Bridge apparent disease intelligent detection device based on machine vision and displacement compensation

Abstract

The application discloses a kind of bridge apparent disease intelligent detection device based on machine vision and displacement compensation, belong to bridge detection technical field.The device includes image acquisition module, light supplement module, laser ranging module, displacement compensation detection beam frame module, image acquisition engineering computer module and image analysis processing module, displacement compensation detection beam frame module is used to absorb the shaking displacement in detection process and record horizontal direction relative coordinate information;Laser ranging module records vertical direction relative coordinate information in real time, and the displacement coordinate information obtained by laser ranging module and electronic digital display rolling ruler range finder is used to carry out displacement compensation calibration to image data, and the automatic identification of disease is realized in combination with convolution neural network algorithm.The application has high integration degree, has displacement compensation function, can effectively eliminate the influence of cantilever shaking on image acquisition quality during the running of bridge detection vehicle, adapts to complex lighting environment, realizes the efficient, accurate and automatic detection of bridge apparent disease.

Description

Technical Field

[0001] This invention relates to the field of bridge inspection technology, specifically to an intelligent detection device for bridge surface defects based on machine vision and displacement compensation. Background Technology

[0002] As a crucial component of transportation infrastructure, the structural safety of bridges directly impacts the safety of people's lives and property and the smooth flow of traffic. With the increasing service life of bridges, factors such as environmental erosion, material aging, and repeated vehicle loads can cause surface defects such as cracks, spalling, and rust to appear on the bridge structure. If these defects are not detected and addressed in a timely manner, they may develop into structural damage or even lead to major safety accidents. Therefore, regular inspections of bridge surface defects are of paramount importance.

[0003] Currently, bridge surface defects detection mainly relies on manual visual inspection and traditional inspection equipment. Manual inspection requires close observation using bridge inspection vehicles, scaffolding, and other tools, which has drawbacks such as low inspection efficiency, high subjectivity, and high risks associated with working at heights. Traditional inspection equipment, such as optical telescopes and ordinary cameras, can assist in recording defects, but still requires manual interpretation, resulting in low automation.

[0004] In recent years, intelligent detection methods for bridge surface defects based on machine vision have gradually attracted attention. However, existing technologies still face many technical bottlenecks: First, the bottom of bridges generally suffers from low and uneven illumination, which seriously affects the quality of image acquisition. Second, during the movement of bridge inspection vehicles, factors such as uneven road surfaces and vehicle dynamic loads cause the inspection cantilever to sway, resulting in dynamic changes in the distance between the camera lens and the bottom of the beam. This leads to problems such as blurry and inconsistent scale in the acquired images, and existing technologies lack effective displacement compensation mechanisms. Third, existing inspection equipment is mostly general-purpose and lacks a dedicated integrated design for bridge inspection vehicle operation scenarios. The functional modules are scattered, making it difficult to flexibly adjust according to on-site conditions. Furthermore, the correlation between image data and spatial location information is weak, which is not conducive to accurate defect localization.

[0005] Therefore, there is an urgent need in this field for an intelligent detection device and method for bridge surface defects that is highly integrated, has displacement compensation function, and can adapt to complex lighting environments. Summary of the Invention

[0006] This invention aims to solve the following technical problems existing in current bridge surface defect detection technologies: First, manual inspection is inefficient, subjective, and carries high safety risks; second, existing intelligent inspection equipment produces poor image quality under low illumination and uneven lighting conditions at the bottom of the bridge; third, the swaying of the inspection cantilever during bridge inspection vehicle movement leads to blurred images and inconsistent scales, and there is a lack of effective displacement compensation mechanisms; fourth, existing inspection equipment has low integration, with functional modules scattered and difficult to adjust flexibly according to on-site conditions, and the correlation between image data and spatial location information is weak, which is not conducive to accurate defect location.

[0007] The purpose of this invention is to provide an intelligent detection device for bridge surface defects based on machine vision and displacement compensation, so as to achieve efficient, accurate and automated detection of bridge surface defects.

[0008] To achieve the above objectives, the present invention provides the following technical solution: A smart detection device for bridge surface defects based on machine vision and displacement compensation includes a hardware system and a software system.

[0009] The hardware system includes: (1) Image acquisition module, used to acquire images of the surface of the bridge structure. This module uses multiple 13-megapixel zoom lens industrial cameras, which are positioned within 100cm of the bottom of the beam, and the lens orientation is perpendicular to the bottom of the beam to ensure the clarity and consistency of the acquired images.

[0010] (2) Supplemental lighting module, used to provide supplemental lighting for the image acquisition module. This module includes a lamp holder, an air-cooled heat sink and LED beads. The LED beads use blue-violet light sources, which can effectively filter ambient light interference and solve the problems of low and uneven illumination at the bottom of the bridge.

[0011] (3) Laser ranging module, used to measure the distance change of the image acquisition module relative to the bottom of the beam and the vertical relative coordinate information in real time. This module includes a laser rangefinder probe and a sensor, which can record the lens displacement change on each time slice for image data calibration and error data removal.

[0012] (4) Displacement compensation detection beam frame module, used to install and fix the image acquisition module, supplementary lighting module and laser ranging module. This module integrates a rigid spring compensation structure for absorbing equipment sway displacement and a ranging component for recording horizontal relative coordinate information, and is installed and fixed on the upper truss beam of the bridge inspection vehicle's inspection cantilever. Specifically, this module includes a plum blossom-shaped cross-section aluminum alloy beam frame body, a sliding fixing component that cooperates with a wedge-shaped slot, and an electronic digital display roller ranging instrument with a rigid spring and rollers. During the inspection process, the rollers contact the bottom of the beam, the rigid springs absorb vertical sway displacement, and simultaneously record horizontal relative coordinate information.

[0013] (5) Image acquisition engineering computer module, used to control image acquisition, store data and run image analysis algorithms. This module includes a graphics workstation, a hub and a general-purpose computer, which respectively undertake functions such as image acquisition control and calculation, signal integration and laser ranging data acquisition.

[0014] The software system includes an image analysis and processing module, which is based on a convolutional neural network algorithm and is used to identify and analyze diseases in the acquired images.

[0015] Preferably, the hardware system specifically includes three sets of image acquisition modules, four sets of supplementary lighting modules, one set of laser ranging modules, and one set of electronic digital display roller rangefinders, all of which are fixedly installed on the displacement compensation detection beam.

[0016] This invention also provides an intelligent detection method for bridge surface defects based on machine vision and displacement compensation, comprising the following steps: Step 1: Install the displacement compensation detection beam frame onto the detection cantilever of the bridge inspection vehicle, and adjust the relative positions of the image acquisition module, the supplementary lighting module, and the laser ranging module; Step 2: The roller of the electronic digital display roller measuring instrument contacts the bottom of the beam, and the rigid spring absorbs the vertical swaying displacement during the detection process. Step 3: During the movement of the bridge inspection vehicle, the image acquisition module acquires images of the bridge's appearance, while the laser ranging module and the electronic digital display roller rangefinder record the vertical and horizontal relative coordinate information of the image acquisition module in real time. Step 4: Using the displacement coordinate information recorded by the laser rangefinder module and the electronic digital display roller rangefinder, the acquired image data is calibrated for displacement compensation. Based on the lens displacement changes in each time slice, the image data is matched and corrected with the corresponding spatial coordinate information, and image data with errors caused by excessive shaking is eliminated. Step 5: Use the image analysis and processing module based on the convolutional neural network algorithm to identify and analyze the defects in the calibrated image.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. Achieving automation and intelligence in bridge defect detection. This invention combines an industrial camera image acquisition module with an image analysis and processing module based on a convolutional neural network algorithm to achieve automatic acquisition and intelligent identification of bridge surface defects. This eliminates the need for manual visual interpretation, significantly improving detection efficiency, reducing labor costs, and eliminating subjective differences in detection results caused by human factors.

[0018] 2. Effectively solves the problem of lighting environment interference. This invention uses a blue-violet light source supplementary lighting module, combined with an air-cooled heat dissipation plate, to provide stable and uniform supplementary lighting in complex environments with low illumination and uneven lighting at the bottom of bridges. It effectively filters ambient light interference, ensures the quality of acquired images, and provides a reliable data foundation for subsequent defect identification.

[0019] 3. An innovative displacement compensation mechanism is introduced to eliminate the effects of swaying. This invention uses a laser ranging module to record the vertical displacement of the camera lens in real time, and a rigid spring structure in an electronic digital display roller rangefinder to absorb vertical swaying while recording horizontal coordinate information, thus constructing a complete displacement compensation mechanism. This mechanism effectively eliminates the impact of cantilever swaying on image acquisition quality during the bridge inspection vehicle's movement, performs spatial coordinate matching and calibration on the acquired images, eliminates erroneous data, and significantly improves the accuracy and reliability of defect identification.

[0020] 4. Integrated and modular design with strong adaptability. This invention uses a quincunx-shaped aluminum alloy beam frame as the installation platform. Through wedge-shaped slots and clamps, it enables flexible translation and rapid assembly / disassembly of the image acquisition module, supplementary lighting module, and laser ranging module. The position and quantity of modules can be freely adjusted according to different bridge site conditions. The beam frame itself has high bending stiffness and light weight, facilitating rapid on-site assembly and subsequent functional upgrades, exhibiting high on-site adaptability and expandability.

[0021] 5. Achieving precise location and source tracing of defects. This invention uses a laser ranging module and an electronic digital display roller rangefinder to record the vertical and horizontal coordinate information of each time slice, accurately linking image data with spatial location information. This allows identified defects to be mapped to their actual spatial location on the bridge structure, facilitating precise subsequent maintenance and repair work.

[0022] In summary, this invention provides a highly integrated intelligent detection device and method for bridge surface defects that has displacement compensation function and can adapt to complex lighting environments. It effectively overcomes the shortcomings of existing technologies, such as low detection efficiency, unstable image quality, and high susceptibility to environmental interference, and achieves efficient, accurate, and automated detection of bridge surface defects, which has good application value. Attached Figure Description

[0023] To more clearly illustrate the technical solution of the present invention, the accompanying drawings are described as follows: Figure 1 This is a schematic diagram of the overall model structure of the inspection beam and inspection vehicle of the present invention; Figure 2 This is a schematic diagram of the supplementary lighting module of the present invention; Figure 3 This is a structural schematic diagram of the displacement compensation detection beam frame module of the present invention; Figure 4 This is a schematic diagram of the output of the disease identification results of the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention.

[0025] This embodiment provides an intelligent detection device for bridge surface defects based on machine vision and displacement compensation, the overall structure of which is as follows: Figure 1 As shown, the inspection beam is installed and fixed on the truss beam of the bridge inspection vehicle's inspection cantilever.

[0026] I. Hardware System Composition and Layout 1. Image acquisition module In this embodiment, see Figure 2 and Figure 3 The image acquisition module employs three sets of industrial cameras, each equipped with a 13-megapixel zoom lens. The distance between the cameras and the bottom of the beam is controlled within 100cm, and the lenses are perpendicular to the beam bottom. The vertical distance between the optical axis of the camera lens and the plane of the beam bottom is denoted as... ,satisfy: (1) To ensure that the image resolution meets the requirements for disease identification, the object-space resolution of the camera lens... Determined by the following formula: (2) in, For working distance, For pixel size, This refers to the camera lens focal length. Adjust the camera mounting position to ensure... To meet the requirements for identifying minute cracks (width ≥ 0.2 mm).

[0027] 2. Fill light module like Figure 2 and Figure 3 As shown, the supplementary lighting module includes a lamp holder, an air-cooled heat sink, and LEDs. The air-cooled heat sink is fixedly mounted on the lamp holder, and the LEDs are mounted on the mounting surface of the air-cooled heat sink and are tightly fitted to it. The lamp holder provides overall support and fixation for the air-cooled heat sink and LEDs, while the air-cooled heat sink dissipates heat generated by the LEDs during operation, ensuring stable LED operation. This embodiment uses four sets of supplementary lighting modules, and the LEDs are blue-violet light sources with a center wavelength of [insert wavelength here]. Illuminance of the supplementary lighting module Must meet: (3) in, Φ This represents the total luminous flux of the LED chips (unit: lm). η For the efficiency of the optical system, A Area of ​​the target region at the bottom of the beam (unit: m²) 2 ), E Illuminance for the target area. Center wavelength. λ The optimal value is determined by the factory parameters or spectral calibration results of the selected blue-violet LED chips, with a preferred value of 400. nm -450 nm Within the specified range; the operating temperature of the LED beads is controlled by an air-cooled heat sink to ensure stable light source output.

[0028] 3. Laser ranging module This embodiment employs a laser rangefinder assembly, consisting of a laser rangefinder probe and a sensor. The laser rangefinder module measures the distance change between the camera lens and the bottom of the beam in real time. (Setting a time interval...) The distance between the measured lens and the bottom of the beam is The initial distance is Then the change in vertical displacement is: (4) The sampling frequency of the laser ranging module is set to f , f Based on image acquisition frame rate f I Confirmed, and f Not less than f I ;when f Greater than f I At that time, adjacent ranging data are interpolated according to the image timestamp to ensure that displacement changes on each time slice can be accurately recorded.

[0029] 4. Displacement Compensation Detection Beam Frame Module like Figure 3 As shown, the displacement compensation detection beam frame module includes a plum blossom-shaped cross-section aluminum alloy beam frame body, a sliding fixing assembly composed of wedge-shaped slots and clips, and an electronic digital display roller rangefinder with a rigid spring.

[0030] The roller of the electronic digital display roller rangefinder contacts the bottom of the beam, and the elastic coefficient of the rigid spring is... The initial compression amount is The preload provided by the spring is: (5) When the detection cantilever wobbles vertically, the change in spring compression is: The actual absorbed displacement is: in, The change in force generated by the shaking. This represents the vertical displacement absorbed by the spring. The horizontal relative coordinate information collected by the roller is denoted as... , The value is calculated from the roller angle and roller circumference of the electronic digital display roller rangefinder, and is used to record the positional changes of the camera lens in the horizontal direction. Let t0 be the horizontal coordinates of the position.

[0031] 5. Image Acquisition Engineering Computer Module The image acquisition engineering computer module includes a graphics workstation, a hub, and a general-purpose computer. The graphics workstation is configured with a CPU clock speed ≥ 3.5GHz, a GPU memory ≥ 11GB, and RAM ≥ 32GB, used to run the image acquisition control program and convolutional neural network algorithms. The hub is a USB 3.0 hub with a transmission rate ≥ 5Gbps, used to integrate the signal lines of various modules. The general-purpose computer is used to acquire the vertical coordinate data fed back from the laser ranging module.

[0032] II. Displacement Compensation and Image Calibration Methods During the movement of the bridge inspection vehicle, factors such as uneven road surface and vehicle dynamic load cause the inspection cantilever to sway, resulting in dynamic changes in the distance between the camera lens and the bottom of the beam. This embodiment uses the following displacement compensation method to calibrate the acquired images.

[0033] Let the acquisition time of a certain frame of image be... The vertical distance recorded by the laser ranging module is The horizontal coordinates recorded by the electronic digital display roller rangefinder are: Relative to the initial time The displacement change is: (6) (7) The conversion between the actual object size and pixel size of an image is determined by calibration parameters. Let the pixel equivalent obtained from camera calibration be... (Unit: mm / pixel), then the image scale correction coefficient due to distance changes. for: (8) Therefore, any pixel in the image Corresponding actual spatial coordinates It needs to be corrected according to the following formula: (9-1) (9-2) in, X, Y These are the spatial coordinates (unit: mm) along the longitudinal and transverse directions of the bridge. u、v These are the pixel coordinates of the image.

[0034] For cases where image blurring is caused by excessive shaking, a shaking amplitude threshold is defined. An image frame is considered an error image and is discarded if the following conditions are met: or (10) in, This is the threshold for vertical displacement change. The threshold for horizontal velocity variation is determined by equipment calibration tests or on-site no-load test run data. The preferred threshold is the average value under normal operating conditions plus three times the standard deviation, in order to eliminate abnormal image frames caused by cantilever swaying or sudden vehicle stops.

[0035] III. Software System and Disease Identification Algorithm The software system includes a software module for analyzing and processing surface defects in bridge structures based on convolutional neural network (CNN) algorithms. This embodiment uses the Swin-Unet network structure for semantic segmentation (U-Net, DeepLab, Mask R-CNN, and other semantic segmentation networks can also be used) to achieve pixel-level identification of defects such as cracks and spalling.

[0036] Let the input image be ,in and These represent the height and width of the image, respectively. Convolutional neural networks extract image features through an encoder-decoder structure. The encoder consists of multiple convolutional and pooling layers, and outputs a feature map. ,in This represents the number of feature channels. The decoder restores spatial resolution through upsampling and skip connections, ultimately outputting a disease segmentation probability map. ,in Number of disease categories (in this embodiment, we take...) (These are, respectively, background, cracks, peeling, and rust swelling).

[0037] The segmentation loss function is a weighted sum of cross-entropy loss and Dice loss: (11) Among them, cross-entropy loss Defined as: (12) Dice loss Defined as: (13) In the formula, For pixels Category The actual label (with a value of 0 or 1). The model predicts the category to which this pixel belongs. The probability, and These are the true label and predicted probability corresponding to the target category, respectively. The weighting coefficients are... .

[0038] The model was trained using the Adam optimizer, with an initial learning rate set to... Batch size Number of training rounds The training dataset contains 5,000 labeled images of bridge surface defects, divided into training, validation, and test sets in an 8:1:1 ratio.

[0039] IV. Testing Process The detection process in this embodiment is as follows: Step 1: Equipment Installation and Debugging The displacement compensation detection beam is installed on the truss beam of the cantilever bridge inspector. Through the engagement of wedge-shaped slots and clamps, the relative positions of three industrial cameras, four supplementary lighting modules, and one laser ranging module are adjusted to ensure the distance between the camera lens and the bottom of the beam is [not specified]. The lens should be perpendicular to the bottom of the beam. Adjust the rollers of the electronic digital roller rangefinder to contact the bottom of the beam, ensuring the initial compression of the rigid spring. Spring preload Ensure that the rollers maintain reliable contact with the bottom of the beam.

[0040] Step 2: System Startup and Data Acquisition Start the image acquisition computer module and set the graphics workstation acquisition parameters: image resolution 1920×1080, acquisition frame rate 30. fps Start the laser ranging module and set the sampling frequency. Start the general-purpose computer and begin recording laser ranging data.

[0041] Step 3: Dynamic detection, coordinate recording and displacement compensation Bridge inspection vehicle at speed Proceeding at a constant speed. During the movement, the image acquisition module continuously captures images of the bridge's appearance, while the laser ranging module records the vertical distance in each time slice. Electronic digital display roller rangefinder records horizontal coordinates The displacement compensation parameters are calculated according to formulas (6)-(9), and the spatial coordinates of each frame of the image are calibrated.

[0042] Step 4: Error Data Removal Image frames that satisfy formula (10) are identified as error images, automatically removed, and not included in subsequent analysis.

[0043] Step 5: Disease Identification and Analysis The displacement-compensated and calibrated image is input into the image analysis and processing module based on a convolutional neural network. The graphics workstation loads the pre-trained model and performs lesion segmentation and identification on each frame of the image. The identification results are as follows: Figure 4 As shown, the output includes the disease type, location coordinates, and geometric parameters.

[0044] For the identified crack defects, their width Calculated using the following formula: (14) in, Let be the pixel span of the crack region in the vertical direction. This represents the number of normal pixels in the crack region.

[0045] For the identified peeling areas, their area Calculated using pixel counting: (15) In the formula, This is the set of pixels in the peeled-off area.

[0046] Step 6: Output and Storage of Results The disease identification results are associated with and stored along with the corresponding spatial coordinate information to generate an inspection report. The inspection report includes the disease type, spatial coordinates, geometric dimensions, and disease image, facilitating subsequent maintenance and repair work.

[0047] V. Implementation Results This embodiment achieves the following technical effects: 1. Detection efficiency: Taking a bridge with a single span of 30m as an example, the time to complete the detection of the appearance defects of the entire bridge using the device of this invention is about 15 minutes, which is 87.5% more efficient than manual detection (about 2 hours).

[0048] 2. Image quality: The light intensity at the bottom of the bridge is only 50. lux In this environment, the fill light module provides 500 lux With supplemental illumination, the image signal-to-noise ratio increased from 18dB before illumination to 35dB, and the image clarity was significantly improved.

[0049] 3. Displacement compensation effect: During the movement of the detection vehicle, the vertical sway of the detection cantilever reaches a maximum of ±35mm. After being absorbed by the rigid spring, the sway transmitted to the camera lens is reduced to within ±5mm. The matching error between the displacement-compensated image and the standard image is ≤2%.

[0050] 4. Recognition accuracy: The disease recognition model based on the convolutional neural network algorithm achieved an average intersection-union ratio (mIoU) of 87.6% on the test set, with crack recognition accuracy of 91.2% and spalling recognition accuracy of 85.3%, meeting the requirements of engineering applications.

[0051] In summary, the intelligent bridge surface defect detection device and method based on machine vision and displacement compensation provided by this invention achieves efficient, accurate, and automated detection of bridge surface defects through integrated hardware design, displacement compensation mechanism, and deep learning algorithm, demonstrating significant technological advancement and practical value.

[0052] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. An intelligent detection device for bridge surface defects based on machine vision and displacement compensation, characterized in that, include: The image acquisition module uses an industrial camera to acquire images of the bridge structure's appearance. The supplementary lighting module is used to provide supplementary illumination for the image acquisition module; The laser ranging module is used to measure the distance change of the industrial camera relative to the bottom of the beam and the vertical relative coordinate information in real time. The displacement compensation detection beam module is used to install the image acquisition module, the supplementary lighting module and the laser ranging module. One end of the displacement compensation detection beam module integrates a rigid spring compensation structure for absorbing the shaking of the detection device and a roller rangefinder for recording the horizontal relative coordinate information. The other end is fixed to the upper truss beam of the bridge inspection vehicle's inspection cantilever. The image acquisition engineering computer module is used to control image acquisition, store image data and coordinate information before and after calibration, perform displacement compensation calibration on the acquired images based on the vertical and horizontal relative coordinate information of the industrial camera, and perform defect identification and analysis on the calibrated images based on convolutional neural networks.

2. The intelligent bridge surface defect detection device based on machine vision and displacement compensation according to claim 1, characterized in that, The displacement compensation detection beam includes: The main body of the beam frame is a plum blossom-shaped cross-section aluminum alloy beam frame, and each side of it is provided with a wedge-shaped groove; The sliding fixing assembly includes a wedge-shaped clip that mates with a wedge-shaped slot. The image acquisition module, the supplementary lighting module, and the laser ranging module can be translatably fixed to the main beam frame through the wedge-shaped clip. The rigid spring compensation structure is a rigid spring, and the roller rangefinder is an electronic digital display roller rangefinder. One end of the rigid spring is installed on the main beam frame, and the other end is equipped with an electronic digital display roller rangefinder, which is used to contact the bottom of the beam during the detection process, absorb vertical sway displacement, and record the horizontal relative coordinate information of the image acquisition module.

3. The intelligent bridge surface defect detection device based on machine vision and displacement compensation according to claim 1, characterized in that, The image acquisition module includes an industrial camera with a 13-megapixel zoom lens. The camera is positioned within 100cm of the bottom of the beam, and the lens is perpendicular to the bottom of the beam.

4. The intelligent bridge surface defect detection device based on machine vision and displacement compensation according to claim 1, characterized in that, The supplementary lighting module includes a lamp holder, an air-cooled heat sink, and LED beads. The LED beads use a blue-violet light source to filter ambient light interference.

5. The intelligent bridge surface defect detection device based on machine vision and displacement compensation according to claim 1, characterized in that, The laser ranging module includes a laser rangefinder, which records the lens displacement changes at each time slice for image data calibration and error data removal.

6. The intelligent bridge surface defect detection device based on machine vision and displacement compensation according to claim 1, characterized in that, It includes several image acquisition modules and several supplementary lighting modules. The supplementary lighting modules are evenly installed on the displacement compensation detection beam frame module, and the image acquisition modules are evenly installed between two adjacent supplementary lighting modules.

7. The intelligent bridge surface defect detection device based on machine vision and displacement compensation according to claim 1, characterized in that, The image acquisition module includes several industrial cameras, the supplementary lighting module includes several supplementary lights, at least one set of laser ranging modules and at least one set of electronic digital display roller rangefinders. The supplementary lights are evenly spaced on the displacement compensation detection beam frame module, and an industrial camera is evenly spaced between two adjacent supplementary lights on the displacement compensation detection beam frame module. The displacement compensation detection beam frame module is fixed to the detection cantilever of the bridge inspection vehicle.

8. A method for intelligent detection of bridge surface defects based on machine vision and displacement compensation, applied to the apparatus described in any one of claims 1-7, characterized in that, Includes the following steps: Step 1: Install the displacement compensation detection beam frame module on the detection cantilever of the bridge inspection vehicle, and adjust the relative positions of the image acquisition module, the supplementary lighting module, and the laser ranging module; Step 2: The roller of the roller measuring instrument contacts the bottom of the beam, and a rigid spring is used to compensate for the vertical swaying displacement of the structure during the detection process. Step 3: During the movement of the bridge inspection vehicle, the image acquisition module acquires images of the bridge's appearance, while the laser ranging module and the roller rangefinder record the vertical and horizontal relative coordinate information of the image acquisition module in real time. Step 4: Using the displacement coordinate information recorded by the laser ranging module and the roller rangefinder, perform displacement compensation calibration on the acquired image data and discard image data with errors caused by excessive shaking. Step 5: Use the image analysis and processing module based on the convolutional neural network algorithm to identify and analyze the defects in the calibrated image.

9. The intelligent bridge surface defect detection device based on machine vision and displacement compensation according to claim 8, characterized in that, In step 1, the supplementary lighting module uses a blue-violet light source for supplementary lighting and a heat dissipation plate for heat dissipation to ensure high-quality image acquisition in low-light or uneven lighting environments.

10. The intelligent bridge surface defect detection device based on machine vision and displacement compensation according to claim 8, characterized in that, In step 4, the displacement compensation calibration specifically includes: matching and correcting the image data with the corresponding spatial coordinate information based on the lens displacement changes in each time slice.

Images