Concrete pouring quality detection method based on visual identification

Abstract

The invention provides a concrete pouring quality detection method based on visual identification, and relates to the technical field of concrete quality detection.The method comprises the steps that a crack area is screened out according to different drying speeds of the crack area and a normal area in the drying process, and the drying process of the peripheral area of the crack is captured through pouring face images at different times; a crack line is obtained according to a crack area, the type of the crack line is judged according to the number of endpoints of the crack line, different weights are given, the drying duration is judged according to the initial time and the end time of the drying process, the drying duration represents the water storage capacity in the crack to judge the depth of the crack, and the size information of the crack is reflected according to the number of pixel blocks occupied by the crack line. Therefore, each crack score is constructed, early warning is carried out according to the score threshold, the type of the crack is judged while the crack is identified, and a quantitative index is formed according to the damage degree of the crack score to the crack.

Description

Technical Field

[0001] This invention relates to the field of concrete quality testing technology, specifically a method for testing the quality of concrete pouring based on visual recognition. Background Technology

[0002] Concrete pouring is a core part of engineering construction, and its quality directly affects the overall quality of the project. Improper operation during pouring can lead to cracks, such as network-like drying cracks caused by improper mix proportions and stress cracks caused by unreasonable structural design. Different types of cracks require different rectification methods; therefore, it is necessary to assess the nature and severity of the cracks. Due to the large area of ​​the poured surface and the fact that some cracks are not very obvious, they are difficult to find directly in a dry state. However, as the poured surface dries, the wet and dry areas show a significant color difference, making it easier to locate cracks. Furthermore, the moisture stored in the cracks causes them to dry more slowly than normal areas, greatly aiding in crack identification. However, in practice, gravity, especially in inclined and vertical poured surfaces, causes water to flow downwards, making it difficult to obtain a complete picture of the moisture around the cracks at any given moment. Therefore, a method is needed to capture the crack area throughout the drying process and to assess the crack information.

[0003] The prior art, disclosed in CN120339222A, is a method and system for detecting cracks in hydraulic engineering based on intelligent visual recognition. This method acquires crack images, enhances brightness and contrast in different regions, extracts and optimizes crack edges, performs morphological processing for noise reduction and smoothing, segments crack regions and determines connectivity, analyzes geometric features to detect connectivity, and calibrates crack positions to obtain localization results. This invention, through dynamic adjustment of image brightness and contrast, makes cracks more prominent in complex backgrounds, especially under uneven lighting, reducing the difficulty of crack identification caused by inconsistent lighting. Morphological operations optimize crack edges, filling in missing parts of the crack, smoothing crack contours, and eliminating noise, thereby ensuring the clarity and integrity of crack features. In the crack segmentation stage, adaptive thresholding and geometric feature analysis accurately determine crack fracture and connectivity, achieving localization of crack endpoints and their continuous regions.

[0004] While publicly available technical documents can identify cracks in the cast surface, they cannot provide information on the type or depth of the cracks, thus failing to offer a reference for assessing the potential harm caused by the cracks.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a visual recognition-based method for detecting the quality of concrete pouring, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: The concrete pouring quality inspection method based on vision recognition includes the following steps: Step 1: Spray the pouring surface and acquire single-channel images of the pouring surface at equal time intervals during its drying process. Establish the same coordinate system for each pixel block in the pouring surface image and add coordinates. Step 2: Based on each poured surface image, construct the gradient value of each pixel block according to the pixel value changes of the eight neighboring areas around each pixel block, set the gradient threshold, and mark all coordinates of all poured surface images that have gradient values ​​exceeding the gradient threshold on the blank image to obtain the crack mask image. Step 3: Analyze the marked pixel blocks in the crack mask image to obtain each independent crack region, obtain crack lines according to the ridge detection algorithm, and determine whether the crack is a stress crack or a drying crack based on the number of endpoints of the crack line, and assign different weights. Step 4: Record the generation time of each pixel block in the mask image that is marked because the gradient value exceeds the gradient threshold. For each crack region, the drying time is the time interval between the generation of the first marked pixel block and the generation of the last marked pixel block within its range. Combine the number of pixel blocks occupied by each crack line and the weight to construct a crack score for each crack. Set a score threshold and determine whether to issue an alert based on the relationship between the crack and the score threshold.

[0008] Furthermore, water is sprayed onto the entire cast surface so that water flows over every part of the cast surface. Images of the poured surface are acquired at a fixed position and at a fixed frequency using a single-channel camera until the poured surface is completely dry. All poured surface images are then sorted according to time sequence. The same Cartesian coordinate system is constructed for each poured surface image, with the unit length of the coordinate system being the width of a pixel block. Coordinates are labeled for each pixel block.

[0009] Furthermore, the gradient value of each pixel block is determined for each poured surface image, with the following logic; Within a 3x3 grid centered on the current pixel block, the pixel value change rates of the top-left and bottom-right cells, the top-right and bottom-left cells, and the bottom-left and top-right cells are obtained respectively, and the gradient value of the current pixel block is constructed based on the following formula: in, This represents the gradient value of the current pixel block. This indicates the rate of change of pixel values ​​between the top-left and bottom-right cells. This indicates the rate of change of pixel values ​​between the top and bottom pixels. This indicates the rate of change of pixel values ​​between the bottom left and top right cells; The gradient values ​​of the outermost pixel block of the poured surface image are not calculated.

[0010] Furthermore, a gradient threshold is set, and the coordinates of all pixel blocks whose gradient values ​​exceed the gradient threshold in all the poured surface images are recorded. A blank image with the same specifications as the poured surface images is constructed, and the same coordinate system is established in the blank image. The coordinates of all pixel blocks whose gradient values ​​exceed the gradient threshold in all the poured surface images are marked in the blank image. The pixel values ​​of the marked pixel blocks are changed to 0, and the pixel values ​​of the remaining pixel blocks are marked as 255. The image formed at this time is marked as a crack mask image.

[0011] Furthermore, the crack mask image is filtered to remove noise, and each independent crack region is obtained through connectivity analysis. The crack region represents the set of pixel blocks with a pixel value of 0 in the crack mask image. The Euclidean distance between each pixel block in the crack region and the nearest boundary is calculated and marked in the pixel block. The center line of the crack region is obtained through the ridge detection algorithm and marked as the crack line. The logic of the ridge detection algorithm is as follows: scan the Euclidean distance marked in each pixel block. If its value is greater than or equal to the average Euclidean distance of the surrounding 8 adjacent pixel blocks, it is marked as a candidate point. The center line is obtained by linearly connecting the set of candidate points.

[0012] Furthermore, for each crack line obtained through an independent crack region, the number of its endpoints is obtained. If the number of endpoints is 2, the crack line is marked as a stress crack; if the number of endpoints exceeds 2, the crack line is marked as a drying crack. The weights for stress crack generation are respectively, with the weight of stress crack being greater than that of drying crack.

[0013] Furthermore, for each crack region, the acquisition time of the first pixel block in the crack region whose gradient value exceeds the gradient threshold and the acquisition time of the poured surface image that generated the pixel block are marked as the initial time. The acquisition time of the last pixel block in the crack region whose gradient value exceeds the gradient threshold and the acquisition time of the poured surface image that generated the pixel block are marked as the final time. The duration between the initial time and the final time is marked as the drying duration. Obtain the drying time for each crack region.

[0014] Furthermore, the number of pixels occupied by each crack line is obtained to construct a crack score, based on the following formula: in, Indicates crack score, This indicates the number of pixels occupied by the crack line. Indicates drying time. Indicates weight, ,when hour, Represents the weight of stress cracks, when hour, This indicates the weight of drying cracks.

[0015] Furthermore, a scoring threshold is set. When the crack score exceeds the scoring threshold, an early warning is issued for the crack, and the crack location, the corresponding crack line, and the drying time are marked in the image of the poured surface. Otherwise, no warning is issued.

[0016] Compared with the prior art, the beneficial effects of the present invention are: This invention creates a humid environment on the cast surface to obtain images of the drying process. Cracked areas are identified based on their different drying rates compared to normal areas. Furthermore, the drying information from cast surface images at different times captures the complete drying process of the area surrounding the cracks, ensuring complete crack information is obtained even when drying times are inconsistent. Crack lines are obtained from the crack mask images, and their type is determined by the number of endpoints, assigning different weights accordingly. The drying duration is determined based on the initial and final times of the drying process for each crack area, and the drying duration represents the water content in the crack to determine its depth. The number of pixels occupied by the crack line reflects the crack's size information, thus constructing a score for each crack. Warnings are issued based on the score threshold, enabling simultaneous crack identification and type determination. The crack score provides a quantitative indicator of the crack's severity, offering data guidance for quality inspection. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall method flow of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0019] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0020] Example: Please see Figure 1 The present invention provides a technical solution: The concrete pouring quality inspection method based on vision recognition includes the following steps: Step 1: Spray the pouring surface and acquire single-channel images of the pouring surface at equal time intervals during its drying process. Establish the same coordinate system for each pixel block in the pouring surface image and add coordinates. Step 1 includes the following: Spray water onto the entire cast surface so that water flows over every part of the cast surface. Images of the poured surface are acquired at a fixed position and at a fixed frequency using a single-channel camera until the poured surface is completely dry. All poured surface images are then sorted according to time sequence. The same Cartesian coordinate system is constructed for each poured surface image, with the unit length of the coordinate system being the width of a pixel block. Coordinates are labeled for each pixel block.

[0021] As a preferred embodiment, the single-channel camera uses a monochrome camera, and the pixel blocks in the acquired image of the cast surface only have grayscale values, which facilitates subsequent gradient value analysis.

[0022] As a preferred embodiment, in accordance with the requirements of the "Technical Specification for Durability of Concrete Structures" (GB / T50476), tap water is sprayed evenly from one end of the pouring surface. The water temperature is close to the ambient temperature. A parallel scanning path is used to ensure coverage without dead corners, and the spraying is continuous. Spraying continues until a continuous thin water film forms on the surface, achieving a water film coverage rate of over 95% and a relative humidity of 98% or higher. Once these requirements are met, spraying is stopped, and the drying process begins. Randomly selected measurement points are used to measure the surface humidity of the poured surface. When the average relative humidity of all measurement points does not exceed 5% of the ambient humidity, and measurement points are randomly selected again after a 5-minute interval, and the average relative humidity obtained at this time is not lower than 5% of the previous measurement, the poured surface is considered completely dry.

[0023] In a preferred embodiment, the lower left corner of all the images of the poured surface is taken as the origin of the coordinate system, the horizontal direction to the right is taken as the positive direction of the X-axis, and the vertical direction upward is taken as the positive direction of the Y-axis.

[0024] By spraying water onto the pouring surface, a moist environment is created. In real-world scenarios, the upper part of inclined and vertical pouring surfaces dries at different rates than the lower part due to gravity, and this can also be simulated by spraying water onto the pouring surface.

[0025] By spraying and continuously covering the entire poured surface with water, uniform wetting is ensured, simulating the evaporation process of water in a natural drying environment. In actual engineering, this can realistically reproduce the surface change characteristics of the early drying stage after concrete pouring, avoiding detection deviations caused by the initial non-uniformity of dry poured surfaces. At the same time, single-channel images are acquired at a fixed frequency and sorted according to time series to establish a unified rectangular coordinate system and label the pixel block coordinates. This not only facilitates accurate tracking of the dynamic changes of each pixel block but also provides a standardized spatiotemporal data foundation for the entire detection process, ensuring seamless connection from image acquisition to crack analysis, improving the reliability and repeatability of detection. In concrete quality control, this has the practical significance of early detection of potential defects and reducing later maintenance costs.

[0026] Step 2: Based on each poured surface image, construct the gradient value of each pixel block according to the pixel value changes of the eight neighboring areas around each pixel block, set the gradient threshold, and mark all coordinates of all poured surface images that have gradient values ​​exceeding the gradient threshold on the blank image to obtain the crack mask image. Step 2 includes the following: Step 201: Determine the gradient value of each pixel block for each poured surface image, with the following logic; Within a 3x3 grid centered on the current pixel block, the pixel value change rates of the top-left and bottom-right cells, the top-right and bottom-left cells, and the bottom-left and top-right cells are obtained respectively, and the gradient value of the current pixel block is constructed based on the following formula: in, This represents the gradient value of the current pixel block. This indicates the rate of change of pixel values ​​between the top-left and bottom-right cells. This indicates the rate of change of pixel values ​​between the top and bottom pixels. This indicates the rate of change of pixel values ​​between the bottom left and top right cells; Dependent variable The comprehensive intensity reflects the multi-directional pixel value changes within the current pixel block's nine-grid layout. Its meaning lies in quantifying the abrupt changes in surface texture caused by moisture evaporation during the drying process of the poured surface, and capturing early crack edge differences; independent variable This highlights the problem of uneven moisture gradient along the diagonal (45 degrees), such as localized stress concentration during drying. This reflects a vertical humidity imbalance. By capturing the drying anomaly along the other diagonal (135 degrees), these independent variables collectively address the engineering pain point of uneven moisture diffusion in multiple dimensions after concrete pouring, avoiding omissions in a single direction. The correlation with the independent variable is based on multi-directional edge detection, and is fused into a scalar magnitude through the sum of squares and the square root. The positive correlation effect of the independent variable is manifested as a linear amplification of the square contribution for each increase, and a non-linear enhancement. This highlights the dominant changes and suppresses noise. Regarding magnitude relationships, when the independent variable increases... Monotonically increasing, directly proportional but with quadratic acceleration; for example, a significant increase in a single independent variable will result in a substantial increase. To highlight anomalies, and conversely, to maintain a low value for filtering smooth areas when the balance is small, thus accurately distinguishing normal textures from crack precursors.

[0027] The gradient values ​​of the outermost pixel block of the poured surface image are not calculated.

[0028] During the drying process, because moisture is stored inside the cracks, the cracks, their edges, and a certain distance around them dry more slowly than other areas. This stored moisture causes a difference in drying speed compared to a normal poured surface. While some areas dry slowly and others quickly on a normal poured surface, the slower-drying areas are still in a state of near-drying, and the boundary between the dried and incompletely dried areas is blurred, resulting in smaller gradient values ​​in the pixel blocks. However, because of the moisture stored inside the cracks, the drying process is slower than on a normal poured surface. Therefore, there is a clearer boundary between the cracks and the already dried areas, resulting in larger gradient values ​​in the pixel blocks. This allows us to determine the boundary of the area around the crack.

[0029] By calculating gradient values ​​for the multi-directional pixel value change rate of each pixel block in the image of the poured surface, subtle differences in surface texture caused by moisture gradient during the drying process can be captured. This practically simulates the intensity abrupt change characteristics of crack edges, improves the sensitivity to early micro-cracks, and avoids omissions in single-direction detection. By not calculating the outermost pixel blocks, image boundary artifact interference is further reduced, ensuring the accuracy of gradient calculation and providing a reliable quantitative indicator for threshold screening in step 202. This achieves a smooth transition from the original image to the initial crack identification, and practically demonstrates efficient edge detection logic in concrete pouring quality assessment, helping to quickly identify risks caused by uneven drying on the engineering site.

[0030] Step 202: Set a gradient threshold, record the coordinates of all pixel blocks in all cast surface images whose gradient values ​​exceed the gradient threshold, construct a blank image with the same specifications as the cast surface images and establish the same coordinate system in the blank image, mark the coordinates of all pixel blocks in all cast surface images whose gradient values ​​exceed the gradient threshold in the blank image, change the pixel value of the marked pixel blocks to 0, and mark the pixel value of the remaining pixel blocks as 255, and mark the image formed at this time as the crack mask image.

[0031] In a preferred embodiment, the blank image and the casting surface image have the same pixel values, and the same X-axis and Y-axis are established with the lower left corner of the blank image as the origin, and the width of a single pixel block is used as the unit length.

[0032] During the drying process of cracks and their surrounding areas, the start and complete drying times vary greatly depending on factors such as gravity and water storage. For example, in the same crack, the upper part may be completely dry, but the lower part may still have a wet area. This affects the analysis of crack morphology. Since the wet area gradually shrinks towards the crack during the drying process, by controlling the time interval of acquiring images of the poured surface, the boundaries of the wet area around the crack and its surroundings at each moment can be captured. This allows for the layer-by-layer depiction of the wet area around the crack affected by its water storage. Even if the start and end times of drying for each crack are large, the wet area around the crack can still be depicted. Furthermore, the wet conditions on both sides of the crack are almost symmetrical, which facilitates the next step of finding the crack line.

[0033] The process of setting a gradient threshold and recording the coordinates of pixels exceeding the threshold to construct a crack mask image on a blank image effectively transforms scattered gradient anomalies into a binary spatial representation. This simplifies the initial location of cracks in practical engineering applications, reduces computational complexity, and improves noise robustness. By labeling pixels exceeding the threshold as 0 and the rest as 255 in the mask format, it not only facilitates visualization analysis but also directly connects to the region segmentation and centerline extraction in step 3, ensuring the continuity from gradient detection to crack region extraction. In practical terms, this means generating operable intermediate results for concrete quality inspection, supporting automated processing of large-scale pouring surface data, and providing timely feedback on pouring defects to optimize construction processes.

[0034] Step 3: Analyze the marked pixel blocks in the crack mask image to obtain each independent crack region, obtain crack lines according to the ridge detection algorithm, and determine whether the crack is a stress crack or a drying crack based on the number of endpoints of the crack line, and assign different weights. Step 3 includes the following: Step 301: Filter the crack mask image to remove noise, obtain each independent crack region through connectivity analysis, the crack region represents the set of pixel blocks with a pixel value of 0 in the crack mask image, calculate the Euclidean distance between each pixel block in the crack region and the nearest boundary and mark it in the pixel block, obtain the center line of the crack region through the ridge detection algorithm, and mark the center line as the crack line; As a preferred embodiment, noise is removed by median filtering.

[0035] The logic of the ridge detection algorithm is as follows: scan the Euclidean distance marked in each pixel block. If its value is greater than or equal to the average Euclidean distance of the surrounding 8 adjacent pixel blocks, it is marked as a candidate point. The center line is obtained by linearly connecting the set of candidate points.

[0036] The crack mask image is filtered to remove noise, and independent crack regions are obtained through connectivity analysis. The center line is extracted by combining Euclidean distance annotation and ridge detection algorithm, which refines the geometric morphology description of the crack and avoids false cracks caused by environmental interference such as changes in light and shadow, thus improving the accuracy of detection and engineering reliability. The ridge detection is based on the candidate point connection logic of distance mean comparison, which further ensures the continuity and representativeness of the center line and provides stable path data for crack type judgment in step 302. It realizes efficient connection from mask binarization to crack line representation, which actually demonstrates the key role of noise suppression in concrete pouring quality control, supports in-depth analysis from macro areas to micro paths, and reduces the need for human intervention.

[0037] Step 302: For each crack line obtained through an independent crack region, obtain the number of its endpoints. If the number of endpoints is 2, mark the crack line as a stress crack. If the number of endpoints exceeds 2, mark the crack line as a drying crack. The weights for stress crack generation are respectively, with the weight of stress crack being greater than that of drying crack.

[0038] Drying cracks often only exist on the surface of the cast surface and do not penetrate into the cast layer, so their harm is often smaller or even negligible. Stress cracks, on the other hand, tend to penetrate into the interior of the cast layer, causing greater harm and even leading to the failure of the cast layer. Therefore, in the process of quantifying stress cracks, they are given a greater weight value to reflect the degree of harm caused by stress cracks.

[0039] Each crack line is classified as either a stress crack or a drying crack based on the number of endpoints, and a corresponding weight is assigned. This distinguishes the cause and potential hazard level of the crack. The higher weight of stress cracks reflects their greater threat to structural integrity, thereby guiding priority maintenance decisions and enhancing the practical value of the inspection. By using a simple counting logic of endpoint numbers, the computational burden of complex morphology analysis is avoided, and the weights are directly passed as quantitative factors to the scoring calculation in step 4, ensuring a seamless connection from crack classification to risk assessment. In the concrete quality inspection process, this is of practical significance because it transforms abstract crack characteristics into actionable risk indicators, supporting on-site engineers in developing targeted maintenance strategies based on weight differences.

[0040] Step 4: Record the generation time of each pixel block in the mask image that is marked because the gradient value exceeds the gradient threshold. For each crack region, the drying time is the time interval between the generation of the first marked pixel block and the generation of the last marked pixel block within its range. Combine the number of pixel blocks occupied by each crack line and the weight to construct a crack score for each crack. Set a score threshold and determine whether to issue an alert based on the relationship between the crack and the score threshold.

[0041] Step 4 includes the following: Step 401: Record the generation time of each pixel block in the mask image that is marked because its gradient value exceeds the gradient threshold. The generation time is the timestamp of the casting surface image from which the pixel block originates. For each crack region, obtain the acquisition time of the first pixel block in the crack region whose gradient value exceeds the gradient threshold and the acquisition time of the casting surface image that generated the pixel block, and mark it as the initial time. Obtain the acquisition time of the last pixel block in the crack region whose gradient value exceeds the gradient threshold and the acquisition time of the casting surface image that generated the pixel block, and mark it as the final time. The duration between the initial time and the final time is marked as the drying duration. Obtain the drying time for each crack region.

[0042] During the drying process of cracks, cracks with larger water storage often take longer to dry completely. Larger water storage means a deeper crack. The drying time reflects the water storage, and the water storage reflects the crack depth. Therefore, the moment when the first pixel block with a gradient value exceeding the gradient threshold appears in the crack region is the moment when the crack begins to dry independently compared to the normal poured surface. The moment when the last pixel block with a gradient value exceeding the gradient threshold appears is the moment when the crack completes drying independently. Even though gravity and other factors may prolong the drying time of the crack judged by this method, this prolongation is negligible in small cracks. When the crack is longer, the drying time judged by this method will be prolonged more, increasing the crack score during the calculation and reflecting a greater degree of harm. When the crack is long, its inherent harm is indeed significant. This effect of prolonged drying time is consistent with the construction logic of crack scoring.

[0043] The drying time from the initial time to the final time was calculated for each crack area, quantifying the dynamic process of crack evolution related to the moisture gradient on the concrete surface. This revealed the internal stress distribution problem that may be caused by uneven drying, thus providing temporal evidence for early intervention in pouring quality and improving the effectiveness of preventive maintenance in engineering. By tracking the image acquisition time of the first and last super-threshold pixel blocks, the accuracy and traceability of the duration calculation were ensured, and the drying time parameter was input into the scoring formula in step 402, achieving a smooth connection from spatiotemporal sequence to comprehensive index. This demonstrates the core value of time series analysis in the practical application of concrete testing, helping to optimize post-pouring covering and ventilation processes and reduce the risk of crack propagation.

[0044] Step 402: Obtain the number of pixels occupied by each crack line and construct a crack score based on the following formula: in, Indicates crack score, This indicates the number of pixels occupied by the crack line. Indicates drying time. Indicates weight, ,when hour, Represents the weight of stress cracks, when hour, This indicates the weight of drying cracks.

[0045] Dependent variable Specifically, it reflects the overall risk level of cracks, meaning that it quantifies the severity of cracks during the drying process of the poured surface, supporting early warning decisions; independent variable As a weight, it reflects the stress cracks in the real environment ( (High weight) poses a greater threat to structural damage, such as potential collapse caused by internal stress concentration; The number of pixels along the crack line reflects the scale of crack propagation and simulates the length of surface damage caused by uneven drying. As drying time, it captures the dynamic issues related to moisture evaporation time and crack evolution. These independent variables collectively address the practical pain points of early concrete curing, providing a multi-dimensional risk assessment for engineering risks caused by improper humidity control after pouring. The correlation with the independent variable stems from the risk product model, which integrates weights, size, and duration into a single indicator through linear multiplication. The effects are directly proportional and cumulative; an increase in each independent variable directly amplifies the effect. The value emphasizes the priority alert for high-weighted or long-duration cracks. Regarding the relationship between magnitude changes, the independent variable... , or When it increases Monotonically increasing, proportionally linearly amplified, for example Switching from low to high will multiply the effect. To highlight stress risk, conversely, if any independent variable is small... Keep the value low to filter out minor defects, thus accurately distinguishing between serious cracks that require intervention and negligible dry texture.

[0046] A scoring threshold is set. When the crack score exceeds the threshold, an early warning is issued for the crack, and the crack location, the corresponding crack line, and the drying time are marked on the image of the poured surface. Otherwise, no warning is issued.

[0047] Crack scores are calculated by multiplying the number of crack line pixels, drying time, and weights, and a threshold is set for early warning labeling. This integrates multi-dimensional features into a single risk quantification value, facilitating rapid on-site decision-making and report generation, avoiding biases in subjective judgment, and improving the automation level of concrete pouring quality management. By labeling crack locations, crack lines, and drying times in the pouring surface image, not only is visual feedback enhanced, but the output link of the entire detection chain is also completed, ensuring a closed-loop connection from score calculation to early warning response. In practical terms, this highlights the practicality of comprehensive assessment, supports real-time monitoring and data accumulation for large-scale engineering projects, and contributes to sustainable structural safety assurance.

[0048] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0049] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0050] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0051] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for inspecting the quality of concrete pouring based on visual recognition, characterized in that, The specific steps include: Step 1: Spray the pouring surface and acquire single-channel images of the pouring surface at equal time intervals during its drying process. Establish the same coordinate system for each pixel block in the pouring surface image and add coordinates. Step 2: Based on each poured surface image, construct the gradient value of each pixel block according to the pixel value changes of the eight neighboring areas around each pixel block, set the gradient threshold, and mark all coordinates of all poured surface images that have gradient values ​​exceeding the gradient threshold on the blank image to obtain the crack mask image. Step 3: Analyze the marked pixel blocks in the crack mask image to obtain each independent crack region, obtain crack lines according to the ridge detection algorithm, and determine whether the crack is a stress crack or a drying crack based on the number of endpoints of the crack line, and assign different weights. Step 4: Record the generation time of each pixel block in the mask image that is marked because the gradient value exceeds the gradient threshold. For each crack region, the drying time is the time interval between the generation of the first marked pixel block and the generation of the last marked pixel block within its range. Combine the number of pixel blocks occupied by each crack line and the weight to construct a crack score for each crack. Set a score threshold and determine whether to issue an alert based on the relationship between the crack and the score threshold.

2. The method for detecting the quality of concrete pouring based on visual recognition according to claim 1, characterized in that: Spray water onto the entire cast surface so that water flows over every part of the cast surface. Images of the poured surface are acquired at a fixed location and at a fixed frequency using a single-channel camera, and time stamps are recorded until the poured surface is completely dry. All poured surface images are then sorted according to time sequence, and the same Cartesian coordinate system is constructed for each poured surface image. The unit length of the coordinate system is the width of a pixel block, and the coordinates of each pixel block are labeled.

3. The method for detecting the quality of concrete pouring based on visual recognition according to claim 2, characterized in that: The gradient value of each pixel block is determined for each poured surface image, and the logic is as follows; Within a 3x3 grid centered on the current pixel block, the pixel value change rates of the top-left and bottom-right cells, the top-right and bottom-left cells, and the bottom-left and top-right cells are obtained respectively, and the gradient value of the current pixel block is constructed based on the following formula: in, This represents the gradient value of the current pixel block. This indicates the rate of change of pixel values ​​between the top-left and bottom-right cells. This indicates the rate of change of pixel values ​​between the top and bottom pixels. This indicates the rate of change of pixel values ​​between the bottom left and top right cells; The gradient values ​​of the outermost pixel block of the poured surface image are not calculated.

4. The method for detecting the quality of concrete pouring based on visual recognition according to claim 3, characterized in that: Set a gradient threshold, record the coordinates of all pixel blocks whose gradient values ​​exceed the gradient threshold in all the poured surface images, construct a blank image with the same specifications as the poured surface images and establish the same coordinate system in the blank image, mark the coordinates of all pixel blocks whose gradient values ​​exceed the gradient threshold in all the poured surface images in the blank image, change the pixel value of the marked pixel blocks to 0, and mark the pixel value of the remaining pixel blocks as 255, and mark the image formed at this time as the crack mask image.

5. The method for detecting the quality of concrete pouring based on visual recognition according to claim 4, characterized in that: The crack mask image is filtered to remove noise, and each independent crack region is obtained through connectivity analysis. The crack region represents the set of pixel blocks with a pixel value of 0 in the crack mask image. The Euclidean distance between each pixel block in the crack region and the nearest boundary is calculated and marked in the pixel block. The center line of the crack region is obtained through the ridge detection algorithm and marked as the crack line. The logic of the ridge detection algorithm is as follows: scan the Euclidean distance marked in each pixel block. If its value is greater than or equal to the average Euclidean distance of the surrounding 8 adjacent pixel blocks, it is marked as a candidate point. The center line is obtained by linearly connecting the set of candidate points.

6. The method for detecting the quality of concrete pouring based on visual recognition according to claim 5, characterized in that: For each crack line obtained through an independent crack region, the number of its endpoints is obtained. If the number of endpoints is 2, the crack line is marked as a stress crack. If the number of endpoints exceeds 2, the crack line is marked as a drying crack. The weights for stress crack generation are respectively, with the weight of stress crack being greater than that of drying crack.

7. The method for detecting the quality of concrete pouring based on visual recognition according to claim 6, characterized in that: Record the generation time of each pixel block in the mask image that is marked because the gradient value exceeds the gradient threshold. The generation time is the timestamp of the casting surface image from which the pixel block originates. For each crack region, obtain the generation time of the first pixel block in the crack region and mark it as the initial time. Obtain the generation time of the last pixel block in the crack region and mark it as the final time. The duration between the initial time and the final time is marked as the drying duration. Obtain the drying time for each crack region.

8. The method for detecting the quality of concrete pouring based on visual recognition according to claim 7, characterized in that: The number of pixels occupied by each crack line is obtained to construct a crack score, based on the following formula: in, Indicates crack score, This indicates the number of pixels occupied by the crack line. Indicates drying time. Indicates weight, ,when hour, Represents the weight of stress cracks, when hour, This indicates the weight of drying cracks.

9. The method for detecting the quality of concrete pouring based on visual recognition according to claim 8, characterized in that: A scoring threshold is set. When the crack score exceeds the threshold, an early warning is issued for the crack, and the crack location, the corresponding crack line, and the drying time are marked on the image of the poured surface. Otherwise, no warning is issued.

Images