A bridge engineering construction quality detection method and system

Abstract

The present invention relates to the technical field of bridge construction quality detection, and in particular to a bridge engineering construction quality detection method and system. The method comprises the following steps: periodically capturing detection images of the steel bar binding site during the bridge construction period to obtain clear bridge construction image data; identifying the main direction of steel bar binding on the clear bridge construction image data to generate optimized bridge steel bar binding skeleton data; performing bridge curved surface effect compensation processing on the optimized bridge steel bar binding skeleton data, and performing image pixel ratio scale mapping to generate steel bar binding spacing estimation values; obtaining bridge steel bar spacing design specification data; using the bridge steel bar spacing design specification data to perform construction quality scoring on the steel bar binding spacing estimation values, and generating an engineering quality detection report to obtain a bridge steel bar binding engineering quality detection report. The present invention can automatically detect the steel bar binding spacing during the bridge construction period, and realize fast and accurate spacing measurement and engineering quality assessment.

Description

Technical Field

[0001] The present invention relates to the technical field of bridge construction quality detection, and in particular to a bridge engineering construction quality detection method and system. Background Art

[0002] Bridge engineering is an important part of modern transportation infrastructure construction, and its quality is directly related to traffic safety and the stable development of social economy. As the most widely used structural form in bridge engineering, the quality of reinforced concrete structure has a vital impact on the bearing capacity, durability and safety of the bridge. In the construction process of reinforced concrete structure, the quality control of steel bar engineering is particularly important as it is a key link. As the core process in steel bar engineering, the quality of steel bar binding directly determines the integrity and stability of the steel bar skeleton, which in turn affects the bearing capacity and durability of the concrete structure. The steel bar binding spacing refers to the distance between adjacent steel bars in the reinforced concrete structure. According to the design requirements and specifications, steel bars in different parts and different stress states need to be tied at a specific spacing. Reasonable steel bar spacing can ensure the effective bonding between steel bars and concrete, ensure the overall stress performance of the steel bar skeleton, and thus ensure the bearing capacity and safety of the structure. At present, at the construction site of bridge projects, the traditional method of detecting the spacing between steel bars in binding mainly relies on manual operation. Experienced quality inspectors observe the arrangement of steel bars with the naked eye and use traditional measuring tools such as steel tape measures and rulers to measure the distance between steel bars one by one. This detection method is inefficient, especially in large-scale bridge projects. The number of steel bars that need to be inspected is huge, and measuring one by one is time-consuming and labor-intensive, which seriously affects the construction progress. In addition, in some narrow or hard-to-reach areas, the ruler measurement method is difficult to operate and it is difficult to ensure the accuracy of the measurement, which cannot meet the quality management requirements of bridge steel bar binding projects. Summary of the invention

[0003] Based on this, the present invention provides a bridge engineering construction quality detection method and system to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a bridge engineering construction quality detection method comprises the following steps:

[0005] Step S1: using a multispectral camera array to periodically capture images of the steel bar binding site during the bridge construction period to generate original bridge construction site image data; performing image contrast enhancement on the original bridge construction site image data to obtain clear bridge construction image data;

[0006] Step S2: Recognize the main direction of steel bar binding on the clear bridge construction image data to obtain the main direction data of steel bar binding; perform steel bar skeleton topology network processing according to the main direction data of steel bar binding to obtain initial steel bar skeleton topology data; identify the steel bar centerline according to the initial steel bar skeleton topology data, and optimize the skeleton endpoints to generate optimized bridge steel bar binding skeleton data;

[0007] Step S3: performing bridge surface effect compensation processing on the optimized bridge reinforcement binding skeleton data to generate a bridge surface transformation correction coefficient; performing image pixel scale mapping on the optimized bridge reinforcement binding skeleton data using the bridge surface transformation correction coefficient to obtain skeleton actual scale mapping data;

[0008] Step S4: mapping the clear bridge construction image data to binding nodes by optimizing the bridge reinforcement binding skeleton data to generate binding node pixel mapping data; performing distance scale conversion on the binding node pixel mapping data using the skeleton actual scale scale mapping data to generate a reinforcement binding spacing estimation value;

[0009] Step S5: Obtain bridge reinforcement spacing design specification data; use the bridge reinforcement spacing design specification data to calculate the spacing difference of the reinforcement binding spacing estimate, and perform a construction quality score to obtain bridge construction quality score data; generate a project quality inspection report based on the bridge construction quality score data to obtain a bridge reinforcement binding project quality inspection report.

[0010] Preferably, step S1 comprises the following steps:

[0011] Step S11: using a multispectral camera array to periodically capture images of the steel bar binding site during the bridge construction period to generate original bridge construction site image data;

[0012] Step S12: performing image noise evaluation based on the original bridge construction site image data to obtain shooting noise level data;

[0013] Step S13: performing multi-scale Gaussian blurring on the original bridge construction site image data by shooting noise level data to generate a multi-scale blur map of the construction site;

[0014] Step S14: Calculate illumination components at each scale according to the multi-scale fuzzy map of the construction site to obtain shooting illumination component data;

[0015] Step S15: performing contrast enhancement weight coefficient processing according to the captured illumination component data to generate a contrast enhancement weight coefficient;

[0016] Step S16: Perform multi-scale weight fusion on the multi-scale fuzzy image of the construction site through the contrast enhancement weight coefficient, and perform adaptive image contrast enhancement to obtain clear bridge construction image data.

[0017] Preferably, step S2 comprises the following steps:

[0018] Step S21: performing image multi-scale gradient analysis on the clear bridge construction image data to obtain image multi-scale gradient data;

[0019] Step S22: using the image multi-scale gradient data to perform non-maximum suppression on the clear bridge construction image data, and perform image edge extraction to generate image edge point response data;

[0020] Step S23: Recognize the main direction of steel bar binding according to the image edge point response data to obtain the main direction data of steel bar binding; perform virtual rectangular grid processing according to the main direction data of steel bar binding to generate virtual rectangular grid data;

[0021] Step S24: performing steel bar skeleton topology network processing on the image edge point response data through virtual rectangular grid data and steel bar binding main direction data to obtain initial steel bar skeleton topology data;

[0022] Step S25: identifying the center line of the steel bar according to the initial steel bar skeleton topology data to obtain the center line data of the steel bar;

[0023] Step S26: using the steel bar centerline data to perform skeleton endpoint detection on the initial steel bar skeleton topology data to obtain binding skeleton endpoint marking data;

[0024] Step S27: Optimize the initial steel bar skeleton topology data by using the binding skeleton endpoint marking data to generate optimized bridge steel bar binding skeleton data.

[0025] Preferably, step S23 includes the following steps:

[0026] Step S231: performing edge connection according to the image edge point response data to obtain closed edge contour data;

[0027] Step S232: performing gap area recognition on the image multi-scale gradient data by closing the edge contour data to obtain the steel bar binding gap area;

[0028] Step S233: extracting the outline pixel coordinates of the steel bar binding gap area to obtain outline pixel coordinate data;

[0029] Step S234: Calculate the average values ​​of horizontal and vertical coordinates according to the contour pixel coordinate data, and perform regional centroid coordinate processing to generate gap region centroid coordinate data;

[0030] Step S235: identifying the main direction of steel bar binding according to the center of gravity coordinate data of the gap area to obtain the main direction data of steel bar binding;

[0031] Step S236: Perform virtual rectangular grid processing on the steel bar binding gap area according to the steel bar binding direction data to generate virtual rectangular grid data.

[0032] Preferably, step S24 includes the following steps:

[0033] Step S241: performing edge point confidence calculation according to the image edge point response data to generate image edge point confidence data;

[0034] Step S242: performing image directional weighted neighborhood growth on the clear bridge construction image data based on the main direction data of the steel bar binding and using the image edge point confidence data to generate a directional local weighted area map;

[0035] Step S243: optimizing the edge steel bar line pixels of the directional local weighted area map, and identifying the cross nodes according to the virtual rectangular grid data to obtain the steel bar binding node data;

[0036] Step S244: Connect adjacent nodes according to the steel bar binding node data to obtain steel bar binding connection edge data;

[0037] Step S245: Mark branch points according to the steel bar binding node data and the steel bar binding connection edge data, and perform steel bar skeleton topology processing to generate initial steel bar skeleton topology data.

[0038] Preferably, step S27 includes the following steps:

[0039] Step S271: performing endpoint domain connection node analysis according to the binding skeleton endpoint mark data to obtain skeleton endpoint domain node data;

[0040] Step S272: evaluating the endpoint direction consistency of the binding skeleton endpoint mark data through the skeleton endpoint domain node data to obtain an endpoint direction consistency value;

[0041] Step S273: evaluating the density connection distance threshold of the bound skeleton endpoint mark data through the skeleton endpoint domain node data to obtain a connection distance threshold;

[0042] Step S274: performing a skeleton connectivity impact analysis on the bound skeleton endpoint mark data through the skeleton endpoint domain node data to obtain a connectivity impact value;

[0043] Step S275: weighted false endpoint scoring is performed according to the endpoint direction consistency value, the connection distance threshold, and the connectivity influence value, and endpoint elimination processing is performed on the initial steel skeleton topology data to obtain endpoint optimized steel skeleton data;

[0044] Step S276: performing breakpoint detection on the endpoint optimized steel bar skeleton data, and searching for the optimal breakpoint topological connection path to obtain the optimal breakpoint topological connection path data;

[0045] Step S277: Perform break connection optimization on the endpoint optimized steel bar skeleton data through the optimal breakpoint topology connection path data to generate optimized bridge steel bar binding skeleton data.

[0046] Preferably, step S3 comprises the following steps:

[0047] Step S31: calibrate corner point detection on the optimized bridge reinforcement binding skeleton data to obtain binding skeleton calibration point data;

[0048] Step S32: Based on the preset calibration plate geometry data, calibration point matching is performed on the optimized bridge reinforcement binding skeleton data through the binding skeleton calibration point data, and bridge curved surface effect compensation processing is performed to generate a bridge curved surface transformation correction coefficient;

[0049] Step S33: extracting skeleton line segments according to the optimized bridge reinforcement binding skeleton data to generate bridge reinforcement skeleton line segment data;

[0050] Step S34: performing bridge structure steel bar vanishing point analysis according to the bridge steel bar skeleton line segment data to generate bridge structure skeleton vanishing point fitting data; wherein step S34 is specifically as follows:

[0051] Step S341: classifying the bridge steel skeleton line segment data into bridge structure to obtain bridge structure classification line segment data;

[0052] Step S342: classifying the steel bar binding line segment direction of the bridge structure classification line segment data to obtain steel bar binding direction type data;

[0053] Step S343: extracting parallel line groups according to the classified line segment data of the bridge structure to obtain parallel line group data of steel bars;

[0054] Step S344: calculating the straight line length of the parallel line group data of the steel bars to obtain the straight line length data of the parallel line group of the steel bars;

[0055] Step S345: performing intersection analysis on the direction type data of the steel bar binding by using the straight line length data of the steel bar parallel line group, and performing vanishing point distribution quantity fitting to obtain vanishing point fitting data of the bridge structure skeleton;

[0056] Step S35: performing image perspective transformation processing according to the vanishing point fitting data of the bridge structure skeleton, and performing transformation matrix correction using the bridge surface transformation correction coefficient to generate an image correction transformation matrix;

[0057] Step S36: Perform image pixel scale mapping on the optimized bridge reinforcement binding skeleton data through an image correction transformation matrix to obtain actual skeleton scale mapping data.

[0058] Preferably, step S4 comprises the following steps:

[0059] Step S41: performing binding node mapping on the clear bridge construction image data by optimizing the bridge reinforcement binding skeleton data to generate binding node pixel mapping data;

[0060] Step S42: performing adjacent node coordinate feature vector processing according to the binding node pixel mapping data to generate binding node feature vector data;

[0061] Step S43: performing pixel polar coordinate conversion on the binding node feature vector data to obtain the binding node polar coordinate feature data;

[0062] Step S44: Optimizing the bridge reinforcement tying skeleton data to perform tying node degree statistics, and obtaining tying node connection degree characteristic data;

[0063] Step S45: performing node geometric feature mapping according to the polar coordinate feature data of the binding nodes and the connectivity feature data of the binding nodes, and calculating the pixel distance between the nodes to generate the pixel distance data of the binding nodes;

[0064] Step S46: Use the binding node pixel distance data to match the optimized bridge reinforcement binding skeleton data with adjacent skeleton segments, and use the skeleton actual scale mapping data to perform distance scale conversion to generate an estimated value of the reinforcement binding spacing.

[0065] Preferably, step S5 comprises the following steps:

[0066] Step S51: Obtaining bridge reinforcement spacing design specification data;

[0067] Step S52: using the bridge reinforcement spacing design specification data to calculate the spacing difference of the reinforcement binding spacing estimation value, and generate bridge binding spacing deviation value data;

[0068] Step S53: mapping the tolerance range according to the bridge lashing spacing deviation value data, and making a compliance determination to obtain lashing deviation state data;

[0069] Step S54: marking the unqualified spacing position of the optimized bridge reinforcement binding skeleton data based on the binding deviation state data to generate unqualified reinforcement binding position data;

[0070] Step S55: scoring the construction quality according to the unqualified steel bar binding position data to obtain bridge construction quality scoring data;

[0071] Step S56: Generate a project quality inspection report based on the unqualified steel bar binding position data and the bridge construction quality score data to obtain a bridge steel bar binding project quality inspection report.

[0072] Preferably, the present invention further provides a bridge engineering construction quality detection system, which executes the bridge engineering construction quality detection method as described above, and the bridge engineering construction quality detection system comprises:

[0073] The bridge construction image detection module is used to use a multi-spectral camera array to periodically capture detection images of the steel bar binding site during the bridge construction period to generate original bridge construction site image data; perform image contrast enhancement on the original bridge construction site image data to obtain clear bridge construction image data;

[0074] The bridge reinforcement skeleton extraction module is used to identify the main direction of reinforcement binding for clear bridge construction image data to obtain the main direction data of reinforcement binding; perform reinforcement skeleton topology network processing based on the main direction data of reinforcement binding to obtain initial reinforcement skeleton topology data; perform reinforcement centerline identification based on the initial reinforcement skeleton topology data, and perform skeleton endpoint optimization to generate optimized bridge reinforcement binding skeleton data;

[0075] The skeleton scale mapping module is used to perform bridge surface effect compensation processing on the optimized bridge reinforcement binding skeleton data to generate bridge surface transformation correction coefficients; the bridge surface transformation correction coefficients are used to perform image pixel scale mapping on the optimized bridge reinforcement binding skeleton data to obtain skeleton actual scale mapping data;

[0076] The binding node positioning module is used to map the binding nodes of the clear bridge construction image data by optimizing the bridge reinforcement binding skeleton data to generate binding node pixel mapping data; and to map the binding node pixel mapping data using the skeleton actual scale mapping data;

[0077] The project quality assessment module is used to obtain the bridge reinforcement spacing design specification data; use the bridge reinforcement spacing design specification data to calculate the spacing difference of the reinforcement binding spacing estimate, and perform construction quality scoring to obtain the bridge construction quality scoring data; generate a project quality inspection report based on the bridge construction quality scoring data to obtain a bridge reinforcement binding project quality inspection report.

[0078] The beneficial effect of the present application is that, by using a multi-spectral camera array and non-contact image acquisition, high-quality original image data can be obtained even under poor lighting conditions, effectively avoiding the tediousness of manual measurement one by one, and greatly improving the detection efficiency. At the same time, the method uses computer vision and image processing technology for batch processing, greatly shortens the detection time, speeds up the project progress, reduces the dependence on the experience of quality inspectors, reduces the cost of personnel training, and makes the detection results more objective and consistent. Especially in dense steel grid areas, manual detection is easily hindered, and the advantages of the automated detection of the present invention are more obvious, and its efficiency improvement far exceeds manual detection. The traditional manual measurement method is not only limited in accuracy by manual operation, but also in dense steel interlaced areas or hard-to-reach corners, it is easier to amplify the measurement error, and it is difficult to ensure the accuracy of the measurement. The present invention uses image contrast enhancement to clarify the steel edge, which is convenient for subsequent accurate identification, and accurately extracts the center line and structural features of the steel bar through the main direction identification of steel bar binding, steel bar skeleton topological network processing, steel bar center line identification and skeleton endpoint optimization, avoiding the interference caused by dense steel bar grid. At the same time, through the compensation processing of the bridge surface effect, the surface transformation correction coefficient is generated, which effectively eliminates the influence of the surface deformation on the measurement results, making the measurement results more accurate and reliable, and directly obtains the accurate estimation value of the reinforcement binding spacing through pixel scale mapping and distance scale conversion. This precise measurement method combined with geometric feature analysis can still maintain high accuracy in complex environments, far exceeding manual measurement, effectively avoiding quality defects caused by measurement errors and ensuring the safety performance of the bridge structure. Through large-scale image acquisition, the entire bridge structure or designated area is fully inspected to avoid the omission of quality hazards, and by comparing with the bridge reinforcement spacing design specification data, the spacing difference is calculated, and compliance can be automatically determined and a quality assessment report can be output. This full coverage detection and compliance determination makes quality management more accurate and reliable, and effectively avoids structural safety hazards caused by omissions in random inspections. Therefore, a bridge engineering construction quality inspection method of the present invention realizes rapid, accurate and comprehensive inspection, compliance judgment and quality assessment of steel bar binding spacing by integrating multi-spectral image analysis, computer vision, image processing technology and geometric feature analysis. It not only greatly improves the efficiency and accuracy of engineering quality inspection and reduces labor costs, but also realizes full coverage inspection and intelligent management, effectively avoiding the risk of structural performance degradation due to irregular steel bar binding. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1 It is a schematic diagram of the steps of the bridge engineering construction quality detection method of the present invention;

[0080] Figure 2 for Figure 1 Detailed implementation steps of step S4 in FIG.

[0081] Figure 3 for Figure 1 Detailed implementation steps of step S5;

[0082] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0083] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0084] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0085] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0086] To achieve this, please refer to Figures 1 to 3 The present invention provides a bridge engineering construction quality detection method, comprising the following steps:

[0087] Step S1: using a multispectral camera array to periodically capture images of the steel bar binding site during the bridge construction period to generate original bridge construction site image data; performing image contrast enhancement on the original bridge construction site image data to obtain clear bridge construction image data;

[0088] Step S2: Recognize the main direction of steel bar binding on the clear bridge construction image data to obtain the main direction data of steel bar binding; perform steel bar skeleton topology network processing according to the main direction data of steel bar binding to obtain initial steel bar skeleton topology data; identify the steel bar centerline according to the initial steel bar skeleton topology data, and optimize the skeleton endpoints to generate optimized bridge steel bar binding skeleton data;

[0089] Step S3: performing bridge surface effect compensation processing on the optimized bridge reinforcement binding skeleton data to generate a bridge surface transformation correction coefficient; performing image pixel scale mapping on the optimized bridge reinforcement binding skeleton data using the bridge surface transformation correction coefficient to obtain skeleton actual scale mapping data;

[0090] Step S4: mapping the clear bridge construction image data to binding nodes by optimizing the bridge reinforcement binding skeleton data to generate binding node pixel mapping data; performing distance scale conversion on the binding node pixel mapping data using the skeleton actual scale scale mapping data to generate a reinforcement binding spacing estimation value;

[0091] Step S5: Obtain bridge reinforcement spacing design specification data; use the bridge reinforcement spacing design specification data to calculate the spacing difference of the reinforcement binding spacing estimate, and perform a construction quality score to obtain bridge construction quality score data; generate a project quality inspection report based on the bridge construction quality score data to obtain a bridge reinforcement binding project quality inspection report.

[0092] In an embodiment of the present invention, the bridge engineering construction quality detection method comprises the following steps:

[0093] Step S1: using a multispectral camera array to periodically capture images of the steel bar binding site during the bridge construction period to generate original bridge construction site image data; performing image contrast enhancement on the original bridge construction site image data to obtain clear bridge construction image data;

[0094] In the embodiment of the present invention, a multispectral camera array is set up at a fixed position above the bridge construction site, for example, on a scaffold 5 meters away from the bridge deck. The camera array includes 5 multispectral cameras in the visible light band and 2 near-infrared bands, covering the wavelength range of 400nm to 1000nm, and the resolution of each camera is 8000 6000 pixels, and ensure that the camera array field of view can cover the entire bridge reinforcement binding area. Set the camera to automatically shoot every 30 minutes to obtain periodic detection images of the reinforcement binding site during the construction period. The original bridge construction site image data generated by shooting is stored in RAW format and transmitted to the data processing server. Subsequently, the original image data is preprocessed. The preprocessing process includes lens distortion correction and radiation correction to eliminate the influence of camera lens and lighting conditions on the image. Then, the contrast enhancement algorithm based on histogram equalization is used to process the corrected image. The specific operation is as follows: First, the frequency of occurrence of each pixel value in the image is counted to construct an image histogram. Then, the histogram is nonlinearly stretched to make the pixel value distribution more uniform, thereby enhancing the image contrast. Finally, the enhanced image is saved in TIFF format to obtain clear bridge construction image data.

[0095] Step S2: Recognize the main direction of steel bar binding on the clear bridge construction image data to obtain the main direction data of steel bar binding; perform steel bar skeleton topology network processing according to the main direction data of steel bar binding to obtain initial steel bar skeleton topology data; identify the steel bar centerline according to the initial steel bar skeleton topology data, and optimize the skeleton endpoints to generate optimized bridge steel bar binding skeleton data;

[0096] In an embodiment of the present invention, the clear bridge construction image data is grayed, and then the Sobel operator is used to perform edge detection to extract the steel bar edge information. The straight line segment is detected by Hough transform, and the direction of the main steel bars in the image is counted according to the direction and position information of the straight line segment, and the main direction data of the steel bar binding is identified, such as the main direction angle of 0 degrees and 90 degrees. Then, the identified steel bar edge information is converted into a graph structure, in which the nodes represent the steel bar intersections and the edges represent the steel bar segments. According to the main direction data of the steel bar binding, the graph structure is topologically processed, such as removing the edges with an angle greater than 15 degrees with the main direction, connecting the nodes with a distance less than 10 pixels, etc., to obtain the initial steel bar skeleton topological data. Then, the center line of the steel bar is extracted using a skeleton refinement algorithm, such as the Zhang-Suen refinement algorithm. Due to the influence of image noise and complex binding conditions, the initial skeleton has defects such as breakage or burrs. Therefore, it is necessary to optimize the skeleton endpoints. For example, for two endpoints with a distance less than 20 pixels, they are connected; for short branches with a length less than 15 pixels, they are removed. After the above processing, optimized bridge reinforcement binding skeleton data is generated, which accurately reflects the spatial layout and connection relationship of the reinforcement.

[0097] Step S3: performing bridge surface effect compensation processing on the optimized bridge reinforcement binding skeleton data to generate a bridge surface transformation correction coefficient; performing image pixel scale mapping on the optimized bridge reinforcement binding skeleton data using the bridge surface transformation correction coefficient to obtain skeleton actual scale mapping data;

[0098] In an embodiment of the present invention, a bridge surface model is established according to a bridge design drawing or three-dimensional scanning data. Then, the optimized steel skeleton data is projected onto the bridge surface model, and the actual position of each skeleton point on the surface is calculated. By comparing the coordinate difference of the skeleton points before and after the projection, the bridge surface transformation correction coefficient is calculated. For example, the scaling coefficient of each skeleton point in the three directions of x, y, and z can be calculated. Then, the bridge surface transformation correction coefficient is used to perform image pixel scale mapping on the optimized bridge steel bar binding skeleton data. The specific operation is: multiply the pixel coordinates of each skeleton point by the corresponding scaling coefficient to obtain the coordinates of the skeleton point in the actual three-dimensional space. Then, these three-dimensional coordinates are re-projected onto the image plane to obtain the actual scale mapping data of the skeleton. For example, assuming that the pixel coordinates of a skeleton point are (100, 200), and the scaling coefficients in the x and y directions are 1.1 and 1.2 respectively, the mapped pixel coordinates are (110, 240). In this way, the skeleton data with the influence of the bridge surface effect is eliminated is obtained. The pixel coordinates of each skeleton point are multiplied by the bridge surface transformation correction coefficient to obtain the three-dimensional coordinates of the point in the actual space. Then, the three-dimensional coordinates are projected onto the horizontal plane to obtain the two-dimensional coordinates of the point on the horizontal plane. Finally, the two-dimensional coordinates are converted into actual physical lengths to obtain the actual scale mapping data of the skeleton.

[0099] Step S4: mapping the clear bridge construction image data to binding nodes by optimizing the bridge reinforcement binding skeleton data to generate binding node pixel mapping data; performing distance scale conversion on the binding node pixel mapping data using the skeleton actual scale scale mapping data to generate a reinforcement binding spacing estimation value;

[0100] In an embodiment of the present invention, based on the optimized bridge reinforcement binding skeleton data, the intersection points of the reinforcement, i.e., the binding nodes, are identified. These binding nodes are mapped to clear bridge construction image data to generate binding node pixel mapping data. For example, the coordinates of each binding node are marked on the image. Then, the binding node pixel mapping data is converted to a distance scale using the actual scale mapping data of the skeleton. The specific operation is: calculate the pixel distance between adjacent binding nodes on the image, and then multiply it by the corresponding scale factor to obtain the estimated value of the reinforcement binding spacing. For example, assuming that the pixel distance between two adjacent binding nodes is 50 and the scale factor is 0.5 cm / pixel, the estimated value of the reinforcement binding spacing is 25 cm.

[0101] Step S5: Obtain bridge reinforcement spacing design specification data; use the bridge reinforcement spacing design specification data to calculate the spacing difference of the reinforcement binding spacing estimate, and perform a construction quality score to obtain bridge construction quality score data; generate a project quality inspection report based on the bridge construction quality score data to obtain a bridge reinforcement binding project quality inspection report.

[0102] In an embodiment of the present invention, the bridge reinforcement spacing design specification data is obtained from the bridge design drawings, for example, the longitudinal reinforcement spacing is 20cm, and the transverse reinforcement spacing is 15cm. The bridge reinforcement spacing design specification data is used to calculate the spacing difference of the estimated value of the reinforcement binding spacing. For example, the absolute difference or relative difference between the estimated value and the design value is calculated. According to the calculated spacing difference, combined with the pre-set scoring standard, the construction quality is scored to obtain the bridge construction quality scoring data. The scoring standard can be formulated according to the engineering specifications and project requirements. For example, the spacing difference is scored as 100 points within ±2cm, and points are deducted according to the size of the difference if it exceeds ±2cm. Finally, a project quality inspection report is generated according to the bridge construction quality scoring data. The report content includes information such as the inspection date, the inspection location, the reinforcement spacing design value, the measured value, the spacing difference, the scoring results, and the unqualified areas are marked and explained to form a bridge reinforcement binding engineering quality inspection report.

[0103] Step S1 includes the following steps:

[0104] Step S11: using a multispectral camera array to periodically capture images of the steel bar binding site during the bridge construction period to generate original bridge construction site image data;

[0105] Step S12: performing image noise evaluation based on the original bridge construction site image data to obtain shooting noise level data;

[0106] Step S13: performing multi-scale Gaussian blurring on the original bridge construction site image data by shooting noise level data to generate a multi-scale blur map of the construction site;

[0107] Step S14: Calculate illumination components at each scale according to the multi-scale fuzzy map of the construction site to obtain shooting illumination component data;

[0108] Step S15: performing contrast enhancement weight coefficient processing according to the captured illumination component data to generate a contrast enhancement weight coefficient;

[0109] Step S16: Perform multi-scale weight fusion on the multi-scale fuzzy image of the construction site through the contrast enhancement weight coefficient, and perform adaptive image contrast enhancement to obtain clear bridge construction image data.

[0110] In the embodiment of the present invention, a multispectral camera array is set up at a fixed position above the bridge construction site, for example, on a scaffold 5 meters away from the bridge deck. The camera array includes 5 multispectral cameras in the visible light band and 2 near-infrared bands, covering the wavelength range of 400nm to 1000nm, and the resolution of each camera is 8000 6000 pixels. Ensure that the camera array field of view can cover the entire bridge reinforcement binding area, and perform camera calibration to obtain camera intrinsic and extrinsic parameters. Set the camera to automatically shoot every 30 minutes to obtain periodic detection images of the reinforcement binding site during the construction period. Convert the original bridge construction site image data into a grayscale image. Select a 100x100 pixel flat area in the image and calculate the standard deviation of the pixel values ​​in the area. The standard deviation value is the shooting noise level data, which represents the noise intensity of the image. For example, if the calculated standard deviation is 5, it means that the noise level of the image is 5. According to the shooting noise level data obtained in step S12, for example, the noise level is 5, three scales are selected for Gaussian blur, namely small scale, medium scale and large scale. The standard deviation σ1 of the small scale is set to half of the noise level, that is, 2.5; the standard deviation σ2 of the medium scale is set to the noise level, that is, 5; the standard deviation σ3 of the large scale is set to twice the noise level, that is, 10. Use these three standard deviations to perform Gaussian blur on the original bridge construction site image data respectively to generate three multi-scale blurred images of the construction site. For example, using the GaussianBlur function in the OpenCV library, setting different kernelsize and sigma values, blurring the original image respectively, and obtaining blurred images of three scales. Convert the blurred images of each scale into grayscale images. Then, for each pixel, calculate the mean grayscale value of the small-scale, medium-scale and large-scale blurred images. The mean is the shooting illumination component data of the pixel. For example, if the grayscale values ​​of a certain pixel on the three-scale blurred images are 100, 120 and 150 respectively, the shooting illumination component data of the pixel is (100+120+150) / 3=123.33. Store the shooting illumination component data of all pixels as a new image, namely, the shooting illumination component data map. Calculate the global average brightness value of the shooting illumination component data map, for example, the average brightness value is 128. Then, for each pixel, calculate the difference between its shooting illumination component data and the global average brightness value. Normalize the difference and map it to between 0 and 1 to obtain the contrast enhancement weight coefficient of the pixel. For example, if the shooting illumination component data of a certain pixel is 150 and the global average brightness value is 128, then the difference value of the pixel is 22, and the contrast enhancement weight coefficient after normalization is 0.17. For each pixel, the grayscale values ​​of the blurred images at three scales are multiplied by the corresponding contrast enhancement weight coefficients, and then the three results are added to obtain the fused grayscale value of the pixel. Finally, the fused image is subjected to adaptive image contrast enhancement, for example, using the Contrast-Limited Adaptive Histogram Equalization (CLAHE) algorithm to enhance the local contrast of the image, thereby obtaining clear bridge construction image data.

[0111] Step S2 includes the following steps:

[0112] Step S21: performing image multi-scale gradient analysis on the clear bridge construction image data to obtain image multi-scale gradient data;

[0113] Step S22: using the image multi-scale gradient data to perform non-maximum suppression on the clear bridge construction image data, and perform image edge extraction to generate image edge point response data;

[0114] Step S23: Recognize the main direction of steel bar binding according to the image edge point response data to obtain the main direction data of steel bar binding; perform virtual rectangular grid processing according to the main direction data of steel bar binding to generate virtual rectangular grid data;

[0115] Step S24: performing steel bar skeleton topology network processing on the image edge point response data through virtual rectangular grid data and steel bar binding main direction data to obtain initial steel bar skeleton topology data;

[0116] Step S25: identifying the center line of the steel bar according to the initial steel bar skeleton topology data to obtain the center line data of the steel bar;

[0117] Step S26: using the steel bar centerline data to perform skeleton endpoint detection on the initial steel bar skeleton topology data to obtain binding skeleton endpoint marking data;

[0118] Step S27: Optimize the initial steel bar skeleton topology data by using the binding skeleton endpoint marking data to generate optimized bridge steel bar binding skeleton data.

[0119] In the embodiment of the present invention, a multi-scale gradient analysis is performed on the clear bridge construction image data. First, the image is converted into a grayscale image. Then, the Sobel operator is used to calculate the gradient of the image in the x direction and the y direction respectively. Three scales are selected for gradient calculation, for example, 3 3,5 5, and 7 7's Sobel operator. Calculate the gradient map in the x and y directions at each scale respectively, and calculate the gradient magnitude and direction of each pixel. The formula for calculating the gradient magnitude is: magnitude = sqrt(Gx^2+Gy^2), where Gx and Gy are the gradient values ​​in the x and y directions respectively. The formula for calculating the gradient direction is: direction = arctan(Gy / Gx). For the gradient data at each scale, for each pixel, compare its gradient magnitude with the gradient magnitude of the adjacent pixels in its gradient direction. If the gradient magnitude of the pixel is a local maximum, the gradient magnitude of the pixel is retained, otherwise the gradient magnitude is set to 0. This is called non-maximum suppression, which can refine the edges. Then, the double threshold method is used for edge extraction. Set a high threshold and a low threshold, for example, the high threshold is 100 and the low threshold is 50. If the gradient amplitude of a pixel is greater than a high threshold, the pixel is considered to be an edge point; if the gradient amplitude of a pixel is less than a low threshold, the pixel is considered not to be an edge point; if the gradient amplitude of a pixel is between a high threshold and a low threshold, the pixel is considered to be an edge point only when it is adjacent to an edge point. The extracted edge points and their gradient amplitudes are stored as image edge point response data. The main direction of steel bar binding is identified based on the image edge point response data. The straight line segments in the image are detected using Hough transform. The directions of all straight line segments are counted to find the two most frequently occurring directions perpendicular to each other as the main direction data of steel bar binding. For example, the detected main direction angles are 0 degrees and 90 degrees. Then, a virtual rectangular grid is generated based on the main direction data of steel bar binding and the image size. For example, if the image size is 1000x1000 pixels, the main directions are 0 degrees and 90 degrees, and the grid spacing is 50 pixels, a 20x20 virtual rectangular grid is generated, and the size of each grid is 50 50 pixels. The generated grid data is stored as virtual rectangular grid data. Each edge point in the image edge point response data is assigned to the corresponding virtual rectangular grid unit. For example, an edge point coordinate is (230,470) and the grid size is 50 50 pixels, then the edge point belongs to the grid cell of the 4th row and the 9th column (200-250, 450-500). Then, in each grid cell, the edge points are grouped according to the main direction data of the steel bar binding (for example, 0 degrees and 90 degrees). The edge points whose directions are less than 15 degrees from the main direction are grouped into the same group. In each group, the distance between the edge points is calculated. If the distance between two edge points is less than a preset threshold, for example, 10 pixels, the two edge points are connected to form a line segment. The weight of the line segment can be set to the average of the response values ​​of the two edge points. Traverse all grid cells and repeat the above operation to connect all edge points that meet the conditions. Finally, the line segments with the same direction and close distance in adjacent grid cells are connected to form a connected graph structure, that is, the initial steel bar skeleton topology data. The initial steel bar skeleton is refined using a skeleton refinement algorithm, such as the Zhang-Suen refinement algorithm. The Zhang-Suen algorithm is an iterative algorithm that peels off the skeleton edge pixels layer by layer and finally obtains a center line with a single pixel width. During each iteration, the algorithm determines whether a pixel meets the deletion condition, for example, the pixel must be a point on the skeleton, and deleting it will not change the connectivity of the skeleton. Repeat the iteration until the skeleton no longer changes. The refined skeleton is the steel bar centerline data, which more accurately expresses the direction and position information of the steel bar. Traverse each node in the initial steel bar skeleton topology data. For each node, count the number of edges connected to it. If a node is only connected to one edge, it is considered to be the endpoint of the skeleton, and the coordinates of the node and the information of the connected edge are marked in the binding skeleton endpoint mark data. For example, if the coordinates of an endpoint node are (300,600) and the connected edge is (300,600) to (350,600), the information of the endpoint is recorded in the binding skeleton endpoint mark data as {(300,600), (350,600)}. According to the binding skeleton endpoint mark data, calculate the distance between all endpoints. If the distance between two endpoints is less than a preset threshold, such as 20 pixels, and the directions of the edges connected by the two endpoints are roughly the same (for example, the angle is less than 30 degrees), the two endpoints are connected to form a new edge. This can effectively connect skeleton breaks caused by image noise or occlusion. Secondly, identify and remove short branches. Traverse all edges in the initial steel skeleton topology data. If the length of an edge is less than a preset threshold, such as 15 pixels, and only one of the two nodes connected by the edge is an endpoint, the edge is considered to be a short branch and is removed from the skeleton. This can remove burrs caused by image noise or other factors. After the two steps of endpoint connection and short branch removal, the optimized bridge steel bar binding skeleton data is obtained.

[0120] Step S23 includes the following steps:

[0121] Step S231: performing edge connection according to the image edge point response data to obtain closed edge contour data;

[0122] Step S232: performing gap area recognition on the image multi-scale gradient data by closing the edge contour data to obtain the steel bar binding gap area;

[0123] Step S233: extracting the outline pixel coordinates of the steel bar binding gap area to obtain outline pixel coordinate data;

[0124] Step S234: Calculate the average values ​​of horizontal and vertical coordinates according to the contour pixel coordinate data, and perform regional centroid coordinate processing to generate gap region centroid coordinate data;

[0125] Step S235: identifying the main direction of steel bar binding according to the center of gravity coordinate data of the gap area to obtain the main direction data of steel bar binding;

[0126] Step S236: Perform virtual rectangular grid processing on the steel bar binding gap area according to the steel bar binding direction data to generate virtual rectangular grid data.

[0127] In the embodiment of the present invention, starting from an edge point that has not been visited, other edge points in its 8-neighborhood are searched. If an adjacent edge point is found, the two points are connected and the newly found edge point is marked as visited. The 8-neighborhood of the new edge point is searched continuously until no new adjacent edge point is found. At this point, an edge chain is obtained. The above process is repeated until all edge points have been visited. In order to obtain a closed edge contour, the edge chain needs to be closed. If the distance between the starting point and the end point of an edge chain is less than a preset threshold, such as 10 pixels, and there is a path between them that can be formed by connecting adjacent edge points, the starting point and the end point are connected to form a closed edge contour. According to the closed edge contour data, the image is segmented into different regions. Each region is either an internal region surrounded by an edge contour or a background region outside the edge contour. Then, the multi-scale gradient data of the image is analyzed. The average value of the multi-scale gradient amplitude in each region is calculated. Since the edge of the steel bar usually has a higher gradient value, and the gradient value of the gap area is lower, the steel bar area and the gap area can be distinguished according to the average value of the gradient amplitude. A gradient threshold is set, for example, the gradient threshold is half of the average value of the multi-scale gradient data of the image. If the average value of the multi-scale gradient amplitude in a region is lower than the gradient threshold, the region is considered to be a steel bar binding gap region. For each region identified as a gap region, the pixel coordinates of its boundary contour are extracted. A contour tracing algorithm, such as the Moore-Neighbor tracing algorithm, can be used to extract the contour pixel coordinates. The algorithm starts from a starting point on a contour and searches for the next contour point in its neighborhood according to a certain rule (for example, in a clockwise direction) until it returns to the starting point. The extracted contour pixel coordinates are stored as contour pixel coordinate data. For each steel bar binding gap region, its contour pixel coordinate data contains the coordinates (x, y) of a series of pixel points. First, the average value of the horizontal coordinate x and the average value of the vertical coordinate y of all contour pixels are calculated respectively. These two average values ​​can approximately represent the center position of the gap region. Then, in order to more accurately determine the centroid coordinates of the gap region, a weighted average method based on pixel area can be used. Specifically, the gap region is regarded as a polygon whose vertices are determined by the contour pixel coordinates. Calculate the area of ​​the polygon and the weighted average of the distance from each pixel to a fixed reference point of the polygon (for example, the upper left corner of the image). The final weighted average is the centroid coordinate of the gap area. Construct a two-dimensional histogram to count the distribution of the centroid coordinates of the gap area in different directions. Set the horizontal and vertical axes of the histogram to angle and distance respectively. For each pair of centroid coordinates of the gap area, calculate the distance between them and the direction of the connecting line. Accumulate the distance and direction to the corresponding cell in the histogram.For example, the coordinates of the two centroids are (200, 300) and (400, 700), the distance between them is 500, and the direction of the connection is 53.13 degrees, then 500 is added to the cells in the histogram with an angle of 53.13 degrees and a distance of 500. After the histogram is constructed, the two mutually perpendicular directions with the largest values ​​in the histogram are found, which are the main directions of the steel bar binding. According to the main direction data of the steel bar binding obtained in step S235 (for example, 0 degrees and 90 degrees), and the size of the image, a virtual rectangular grid is constructed. The direction of the grid line is parallel to the main direction of the steel bar binding. The size of the grid can be adjusted according to the actual situation. For example, the grid size can be set to 1 / 2 or 1 / 3 of the average spacing of the steel bars. For example, if the image size is 1000x800 pixels, the main directions of the steel bar binding are 0 degrees and 90 degrees, and the grid size is set to 50x50 pixels, a 20x16 virtual rectangular grid can be generated. The generated grid information, including the coordinates of the grid lines, the size of the grid, etc., is stored as virtual rectangular grid data.

[0128] Step S24 includes the following steps:

[0129] Step S241: performing edge point confidence calculation according to the image edge point response data to generate image edge point confidence data;

[0130] Step S242: performing image directional weighted neighborhood growth on the clear bridge construction image data based on the main direction data of the steel bar binding and using the image edge point confidence data to generate a directional local weighted area map;

[0131] Step S243: optimizing the edge steel bar line pixels of the directional local weighted area map, and identifying the cross nodes according to the virtual rectangular grid data to obtain the steel bar binding node data;

[0132] Step S244: Connect adjacent nodes according to the steel bar binding node data to obtain steel bar binding connection edge data;

[0133] Step S245: Mark branch points according to the steel bar binding node data and the steel bar binding connection edge data, and perform steel bar skeleton topology processing to generate initial steel bar skeleton topology data.

[0134] In an embodiment of the present invention, a method based on gradient amplitude and local direction consistency is used to calculate confidence. First, the gradient amplitude of each edge point is normalized to between 0 and 1. The larger the gradient amplitude, the higher the confidence. Then, for each edge point, the angle between the gradient direction of the edge point in its 8-neighborhood and the gradient direction of the edge point is calculated. The smaller the angle, the higher the direction consistency. The angle is converted to a value between 0 and 1, with a value of 1 when the angle is 0 degrees and a value of 0 when the angle is 90 degrees. Finally, the normalized gradient amplitude and direction consistency are weighted averaged to obtain the confidence of the edge point. The growth direction is determined according to the main direction data of the steel bar binding (for example, 0 degrees and 90 degrees). Then, the edge point with the highest confidence is selected as the seed point from the image edge point confidence data. The seed point is subjected to neighborhood growth. During the growth process, neighborhood pixels in the direction consistent with the main direction of the steel bar binding are given priority. For example, if the main direction of the steel bar binding is 0 degrees, the pixels on the left and right sides of the seed point are given priority. When adding neighborhood pixels to the growth region, the edge point confidence data is considered. The higher the confidence of a pixel, the greater its weight and the easier it is to be added to the growth region. For example, a weighted average method can be used to calculate the pixel value of the growth region. Repeat the above process until all edge points are processed or the growth region is no longer expanded. The resulting growth region is stored as a directional local weighted region map. The directional local weighted region map is refined, for example, using a skeleton refinement algorithm to extract steel bars with a single pixel width. Then, the steel bars are divided into different grid cells based on the virtual rectangular grid data. Within each grid cell, the intersection of the steel bar line is identified. For example, the intersection between each steel bar line and other steel bar lines can be calculated. The identified intersection coordinates are stored as steel bar binding node data. For each pair of steel bar binding nodes, the distance between them is calculated. If the distance between two nodes is less than a preset threshold, such as twice the average diameter of the steel bar, the two nodes are considered to be adjacent nodes and are connected to form an edge. The information of the connected edge, such as the starting node coordinates, the ending node coordinates, etc., is stored as steel bar binding connection edge data. Traverse all nodes in the steel bar binding node data. For each node, count the number of edges connected to it. If the number of edges connected to a node is greater than 2, the node is considered to be a branch point and marked as a branch point in the initial steel bar skeleton topology data. Then, based on the steel bar binding node data and the steel bar binding connection edge data, a graph structure is constructed, in which the nodes represent the steel bar binding nodes and the edges represent the steel bar binding connection edges. This graph structure is the initial steel bar skeleton topology data.

[0135] Step S27 includes the following steps:

[0136] Step S271: performing endpoint domain connection node analysis according to the binding skeleton endpoint mark data to obtain skeleton endpoint domain node data;

[0137] Step S272: evaluating the endpoint direction consistency of the binding skeleton endpoint mark data through the skeleton endpoint domain node data to obtain an endpoint direction consistency value;

[0138] Step S273: evaluating the density connection distance threshold of the bound skeleton endpoint mark data through the skeleton endpoint domain node data to obtain a connection distance threshold;

[0139] Step S274: performing a skeleton connectivity impact analysis on the bound skeleton endpoint mark data through the skeleton endpoint domain node data to obtain a connectivity impact value;

[0140] Step S275: weighted false endpoint scoring is performed according to the endpoint direction consistency value, the connection distance threshold, and the connectivity influence value, and endpoint elimination processing is performed on the initial steel skeleton topology data to obtain endpoint optimized steel skeleton data;

[0141] Step S276: performing breakpoint detection on the endpoint optimized steel bar skeleton data, and searching for the optimal breakpoint topological connection path to obtain the optimal breakpoint topological connection path data;

[0142] Step S277: Perform break connection optimization on the endpoint optimized steel bar skeleton data through the optimal breakpoint topology connection path data to generate optimized bridge steel bar binding skeleton data.

[0143] In an embodiment of the present invention, for each endpoint, a square area centered on the endpoint is defined, which is called the endpoint area. The side length of the area can be determined based on the average diameter or spacing of the steel bars. For example, the side length is set to 3 times the average diameter of the steel bars. In the endpoint area, search for nodes connected to the endpoint, that is, nodes with edges with the endpoint. Record the coordinates of these nodes, the distance from the endpoint, and the direction of the connecting edge as the domain node data of the endpoint. Store the domain node data of all endpoints as skeleton endpoint domain node data. For example, the coordinates of an endpoint are (100, 200), and there are two connecting nodes in its domain, with coordinates of (120, 210) and (80, 190), respectively. The distances from the endpoints are 22.36 and 22.36, respectively, and the directions of the connecting edges are 45 degrees and 135 degrees, respectively. The domain node data of the endpoint is {(120, 210, 22.36, 45), (80, 190, 22.36, 135)}. According to its domain node data, calculate the consistency of the direction of the connecting edge. First, calculate the average direction of all connecting edges. Then, calculate the angle between each connecting edge and the average direction. The average or maximum value of all angles is used as the endpoint direction consistency value. The smaller the consistency value, the more consistent the direction of the connecting edges. For example, an endpoint has three connecting edges with directions of 30 degrees, 40 degrees, and 50 degrees, respectively. The average direction is 40 degrees, the angles are 10 degrees, 0 degrees, and 10 degrees, respectively, and the average angle is 6.67 degrees. 6.67 degrees can be used as the endpoint direction consistency value. In areas with dense steel bars, the distance between endpoints is usually small, while in areas with sparse steel bars, the distance between endpoints is usually large. Therefore, it is necessary to adaptively adjust the connection distance threshold according to the density of steel bars. The density of steel bars can be evaluated based on the node data of the skeleton endpoint area. For example, the average distance of the connecting nodes in the endpoint area can be calculated. The larger the average distance, the sparser the steel bars, and the larger the connection distance threshold should be. The average distance can be multiplied by a coefficient as the connection distance threshold, for example, the coefficient value is 2 or 3. For example, if there are two connected nodes in the domain of an endpoint, and the distances from the endpoint are 20 and 30 respectively, then the average distance is 25, and the connection distance threshold can be set to 25. 2=50. For each endpoint, analyze its domain node data. If after removing the endpoint, the connected nodes in its domain can still be connected to other parts of the skeleton through other paths, the connectivity impact value is low. Conversely, if after removing the endpoint, the connected nodes in its domain are disconnected from other parts of the skeleton, the connectivity impact value is high. The connected component algorithm in graph theory can be used to evaluate the connectivity impact. For example, after removing an endpoint, calculate the number of connected components of the skeleton. The more the number of connected components increases, the greater the connectivity impact. The change in the number of connected components can be used as the connectivity impact value. According to the endpoint direction consistency value, connection distance threshold and connectivity impact value obtained in steps S272, S273 and S274, calculate the false endpoint score of the endpoint. The scoring formula can be adjusted according to the actual situation. For example, a linear weighted method can be used: Score=w1 Consistency value + w2 (distance-threshold)+w3 Connectivity influence value, where w1, w2 and w3 are weight coefficients, distance is the distance from the endpoint to its nearest neighbor node, and threshold is the connection distance threshold. Set a score threshold, such as 0.5. If the score of an endpoint is higher than the threshold, remove the endpoint from the initial reinforcement skeleton topology data. After removing the endpoint, the skeleton topology data needs to be updated, such as deleting the edge connected to the endpoint. The skeleton data after removing the false endpoint is stored as endpoint optimized reinforcement skeleton data. A graph traversal algorithm, such as depth first search or breadth first search, can be used to detect breakpoints. If it is impossible to reach another node from one node during the traversal process, it is considered that there is a breakpoint between the two nodes. For the detected breakpoints, search for the optimal topological connection path connecting the two breakpoints. The shortest path algorithm, such as Dijkstra algorithm or A* algorithm, can be used to search for the shortest path connecting two breakpoints in the skeleton graph. The found optimal path is stored as the optimal breakpoint topological connection path data. Add the nodes and edges on the path to the skeleton graph. If the nodes on the path already exist in the skeleton graph, there is no need to add the nodes, only the edges connecting the nodes need to be added. After adding the optimal path, the final optimized bridge reinforcement binding skeleton data is obtained.

[0144] Step S3 includes the following steps:

[0145] Step S31: calibrate corner point detection on the optimized bridge reinforcement binding skeleton data to obtain binding skeleton calibration point data;

[0146] Step S32: Based on the preset calibration plate geometry data, calibration point matching is performed on the optimized bridge reinforcement binding skeleton data through the binding skeleton calibration point data, and bridge curved surface effect compensation processing is performed to generate a bridge curved surface transformation correction coefficient;

[0147] Step S33: extracting skeleton line segments according to the optimized bridge reinforcement binding skeleton data to generate bridge reinforcement skeleton line segment data;

[0148] Step S34: performing bridge structure steel bar vanishing point analysis according to the bridge steel bar skeleton line segment data to generate bridge structure skeleton vanishing point fitting data; wherein step S34 is specifically as follows:

[0149] Step S341: classifying the bridge steel skeleton line segment data into bridge structure to obtain bridge structure classification line segment data;

[0150] Step S342: classifying the steel bar binding line segment direction of the bridge structure classification line segment data to obtain steel bar binding direction type data;

[0151] Step S343: extracting parallel line groups according to the classified line segment data of the bridge structure to obtain parallel line group data of steel bars;

[0152] Step S344: calculating the straight line length of the parallel line group data of the steel bars to obtain the straight line length data of the parallel line group of the steel bars;

[0153] Step S345: performing intersection analysis on the direction type data of the steel bar binding by using the straight line length data of the steel bar parallel line group, and performing vanishing point distribution quantity fitting to obtain vanishing point fitting data of the bridge structure skeleton;

[0154] Step S35: performing image perspective transformation processing according to the vanishing point fitting data of the bridge structure skeleton, and performing transformation matrix correction using the bridge surface transformation correction coefficient to generate an image correction transformation matrix;

[0155] Step S36: Perform image pixel scale mapping on the optimized bridge reinforcement binding skeleton data through an image correction transformation matrix to obtain actual skeleton scale mapping data.

[0156] In the embodiment of the present invention, a calibration plate with known geometric dimensions, such as a checkerboard calibration plate, is placed in advance at the bridge construction site. The dimensions and corner positions of the calibration plate are known. For example, the dimensions of the calibration plate are 30 cm 20cm, the size of each grid is 2cm 2cm. The calibration plate should be placed within the camera's field of view and have a certain relative position relationship with the bridge structure so that the bridge surface effect can be compensated later. Use a corner detection algorithm, such as the Harris corner detection algorithm or the Shi-Tomasi corner detection algorithm, to detect the corners of the calibration plate on the image corresponding to the optimized bridge reinforcement binding skeleton data. The Harris corner detection algorithm detects corners by calculating the gradient change of the local area of ​​the image, while the Shi-Tomasi corner detection algorithm selects the best corner based on eigenvalue analysis. The detected corner coordinates are recorded and matched with the known corner information of the calibration plate to obtain the binding skeleton calibration point data. The calibration corner points detected in step S31 are matched with the corner points in the preset calibration plate geometry data. A descriptor-based matching method, such as the SIFT or SURF algorithm, or a geometric relationship-based matching method, such as the PnP algorithm, can be used. After matching, the position and posture of the calibration plate in the image can be obtained. Then, the bridge surface transformation correction coefficient is calculated based on the position and posture of the calibration plate and the surface model of the bridge. The bridge surface model can be obtained from the bridge design drawings or 3D scanning data. The surface transformation correction coefficient can be expressed as a transformation matrix that maps points in the image coordinate system to points in the bridge surface coordinate system. For example, perspective transformation or nonlinear transformation can be used to describe the bridge surface effect. The bridge surface transformation correction coefficient can be obtained by calculating the transformation relationship between the calibrated corner points in the image coordinate system and the corresponding points in the bridge surface coordinate system. The optimized bridge reinforcement binding skeleton data is usually represented as a graph structure, where the nodes represent the intersections or endpoints of the reinforcement and the edges represent the reinforcement segments. Traverse all the edges in the skeleton graph and extract the coordinates of the start and end nodes of each edge. Represent each edge as a line segment and record its start and end point coordinates. Store the data of all line segments as bridge reinforcement skeleton line segment data. For example, if the start node coordinate of an edge is (100,200) and the end node coordinate is (300,400), the corresponding line segment data is {(100,200),(300,400)}. According to the bridge design drawings or other prior knowledge, the steel skeleton line segment data is divided into different categories, such as bridge deck, bridge pier, beam, etc. The division can be based on the position, direction, length and other features of the line segment. For example, the line segment can be divided into horizontal line segments, vertical line segments and oblique line segments according to the direction of the line segment. Then, according to the position of the line segment in the image, the horizontal line segment is further divided into bridge deck line segments and other line segments. According to the direction of the line segment, the line segment is divided into different direction types, such as horizontal direction, vertical direction, 45 degree direction, 135 degree direction, etc. The angle interval can be used to divide the direction type, for example, the line segment with a direction between -15 degrees and 15 degrees is classified as horizontal direction, and the line segment with a direction between 75 degrees and 105 degrees is classified as vertical direction.For each type of bridge structure, such as a bridge deck, extract the mutually parallel line segments to form a parallel line group. The angle between the line segment directions can be used to determine whether two line segments are parallel. If the angle between the two line segment directions is less than a preset threshold, such as 5 degrees, the two line segments are considered to be parallel. For each line segment in each parallel line group, calculate its length. The line segment length can be calculated using the coordinates of the two end points of the line segment. The calculated line segment length data is stored as the straight line length data of the steel bar parallel line group. For each parallel line group, extend the line segments therein and calculate their intersections. These intersections are the vanishing points. Due to the existence of errors, the line segments in the parallel line group will not intersect at the same point after extension. Therefore, it is necessary to fit the vanishing points. The least squares method or other fitting methods can be used to fit multiple intersections into one vanishing point. The number distribution of vanishing points in different direction types is counted and fitted to obtain the vanishing point fitting data of the bridge structure skeleton. According to the position of the vanishing point, the perspective transformation matrix of the image can be calculated. The perspective transformation matrix can convert the image into a front view and eliminate perspective distortion. Then, the perspective transformation matrix is ​​corrected using the bridge surface transformation correction coefficient obtained in step S32 to obtain the final image correction transformation matrix. The image correction transformation matrix is ​​applied to optimize the coordinates of each node in the bridge reinforcement binding skeleton data, and the pixel coordinates are converted into actual physical coordinates. Then, based on the known size of the calibration plate and its size in the image, the proportional relationship between the image pixel and the actual physical size, i.e., the scale factor, is calculated. The physical coordinates of each node are multiplied by the scale factor to obtain the actual scale mapping data of the skeleton.

[0157] As an example of the present invention, refer to Figure 2 As shown, Figure 1 Detailed implementation steps of step S4 in the embodiment are shown in the flowchart. In this embodiment, step S4 includes:

[0158] Step S41: performing binding node mapping on the clear bridge construction image data by optimizing the bridge reinforcement binding skeleton data to generate binding node pixel mapping data;

[0159] In the embodiment of the present invention, in the skeleton data, each node represents a steel bar binding point. For each node, a pixel coordinate of size n is constructed with the pixel coordinate of the node as the center. A rectangular window of n pixels (for example, n=5) is selected. Within the window, pixels with drastic changes in grayscale values ​​are searched. These pixels correspond to the actual positions of the steel bar binding points in the image. These pixels can be identified using edge detection operators (for example, Sobel operator) or corner detection operators (for example, Harris operator). In order to improve the mapping accuracy, a sub-pixel precision positioning method can be used. For example, the grayscale values ​​of the pixels around the node can be interpolated to obtain sub-pixel precision coordinates. The pixel coordinates of all nodes are stored as binding node pixel mapping data. For example, the coordinates of a node in the skeleton data are (1.2, 2.3). After being mapped to the image, its pixel coordinates are (100.5, 200.7). The information of the node stored in the binding node pixel mapping data is (100.5, 200.7).

[0160] Step S42: performing adjacent node coordinate feature vector processing according to the binding node pixel mapping data to generate binding node feature vector data;

[0161] In an embodiment of the present invention, for each binding node, its adjacent nodes in the skeleton graph are found. Adjacent nodes refer to nodes that are directly connected to the node. The pixel coordinate difference between the node and each adjacent node is calculated, and these differences are combined into a feature vector. The dimension of the feature vector is equal to the number of adjacent nodes multiplied by 2, where each dimension represents a coordinate difference (x difference or y difference). For example, the pixel coordinates of a node are (100, 200), and it has two adjacent nodes with pixel coordinates of (120, 210) and (80, 190), then the feature vector of the node is (20, 10, -20, -10). If the number of adjacent nodes of a node is different, the feature vector can be padded to a fixed length, for example, using zero padding or other padding methods.

[0162] Step S43: performing pixel polar coordinate conversion on the binding node feature vector data to obtain the binding node polar coordinate feature data;

[0163] In an embodiment of the present invention, for the feature vector of each binding node, each coordinate difference (Δx, Δy) therein is converted into polar coordinates (ρ, θ). Wherein, ρ represents the distance, and the calculation formula is: ρ=sqrt(Δx^2+Δy^2); θ represents the angle, and the calculation formula is: θ=arctan(Δy / Δx). The converted polar coordinates (ρ, θ) are combined into a new feature vector as the polar coordinate feature data of the node. For example, if the feature vector of a node is (20, 10, -20, -10), then its polar coordinate feature data is (22.36, 26.57, 22.36, -26.57). Polar coordinate transformation can better express the relative position relationship between nodes and eliminate the influence of translation.

[0164] Step S44: Optimizing the bridge reinforcement tying skeleton data to perform tying node degree statistics, and obtaining tying node connection degree characteristic data;

[0165] In an embodiment of the present invention, in the skeleton data, each node represents a steel bar binding point, and the degree of the node represents the number of edges directly connected to the node. Traverse all nodes in the skeleton data and count the degree of each node. For example, if a node is connected to three edges, its degree is 3. The degree of each node is saved as binding node connection degree feature data. This data reflects the connection status of the node in the skeleton, for example, a node with a degree of 1 represents an endpoint, a node with a degree of 2 represents an ordinary point on the skeleton, and a node with a degree greater than 2 represents a branch point.

[0166] Step S45: performing node geometric feature mapping according to the polar coordinate feature data of the binding nodes and the connectivity feature data of the binding nodes, and calculating the pixel distance between the nodes to generate the pixel distance data of the binding nodes;

[0167] In an embodiment of the present invention, nodes are divided into different types according to the binding node connection degree feature data. For example, nodes with a degree of 2 generally represent the middle point of the steel bar, and nodes with a degree of 3 or 4 generally represent the intersection point of the steel bar. Then, the pixel distance between the nodes is calculated according to the type of the node and the binding node polar coordinate feature data. For a node with a degree of 2, the distance between the node and its two adjacent nodes is calculated. For a node with a degree of 3 or 4, the distance between the node and all its adjacent nodes is calculated. The pixel distance can be calculated using Euclidean distance or other distance measurement methods. The calculated distance is stored as binding node pixel distance data. For example, the pixel coordinates of a node are (100, 200), and the pixel coordinates of its adjacent nodes are (120, 210) and (80, 190), respectively. The pixel distances between the node and its two adjacent nodes are 22.36 and 22.36, respectively.

[0168] Step S46: Use the binding node pixel distance data to match the optimized bridge reinforcement binding skeleton data with adjacent skeleton segments, and use the skeleton actual scale mapping data to perform distance scale conversion to generate an estimated value of the reinforcement binding spacing.

[0169] In an embodiment of the present invention, all line segments in the skeleton are extracted based on the optimized bridge reinforcement binding skeleton data. Then, the adjacent line segments are matched according to the binding node pixel distance data. For example, if two line segments share a node and their directions are roughly perpendicular, the two line segments are considered to be adjacent line segments. For each pair of adjacent line segments, the distance between them is calculated. The shortest distance between line segments or other distance measurement methods can be used. Then, the pixel distance is converted into the actual physical distance using the skeleton actual scale mapping data. The skeleton actual scale mapping data contains the proportional relationship between the image pixels and the actual physical size. Multiplying the pixel distance by the scale factor can obtain the actual physical distance, that is, the estimated value of the reinforcement binding spacing. For example, if the pixel distance between two adjacent line segments is 50 and the scale factor is 0.5 cm / pixel, the estimated value of the reinforcement binding spacing is 25 cm.

[0170] As an example of the present invention, refer to Figure 3 As shown, Figure 1 Detailed implementation steps of step S5 in the flowchart, in this example, step S5 includes:

[0171] Step S51: Obtaining bridge reinforcement spacing design specification data;

[0172] In an embodiment of the present invention, bridge reinforcement spacing design specification data is obtained from bridge design drawings or related specification documents. The data specifies the spacing requirements of different types of reinforcements, such as longitudinal reinforcement spacing, transverse reinforcement spacing, stirrup spacing, etc. The design specification data is usually given in a table, which contains information such as reinforcement type, diameter, spacing range, etc. For example, for longitudinal reinforcement with a diameter of 20 mm, the design specification specifies that its spacing range is 15 cm to 20 cm.

[0173] Step S52: using the bridge reinforcement spacing design specification data to calculate the spacing difference of the reinforcement binding spacing estimation value, and generate bridge binding spacing deviation value data;

[0174] In an embodiment of the present invention, the type of each steel bar and the corresponding design spacing range are determined based on the estimated value of the steel bar binding spacing obtained in step S46 and the bridge steel bar spacing design specification data obtained in step S51. Then, the deviation between the estimated spacing value of each steel bar and the design spacing range is calculated. The deviation can be defined as the difference between the estimated spacing value and the center value of the design spacing, or the difference between the estimated spacing value and the upper and lower limits of the design spacing. For example, the estimated spacing value of a longitudinal steel bar with a diameter of 20 mm is 18 cm, and the design spacing range is 15 cm to 20 cm. The center value of the design spacing is 17.5 cm, and the deviation is 18 cm-17.5 cm=0.5 cm.

[0175] Step S53: mapping the tolerance range according to the bridge lashing spacing deviation value data, and making a compliance determination to obtain lashing deviation state data;

[0176] In an embodiment of the present invention, the bridge reinforcement spacing design specification usually stipulates a tolerance range, such as ±2cm. The spacing deviation value calculated in step S52 is compared with the tolerance range. If the deviation value is within the tolerance range, it is considered that the spacing of the reinforcement meets the requirements of the specification; if the deviation value exceeds the tolerance range, it is considered that the spacing of the reinforcement does not meet the requirements of the specification. For example, the spacing deviation value of a reinforcement is 1cm, and the tolerance range is ±2cm, then the spacing of the reinforcement meets the requirements of the specification. The spacing deviation value of another reinforcement is 3cm, then the spacing of the reinforcement does not meet the requirements of the specification. The compliance judgment result of each reinforcement is stored as binding deviation status data. For example, a Boolean value can be used to represent the compliance status, True means compliance with the specification, and False means non-compliance with the specification. A numerical value can also be used to represent the deviation status, for example, 1 means compliance with the specification, 0 means non-compliance with the specification, and -1 means the deviation is too large.

[0177] Step S54: marking the unqualified spacing position of the optimized bridge reinforcement binding skeleton data based on the binding deviation state data to generate unqualified reinforcement binding position data;

[0178] In an embodiment of the present invention, the binding deviation state data generated in step S53 is traversed to find all steel bars that do not meet the requirements of the specification. The skeleton data is marked according to the position information of these steel bars in the optimized bridge steel bar binding skeleton data. The marking method can be to mark the edges or nodes corresponding to the unqualified steel bars with different colors or marking symbols in the skeleton diagram. For example, the edges corresponding to the unqualified steel bars can be set to red, and the unqualified nodes can be marked with asterisks. The coordinate information of the unqualified steel bars and the corresponding deviation values ​​can also be stored in a new data structure, such as a list or a dictionary. The marked skeleton data or the data structure storing the unqualified steel bar information is stored as unqualified steel bar binding position data.

[0179] Step S55: scoring the construction quality according to the unqualified steel bar binding position data to obtain bridge construction quality scoring data;

[0180] In an embodiment of the present invention, the construction quality score can be calculated based on a variety of indicators, such as the number of unqualified steel bars, the length of unqualified steel bars, the deviation value of unqualified steel bars, etc. Different scoring rules can be defined according to the specific requirements and specifications of the project. For example, the score can be scored according to the proportion of unqualified steel bars to the total number of steel bars. The higher the proportion, the lower the score. The score can also be scored according to the deviation value of the unqualified steel bars. The larger the deviation value, the lower the score. Different weight coefficients can also be set according to different unqualified types. For example, for steel bars in key parts, the weight coefficient can be set higher. For example, a percentage system can be used for scoring, with a full score of 100 points. A certain score is deducted for each unqualified steel bar found, and the deduction ratio can be adjusted according to the size of the deviation value. The calculated scoring result is stored as bridge construction quality scoring data.

[0181] Step S56: Generate a project quality inspection report based on the unqualified steel bar binding position data and the bridge construction quality score data to obtain a bridge steel bar binding project quality inspection report.

[0182] In an embodiment of the present invention, the test report should include the following information: test time, test location, test object, test method, test result, quality score, and specific location and deviation value of unqualified steel bars. The location and distribution of unqualified steel bars can be intuitively displayed in the form of charts or images. For example, unqualified steel bars can be marked with different colors on the bridge steel bar binding skeleton diagram, and the meaning of different colors can be explained in the legend. In addition, the report should also include an analysis and evaluation of the test results, as well as suggestions for subsequent construction. For example, if the quality of steel bar binding in a certain area is found to be poor, it is necessary to point out the problems in the area in the report, and suggest that the construction personnel conduct key inspections and rectifications in the area to carry out auxiliary operations for bridge engineering construction quality inspection.

[0183] The present invention also provides a bridge engineering construction quality detection system, which executes the bridge engineering construction quality detection method as described above, and the bridge engineering construction quality detection system comprises:

[0184] The bridge construction image detection module is used to use a multi-spectral camera array to periodically capture detection images of the steel bar binding site during the bridge construction period to generate original bridge construction site image data; perform image contrast enhancement on the original bridge construction site image data to obtain clear bridge construction image data;

[0185] The bridge reinforcement skeleton extraction module is used to identify the main direction of reinforcement binding for clear bridge construction image data to obtain the main direction data of reinforcement binding; perform reinforcement skeleton topology network processing based on the main direction data of reinforcement binding to obtain initial reinforcement skeleton topology data; perform reinforcement centerline identification based on the initial reinforcement skeleton topology data, and perform skeleton endpoint optimization to generate optimized bridge reinforcement binding skeleton data;

[0186] The skeleton scale mapping module is used to perform bridge surface effect compensation processing on the optimized bridge reinforcement binding skeleton data to generate bridge surface transformation correction coefficients; the bridge surface transformation correction coefficients are used to perform image pixel scale mapping on the optimized bridge reinforcement binding skeleton data to obtain skeleton actual scale mapping data;

[0187] The binding node positioning module is used to map the binding nodes of the clear bridge construction image data by optimizing the bridge reinforcement binding skeleton data to generate binding node pixel mapping data; and to map the binding node pixel mapping data using the skeleton actual scale mapping data;

[0188] The project quality assessment module is used to obtain the bridge reinforcement spacing design specification data; use the bridge reinforcement spacing design specification data to calculate the spacing difference of the reinforcement binding spacing estimate, and perform construction quality scoring to obtain the bridge construction quality scoring data; generate a project quality inspection report based on the bridge construction quality scoring data to obtain a bridge reinforcement binding project quality inspection report.

[0189] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0190] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A bridge engineering construction quality detection method, characterized in that: The following steps are involved: Step S1: using a multispectral camera array to periodically capture images of the steel bar binding site during the bridge construction period to generate original bridge construction site image data; performing image contrast enhancement on the original bridge construction site image data to obtain clear bridge construction image data; Step S2: Recognize the main direction of steel bar binding on the clear bridge construction image data to obtain the main direction data of steel bar binding; perform steel bar skeleton topology network processing according to the main direction data of steel bar binding to obtain initial steel bar skeleton topology data; identify the steel bar centerline according to the initial steel bar skeleton topology data, and optimize the skeleton endpoints to generate optimized bridge steel bar binding skeleton data; Step S3: performing bridge surface effect compensation processing on the optimized bridge reinforcement binding skeleton data to generate a bridge surface transformation correction coefficient; performing image pixel scale mapping on the optimized bridge reinforcement binding skeleton data using the bridge surface transformation correction coefficient to obtain skeleton actual scale mapping data; Step S4: mapping the clear bridge construction image data to binding nodes by optimizing the bridge reinforcement binding skeleton data to generate binding node pixel mapping data; The actual scale mapping data of the skeleton is used to convert the pixel mapping data of the binding nodes into distance scales to generate the estimated value of the steel bar binding spacing; Step S5: Obtain bridge reinforcement spacing design specification data; use the bridge reinforcement spacing design specification data to calculate the spacing difference of the reinforcement binding spacing estimate, and perform construction quality scoring to obtain bridge construction quality scoring data; generate a project quality inspection report based on the bridge construction quality scoring data to obtain a bridge reinforcement binding project quality inspection report; Step S2 includes the following steps: Step S21: performing image multi-scale gradient analysis on the clear bridge construction image data to obtain image multi-scale gradient data; Step S22: using the image multi-scale gradient data to perform non-maximum suppression on the clear bridge construction image data, and perform image edge extraction to generate image edge point response data; Step S23: Recognize the main direction of steel bar binding according to the image edge point response data to obtain the main direction data of steel bar binding; perform virtual rectangular grid processing according to the main direction data of steel bar binding to generate virtual rectangular grid data; Step S24: performing steel bar skeleton topology network processing on the image edge point response data through virtual rectangular grid data and steel bar binding main direction data to obtain initial steel bar skeleton topology data; Step S25: identifying the center line of the steel bar according to the initial steel bar skeleton topology data to obtain the center line data of the steel bar; Step S26: using the steel bar centerline data to perform skeleton endpoint detection on the initial steel bar skeleton topology data to obtain binding skeleton endpoint marking data; Step S27: Optimize the initial steel bar skeleton topology data through the binding skeleton endpoint marking data to generate optimized bridge steel bar binding skeleton data.

2. The bridge engineering construction quality detection method according to claim 1 is characterized in that: Step S1 includes the following steps: Step S11: using a multispectral camera array to periodically capture images of the steel bar binding site during the bridge construction period to generate original bridge construction site image data; Step S12: performing image noise evaluation based on the original bridge construction site image data to obtain shooting noise level data; Step S13: performing multi-scale Gaussian blurring on the original bridge construction site image data by shooting noise level data to generate a multi-scale blur map of the construction site; Step S14: Calculate illumination components at each scale according to the multi-scale fuzzy map of the construction site to obtain shooting illumination component data; Step S15: performing contrast enhancement weight coefficient processing according to the captured illumination component data to generate a contrast enhancement weight coefficient; Step S16: Perform multi-scale weight fusion on the multi-scale fuzzy image of the construction site through the contrast enhancement weight coefficient, and perform adaptive image contrast enhancement to obtain clear bridge construction image data.

3. The bridge engineering construction quality detection method according to claim 1 is characterized in that: Step S23 includes the following steps: Step S231: performing edge connection according to the image edge point response data to obtain closed edge contour data; Step S232: performing gap area recognition on the image multi-scale gradient data by closing the edge contour data to obtain the steel bar binding gap area; Step S233: extracting the outline pixel coordinates of the steel bar binding gap area to obtain outline pixel coordinate data; Step S234: Calculate the average values ​​of horizontal and vertical coordinates according to the contour pixel coordinate data, and perform regional centroid coordinate processing to generate gap region centroid coordinate data; Step S235: identifying the main direction of steel bar binding according to the center of gravity coordinate data of the gap area to obtain the main direction data of steel bar binding; Step S236: Perform virtual rectangular grid processing on the steel bar binding gap area according to the steel bar binding direction data to generate virtual rectangular grid data.

4. The bridge engineering construction quality detection method according to claim 1 is characterized in that: Step S24 includes the following steps: Step S241: performing edge point confidence calculation according to the image edge point response data to generate image edge point confidence data; Step S242: performing image directional weighted neighborhood growth on the clear bridge construction image data based on the main direction data of the steel bar binding and using the image edge point confidence data to generate a directional local weighted area map; Step S243: optimizing the edge steel bar line pixels of the directional local weighted area map, and identifying the cross nodes according to the virtual rectangular grid data to obtain the steel bar binding node data; Step S244: Connect adjacent nodes according to the steel bar binding node data to obtain steel bar binding connection edge data; Step S245: Mark branch points according to the steel bar binding node data and the steel bar binding connection edge data, and perform steel bar skeleton topology processing to generate initial steel bar skeleton topology data.

5. The bridge engineering construction quality detection method according to claim 1 is characterized in that: Step S27 includes the following steps: Step S271: performing endpoint domain connection node analysis according to the binding skeleton endpoint mark data to obtain skeleton endpoint domain node data; Step S272: evaluating the endpoint direction consistency of the binding skeleton endpoint mark data through the skeleton endpoint domain node data to obtain an endpoint direction consistency value; Step S273: evaluating the density connection distance threshold of the bound skeleton endpoint mark data through the skeleton endpoint domain node data to obtain a connection distance threshold; Step S274: performing a skeleton connectivity impact analysis on the bound skeleton endpoint mark data through the skeleton endpoint domain node data to obtain a connectivity impact value; Step S275: weighted false endpoint scoring is performed according to the endpoint direction consistency value, the connection distance threshold, and the connectivity influence value, and endpoint elimination processing is performed on the initial steel skeleton topology data to obtain endpoint optimized steel skeleton data; Step S276: performing breakpoint detection on the endpoint optimized steel bar skeleton data, and searching for the optimal breakpoint topological connection path to obtain the optimal breakpoint topological connection path data; Step S277: Perform break connection optimization on the endpoint optimized steel bar skeleton data through the optimal breakpoint topology connection path data to generate optimized bridge steel bar binding skeleton data.

6. The bridge engineering construction quality detection method according to claim 1 is characterized in that: Step S3 includes the following steps: Step S31: performing calibration corner point detection on the optimized bridge reinforcement binding skeleton data to obtain binding skeleton calibration point data; Step S32: Based on the preset calibration plate geometry data, calibration point matching is performed on the optimized bridge reinforcement binding skeleton data through the binding skeleton calibration point data, and bridge curved surface effect compensation processing is performed to generate a bridge curved surface transformation correction coefficient; Step S33: extracting skeleton line segments according to the optimized bridge reinforcement binding skeleton data to generate bridge reinforcement skeleton line segment data; Step S34: performing bridge structure steel bar vanishing point analysis based on the bridge steel bar skeleton line segment data to generate bridge structure skeleton vanishing point fitting data; wherein step S34 is specifically as follows: Step S341: classifying the bridge steel skeleton line segment data into bridge structure to obtain bridge structure classification line segment data; Step S342: classifying the steel bar binding line segment direction of the bridge structure classification line segment data to obtain steel bar binding direction type data; Step S343: extracting parallel line groups according to the classified line segment data of the bridge structure to obtain parallel line group data of steel bars; Step S344: calculating the straight line length of the parallel line group data of the steel bars to obtain the straight line length data of the parallel line group of the steel bars; Step S345: performing intersection analysis on the direction type data of the steel bar binding by using the straight line length data of the steel bar parallel line group, and performing vanishing point distribution quantity fitting to obtain vanishing point fitting data of the bridge structure skeleton; Step S35: performing image perspective transformation processing according to the vanishing point fitting data of the bridge structure skeleton, and performing transformation matrix correction using the bridge surface transformation correction coefficient to generate an image correction transformation matrix; Step S36: Perform image pixel scale mapping on the optimized bridge reinforcement binding skeleton data through an image correction transformation matrix to obtain actual skeleton scale mapping data.

7. The bridge engineering construction quality detection method according to claim 1 is characterized in that: Step S4 includes the following steps: Step S41: mapping binding nodes to the clear bridge construction image data by optimizing the bridge reinforcement binding skeleton data to generate binding node pixel mapping data; Step S42: performing adjacent node coordinate feature vector processing according to the binding node pixel mapping data to generate binding node feature vector data; Step S43: performing pixel polar coordinate conversion on the binding node feature vector data to obtain the binding node polar coordinate feature data; Step S44: Optimizing the bridge reinforcement tying skeleton data to perform tying node degree statistics, and obtaining tying node connection degree characteristic data; Step S45: performing node geometric feature mapping according to the polar coordinate feature data of the binding nodes and the connectivity feature data of the binding nodes, and calculating the pixel distance between the nodes to generate the pixel distance data of the binding nodes; Step S46: Use the binding node pixel distance data to match the optimized bridge reinforcement binding skeleton data with adjacent skeleton segments, and use the skeleton actual scale mapping data to perform distance scale conversion to generate an estimated value of the reinforcement binding spacing.

8. The bridge engineering construction quality detection method according to claim 1 is characterized in that: Step S5 includes the following steps: Step S51: Obtaining bridge reinforcement spacing design specification data; Step S52: using the bridge reinforcement spacing design specification data to calculate the spacing difference of the reinforcement binding spacing estimation value, and generate bridge binding spacing deviation value data; Step S53: mapping the tolerance range according to the bridge lashing spacing deviation value data, and making a compliance determination to obtain lashing deviation state data; Step S54: marking the unqualified spacing position of the optimized bridge reinforcement binding skeleton data based on the binding deviation state data to generate unqualified reinforcement binding position data; Step S55: scoring the construction quality according to the unqualified steel bar binding position data to obtain bridge construction quality scoring data; Step S56: Generate a project quality inspection report based on the unqualified steel bar binding position data and the bridge construction quality score data to obtain a bridge steel bar binding project quality inspection report.

9. A bridge engineering construction quality inspection system, characterized in that: Used to execute the bridge engineering construction quality detection method as claimed in claim 1, the bridge engineering construction quality detection system comprises: The bridge construction image detection module is used to use a multi-spectral camera array to periodically capture detection images of the steel bar binding site during the bridge construction period to generate original bridge construction site image data; perform image contrast enhancement on the original bridge construction site image data to obtain clear bridge construction image data; The bridge reinforcement skeleton extraction module is used to identify the main direction of reinforcement binding for clear bridge construction image data to obtain the main direction data of reinforcement binding; perform reinforcement skeleton topology network processing based on the main direction data of reinforcement binding to obtain initial reinforcement skeleton topology data; perform reinforcement centerline identification based on the initial reinforcement skeleton topology data, and perform skeleton endpoint optimization to generate optimized bridge reinforcement binding skeleton data; The skeleton scale mapping module is used to perform bridge surface effect compensation processing on the optimized bridge reinforcement binding skeleton data to generate bridge surface transformation correction coefficients; the bridge surface transformation correction coefficients are used to perform image pixel scale mapping on the optimized bridge reinforcement binding skeleton data to obtain skeleton actual scale mapping data; The binding node positioning module is used to map the binding nodes of the clear bridge construction image data by optimizing the bridge reinforcement binding skeleton data to generate binding node pixel mapping data; and to map the binding node pixel mapping data using the skeleton actual scale mapping data; The project quality assessment module is used to obtain the bridge reinforcement spacing design specification data; use the bridge reinforcement spacing design specification data to calculate the spacing difference of the reinforcement binding spacing estimate, and perform construction quality scoring to obtain the bridge construction quality scoring data; generate a project quality inspection report based on the bridge construction quality scoring data to obtain a bridge reinforcement binding project quality inspection report.

Images