-quality image dataset, and extract the abnormal gray-level distribution parameter Gabn of the crack area from the enhanced image. Combined with the crack width uw and depth ud, the crack index Cr is calculated and compared with the crack threshold Tcr to determine whether there is a risk of crack defects in the UHPC component. If so, an appropriate strategy is given. The surface porosity monitoring module is used to extract the surface porosity detection area Apore and the total detection area Atot based on the spatial feature data of depressions and pores in the high-quality image dataset and the image after wetting enhancement. It calculates the surface porosity index Ph and compares it with the surface porosity area ratio threshold TPh to determine whether the surface porosity distribution of the UHPC component is within the allowable range. If not, it provides an appropriate strategy. The color difference monitoring module is used to extract the lightness component L2*, red-green axis color component a2*, and yellow-blue axis color component b2* of the reference area from the high-quality image dataset, and combine them with the lightness component L1*, red-green color component a1*, and yellow-blue color component b1* of the defect area to calculate the color difference index Dc. The index Dc is then compared with the color difference judgment threshold TDc to determine whether there is a color difference defect on the surface of the UHPC component. If a defect exists, an appropriate strategy is applied. The defect linkage and risk assessment module is used to activate the defect linkage assessment mechanism when any two types of defect indices exceed the threshold at the same time. Based on the spatial overlap of defects, it generates linkage strategies for structural review, process tracing, or local repair and re-inspection. When all three types of indices exceed the threshold, the composite defect risk assessment mechanism is triggered, which automatically determines that the component has serious composite defects and triggers high-risk classification, factory freeze, process correction, and stricter re-inspection of the same batch.

[0053] In this embodiment, the present invention establishes a complete detection system from image quality improvement and single-type defect identification to multi-defect coupling judgment by setting up modules such as multi-source image acquisition and enhancement, crack monitoring, surface porosity monitoring, color difference monitoring, and defect linkage and risk assessment. It can not only achieve high-precision identification of typical surface defects such as cracks, surface porosity and color difference in the early stage, but also automatically trigger linkage strategies and risk classification and handling when multiple types of defects occur at the same time. Thus, it takes into account the accuracy of defect detection, the operability of process traceability and the level of intelligent quality control, and significantly improves the appearance quality inspection and risk prevention capabilities of UHPC components.

[0054] The threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by those skilled in the art for each set of sample data; as long as it does not affect the ratio between the parameter and the quantized value, it is acceptable.

[0055] The above formulas are all derived from software simulation using a large amount of data, and are selected to be close to the actual values. The coefficients in the formulas are set by those skilled in the art based on the actual situation. The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or changes made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for detecting the quality of the appearance of UHPC based on an image sensor, characterized in that, Includes the following steps: Step 1: Improve the surface image quality of UHPC components through multi-source image acquisition and enhancement techniques; use polarizing filters to suppress reflection interference and acquire basic images; Spatial features of depressions and holes were extracted using structured light stripe distortion detection. Combine adaptive supplemental lighting to achieve balanced and enhanced illumination; utilize atomized water film to improve the contrast between cracks and holes; Integrate multiple images to construct a high-quality image dataset; Step 2: Based on the base image after reflection suppression and the enhanced image with equalization in the high-quality image dataset, extract the gray-level uniformity parameter Gref of the reference area on the surface of the component from the base image, extract the abnormal gray-level distribution parameter Gabn of the crack area from the enhanced image, and calculate the crack index Cr by combining the crack width uw and depth ud with the crack threshold Tcr. This will help determine whether the UHPC component has a risk of crack defects. If so, appropriate strategies will be given. Step two includes: S21. Based on the base image after reflection suppression and the illumination equalization enhancement image in the high-quality image dataset, extract the gray-level uniformity parameter Gref of the reference area of ​​the component surface from the base image, extract the abnormal gray-level distribution parameter Gabn of the crack area from the enhancement image, and combine the crack width uw and depth ud. S22, based on the enhanced image, extract the geometric feature parameters of the crack region, including the crack width uw, the crack depth estimation value ud, and the gray contrast abnormality degree Gabn, and combine the gray uniformity parameter Gref of the reference region. After non-dimensional processing, the crack index Cr is calculated and obtained S23. By setting a crack threshold Tcr and comparing the crack index Cr with the crack threshold Tcr, the first evaluation results are obtained, including: When the crack index Cr < crack threshold Tcr, it means that the width and depth of the cracks on the surface of the UHPC component are within the allowable range, there is no risk of crack defects, the current detection status is maintained and monitoring continues. This indicates that the width or depth of cracks on the surface of the UHPC component exceeds the allowable range, posing a risk of crack defects. This triggers the first warning instruction and generates the first strategy: pixel-level boundary marking of the abnormal crack area is highlighted in the detection interface, and an audio-visual prompt device is activated to alert quality inspectors; a dynamic tracking subprocess is initiated to continuously collect and update the crack extension length and direction. If the crack propagation rate further increases, a crack risk level upgrade mechanism is triggered, generating an emergency response strategy of "stopping use and scrapping for review". Step 3: Based on the spatial feature data of depressions and holes in the high-quality image dataset, combined with the image after wetting enhancement, extract the surface pore detection area Apore and the total detection area Atot, calculate the surface pore index Ph, and compare it with the surface pore area ratio threshold TPh to determine whether the surface pore distribution of the UHPC component is within the allowable range. If not, give the corresponding strategy. Step 4: Based on the illumination equalization enhanced image and the wet enhancement image in the high-quality image dataset, the lightness component L2*, red-green axis color component a2*, and yellow-blue axis color component b2* of the reference area are extracted and combined with the lightness component L1*, red-green color component a1*, and yellow-blue color component b1* of the defect area to calculate the color difference index Dc. The index Dc is then compared with the color difference judgment threshold TDc to determine whether there is a color difference defect on the surface of the UHPC component. If so, an appropriate strategy is given. Step 5: When any two types of defect indices exceed the threshold at the same time, the defect linkage assessment mechanism is activated, and linkage strategies for structural review, process tracing, or local repair and re-inspection are generated according to the spatial overlap relationship of defects. When all three types of indices exceed the threshold, the composite defect risk judgment mechanism is triggered, and the component is automatically judged to have serious composite defects, triggering high-risk classification, factory freezing, process correction, and stricter re-inspection of the same batch.

2. The image sensor-based UHPC appearance quality detection method according to claim 1, characterized in that, Step one includes: S11. Image sensors are deployed on the surface of the UHPC component, and polarization filters are installed to perform surface imaging acquisition on the component, monitor and suppress reflection interference in the bright areas, and obtain the basic image after reflection suppression. S12. Project structured light stripes onto the surface of the component, collect stripe deformation data through an image sensor, and use distortion analysis to detect tiny pits and pores on the surface to obtain spatial feature data of the pits and pores. S13. During the acquisition process, in conjunction with the adaptive lighting system, the brightness and angle of the light source are automatically adjusted based on the image histogram entropy value feedback to enhance the overall brightness uniformity and obtain an image with balanced and enhanced illumination. S14. Atomize the surface of the component to form a thin water film. Improve the contrast between cracks and holes through the optical effect of the water film and obtain the image after wetting enhancement. S15. Integrate the base image after reflection suppression, the spatial feature data of depressions and holes, the image after illumination equalization enhancement, and the image after wetting enhancement to establish a high-quality image dataset.

3. The UHPC appearance quality inspection method based on an image sensor according to claim 1, characterized in that, Step three includes: S31. Based on the spatial feature data of depressions and holes in the high-quality image dataset, combined with the wet-enhanced image; the spatial feature data is used to perform three-dimensional structural analysis of the hole morphology to determine the geometric distribution and boundary range of the hole; at the same time, the contrast between the hole edge and the background is enhanced by the wet-enhanced image, the surface pore detection area Apore is extracted, and the total detection area Atot is obtained from the overall image.

4. The UHPC appearance quality inspection method based on an image sensor according to claim 3, characterized in that, Step three also includes: S32. After dimensionless processing of the extracted surface pore detection area Apore and the total detection area Atot, the surface porosity index Ph is calculated. S33. By setting a preset surface pore area ratio threshold TPh, and comparing and analyzing the surface porosity index Ph with the surface pore area ratio threshold TPh, the second evaluation result is obtained, including: When the surface porosity index Ph < the surface porosity area ratio threshold TPh, it indicates that the surface porosity distribution of the UHPC component is within the allowable range, the apparent compactness of the component meets the requirements, and continuous monitoring is required. When the surface porosity index Ph is greater than or equal to the surface porosity area ratio threshold TPh, it indicates that the surface porosity distribution of the UHPC component is outside the allowable range, and there are pitting or hole defects. It is determined that there is a risk of abnormal surface porosity, triggering a second warning instruction and generating a second strategy: clustering and marking the concentrated distribution area of ​​surface porosity in the detection interface and generating a surface porosity density heat map; at the same time, linking the automated quality inspection recording equipment to include the current area in the "needs repair and review" list; if the surface porosity distribution spatially overlaps with the crack area, the defect linkage assessment mechanism is triggered, and step five is entered.

5. The UHPC appearance quality inspection method based on an image sensor according to claim 4, characterized in that, Step four includes: S41. Based on the illumination equalization enhanced image and the wet enhancement image in the high-quality image dataset; use CIELab color space conversion to extract color components from the defect area and the reference area respectively; extract the lightness component L2*, red-green axis color component a2* and yellow-blue axis color component b2* of the reference area in the illumination equalization enhanced image, and extract the lightness component L1*, red-green color component a1* and yellow-blue color component b1* of the defect area in the wet enhancement image.

6. The UHPC appearance quality inspection method based on an image sensor according to claim 5, characterized in that, Step four also includes: S42. By combining the extracted lightness component L2*, red-green axis color component a2*, and yellow-blue axis color component b2* of the reference area with the lightness component L1*, red-green color component a1*, and yellow-blue color component b1* of the defect area, and after dimensionless processing, the color difference index Dc is calculated. S43. By using a preset color difference judgment threshold TDc, and comparing and analyzing the color difference index Dc with the color difference judgment threshold TDc, the third evaluation result is obtained, including: When the color difference index Dc < the color difference judgment threshold TDc, it means that there are no color difference defects on the surface of the UHPC component, and continuous monitoring is required. When the color difference index Dc ≥ the color difference judgment threshold TDc, it indicates that there is a color difference defect on the surface of the UHPC component, triggering the third early warning instruction and generating the third strategy: generate a pseudo-color enhanced display for the color difference area in the detection interface, and simultaneously generate a process abnormality prompt, prompting quality inspectors to check the curing process, casting process, UHPC workability control and raw material batch consistency; if the color difference area overlaps with the surface pore or crack area, the composite defect risk judgment mechanism is triggered, and proceed to step five.

7. The UHPC appearance quality inspection method based on an image sensor according to claim 6, characterized in that, Step five includes: S51. When any two or more of the crack index Cr, surface porosity index Ph, and color difference index Dc simultaneously exceed their respective thresholds, the defect linkage assessment mechanism is automatically activated, and the linkage assessment results include: If the crack area and the surface pore area overlap in space, it indicates that the defect poses a structural risk. A linkage strategy is generated: automatically create a three-dimensional spatial defect distribution map, give priority to the display of overlapping areas, and conduct structural performance verification and detection. If the crack area overlaps with the color difference area, it indicates that maintenance or material process issues may lead to crack expansion. A linkage strategy is generated: the process monitoring database is linked to retrieve maintenance curves and raw material batch information, and material batch traceability and maintenance process review are performed. If the surface pore area overlaps with the color difference area, it indicates that there is a coupling defect between surface compactness and raw material uniformity. A linkage strategy is generated: the current area is marked as a "composite appearance anomaly area" and a local compensation repair and secondary imaging re-inspection process is triggered. S52. When the crack index Cr, surface porosity index Ph, and color difference index Dc all exceed their respective thresholds, the system automatically triggers the composite defect risk assessment mechanism and obtains the risk assessment results, including: The UHPC component surface was found to have multiple superimposed defects, posing serious risks to both appearance and structure. A composite defect handling strategy was generated: the composite defect area was automatically classified as a "high-risk component" and simultaneously uploaded to the quality traceability database; an "stop use and scrap review" instruction was automatically generated, and the quality inspection scheduling system was linked to freeze the current UHPC component's factory release authority; the "process correction and feedback" sub-process was initiated, and the production process parameters were retrospectively reviewed based on historical inspection data to generate correction suggestions; and a re-inspection instruction was triggered, so that components in the same batch underwent stricter testing to prevent the spread of batch defects.

8. An image sensor-based UHPC appearance quality inspection system, applied to the image sensor-based UHPC appearance quality inspection method according to any one of claims 1 to 7, characterized in that, include: The multi-source image acquisition and enhancement module is used to improve the surface image quality of UHPC components through multi-source image acquisition and enhancement methods; A polarizing filter is used to suppress reflection interference and obtain the basic image. Spatial features of depressions and holes were extracted using structured light stripe distortion detection. Combine adaptive supplemental lighting to achieve balanced and enhanced illumination; utilize atomized water film to improve the contrast between cracks and holes; Integrate multiple images to construct a high-quality image dataset; The crack monitoring module is used to extract the gray-level uniformity parameter Gref of the reference area on the surface of the component from the base image after reflection suppression and the enhanced image with equal illumination from the high-quality image dataset, and extract the abnormal gray-level distribution parameter Gabn of the crack area from the enhanced image. Combined with the crack width uw and depth ud, the crack index Cr is calculated and compared with the crack threshold Tcr to determine whether there is a risk of crack defects in the UHPC component. If so, an appropriate strategy is given. The surface porosity monitoring module is used to extract the surface porosity detection area Apore and the total detection area Atot based on the spatial feature data of depressions and pores in the high-quality image dataset and the image after wetting enhancement. It calculates the surface porosity index Ph and compares it with the surface porosity area ratio threshold TPh to determine whether the surface porosity distribution of the UHPC component is within the allowable range. If not, it provides an appropriate strategy. The color difference monitoring module is used to extract the lightness component L2*, red-green axis color component a2*, and yellow-blue axis color component b2* of the reference area from the high-quality image dataset, and combine them with the lightness component L1*, red-green color component a1*, and yellow-blue color component b1* of the defect area to calculate the color difference index Dc. The index Dc is then compared with the color difference judgment threshold TDc to determine whether there is a color difference defect on the surface of the UHPC component. If a defect exists, an appropriate strategy is applied. The defect linkage and risk assessment module is used to activate the defect linkage assessment mechanism when any two types of defect indices exceed the threshold at the same time. Based on the spatial overlap of defects, it generates linkage strategies for structural review, process tracing, or local repair and re-inspection. When all three types of indices exceed the threshold, it triggers the composite defect risk assessment mechanism, automatically determines that the component has serious composite defects, and triggers high-risk classification, factory freeze, process correction, and stricter re-inspection of the same batch.

Images