An image recognition-based bridge crack global intelligent inspection and evaluation system and method

Abstract

The application discloses a bridge crack global intelligent inspection and evaluation system based on image recognition, which comprises an unmanned aerial vehicle (UAV) inspection platform, an image acquisition module, a crack identification module and a quantitative analysis and evaluation module. The image acquisition module is integrated in the UAV inspection platform, the UAV inspection platform performs global image acquisition on a bridge structure along a preset route, the image acquisition module acquires multispectral image data and visible light image data of a bridge surface, the crack identification module performs crack detection and segmentation on the acquired image data based on a deep learning model, and outputs a crack region mask, and the quantitative analysis and evaluation module extracts geometric parameters of cracks according to the crack region mask. Compared with the prior art, the bridge crack global intelligent inspection and evaluation system based on image recognition has the advantages that global coverage and high-precision identification are realized, and dynamic risk evaluation is performed.

Description

Technical Field

[0001] This invention relates to the field of bridge safety inspection technology, specifically to a bridge crack full-area intelligent inspection and evaluation system and method based on image recognition. Background Technology

[0002] As a crucial component of transportation infrastructure, the structural safety of bridges is directly related to public safety. Cracks are one of the most common forms of structural damage to bridges, and their occurrence and development often indicate a degradation of the structure's load-bearing capacity.

[0003] Traditional bridge crack detection mainly relies on manual visual inspection or contact measurement, which has problems such as low detection efficiency, strong subjectivity, and high risk of high-altitude operations.

[0004] With the development of drone technology and computer vision technology, bridge inspection methods based on image recognition are gradually being applied. However, existing technologies suffer from unstable image quality due to fluctuations in flight attitude, and often rely on static thresholds for judgment, resulting in evaluation results that lag behind actual structural changes. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to overcome the above-mentioned technical defects and provide an image recognition-based intelligent inspection and assessment system and method for bridge cracks that achieves full coverage, high-precision identification, and dynamic risk assessment.

[0006] To solve the above-mentioned technical problems, the technical solution provided by the present invention is: a bridge crack full-domain intelligent inspection and evaluation system based on image recognition, including a drone inspection platform, an image acquisition module, a crack recognition module, and a quantitative analysis and evaluation module; The image acquisition module is integrated into the UAV inspection platform, which performs full-area image acquisition of the bridge structure along a preset route. The image acquisition module acquires multispectral image data and visible light image data of the bridge surface, and the crack identification module performs crack detection and segmentation on the acquired image data based on a deep learning model, and outputs crack region mask. The quantitative analysis and evaluation module extracts the geometric parameters of the cracks based on the crack area mask, assesses the crack risk level based on the geometric parameters and bridge structural parameters, and generates a crack distribution map of the entire bridge area.

[0007] Preferably, the UAV inspection platform further includes an RTK positioning unit and a gimbal attitude stabilization unit. The RTK positioning unit acquires centimeter-level positioning coordinates of the image acquisition location, and the gimbal attitude stabilization unit is integrated into the image acquisition module to compensate for the impact of UAV flight attitude fluctuations on image acquisition.

[0008] Preferably, the crack identification module includes a preprocessing unit, a feature extraction unit, and a crack segmentation unit; The preprocessing unit performs noise reduction, enhancement, and geometric correction on the image data, while the feature extraction unit extracts multi-scale features of cracks in the processed image data. The crack segmentation unit outputs a pixel-level crack region mask.

[0009] Preferably, the quantitative analysis and evaluation module includes a crack quantification unit and a global evaluation unit; The crack quantization unit extracts crack length, width, area, and orientation parameters based on the crack region mask; The global assessment unit determines the crack risk level based on a comparison of geometric parameters and preset thresholds.

[0010] Preferably, the image acquisition module further includes a lidar scanning unit to acquire three-dimensional point cloud data of the bridge structure, and the crack quantization unit further includes a three-dimensional reconstruction unit; The three-dimensional reconstruction unit calculates the crack depth parameters based on the acquired three-dimensional point cloud data of the crack region.

[0011] Preferably, the global evaluation unit further includes a time-series evolution analysis unit; The time-series evolution analysis unit compares the current inspection data with historical inspection data to analyze the crack expansion trend.

[0012] Preferably, a data management platform is also included; The data management platform stores, queries, and visualizes crack area masks, geometric parameters, risk levels, and global crack distribution maps.

[0013] Another aspect of this invention discloses a method for intelligent inspection and evaluation of bridge cracks across the entire area based on image recognition, comprising the following steps: S1: Divide the inspection area according to the bridge structure parameters and generate the optimal flight path, and control the UAV inspection platform to collect full-area images of the bridge structure along the flight path; S2: Register, fuse, and preprocess the acquired multi-source image data, perform crack detection and segmentation based on a deep learning model, and output pixel-level crack region masks; S3: Perform topological connectivity analysis on the crack region mask to extract the geometric parameters and spatial distribution characteristics of the crack network; S4: Construct a dynamic assessment model for crack risk, evaluate the crack risk level based on geometric parameters, spatial distribution characteristics and bridge structural parameters, and generate a crack distribution map of the entire bridge area.

[0014] Preferably, the crack risk dynamic assessment model in S4 also incorporates historical inspection data, and analyzes the crack propagation rate and evolution trend through time-series comparison.

[0015] The advantages of this invention compared with the prior art are as follows: In this invention, the route planning driven by the bridge structure parameters enables full-area coverage inspection of complex curved bridge structures, reduces image acquisition blind spots, and the gimbal attitude stabilization unit compensates for flight attitude fluctuations, thereby improving image acquisition stability. This invention also introduces a dynamic assessment model for crack risk based on historical data, enabling prediction of crack development trends, improving assessment timeliness, and enhancing detection accuracy. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of a bridge crack full-area intelligent inspection and evaluation system based on image recognition.

[0017] Figure 2 This is a flowchart illustrating a method for intelligent inspection and evaluation of bridge cracks based on image recognition. Detailed Implementation

[0018] The present invention will now be described in further detail with reference to the accompanying drawings.

[0019] Combined with appendix Figure 1-2 As shown, a bridge crack full-area intelligent inspection and evaluation system based on image recognition includes a drone inspection platform, an image acquisition module, a crack recognition module, and a quantitative analysis and evaluation module.

[0020] The image acquisition module is integrated into the UAV inspection platform. The UAV inspection platform performs full-domain image acquisition of the bridge structure along a preset route. The image acquisition module acquires multispectral image data and visible light image data of the bridge surface. The crack recognition module performs crack detection and segmentation on the acquired image data based on a deep learning model and outputs crack area mask. The quantitative analysis and evaluation module extracts the geometric parameters of the cracks based on the crack area mask, assesses the crack risk level based on the geometric parameters and bridge structural parameters, and generates a crack distribution map of the entire bridge area.

[0021] When in use, the UAV inspection platform also includes an RTK positioning unit and a gimbal attitude stabilization unit. The RTK positioning unit obtains centimeter-level positioning coordinates of the image acquisition location, and the gimbal attitude stabilization unit is integrated into the image acquisition module to compensate for the impact of UAV flight attitude fluctuations on image acquisition. The crack recognition module includes a preprocessing unit, a feature extraction unit, and a crack segmentation unit. The preprocessing unit performs noise reduction, enhancement, and geometric correction on the image data. The feature extraction unit extracts multi-scale features of cracks in the processed image data. The crack segmentation unit outputs pixel-level crack region masks. The quantitative analysis and evaluation module includes a crack quantification unit and a global evaluation unit. The crack quantification unit extracts crack length, width, area, and orientation parameters based on crack region masks. The global evaluation unit determines the crack risk level based on the comparison of geometric parameters with preset thresholds. The image acquisition module also includes a lidar scanning unit to acquire three-dimensional point cloud data of the bridge structure. The crack quantification unit also includes a three-dimensional reconstruction unit. The three-dimensional reconstruction unit calculates crack depth parameters based on the acquired three-dimensional point cloud data of the crack region. In one embodiment, the global evaluation unit further includes a time-series evolution analysis unit; The time-series evolution analysis unit compares the current inspection data with historical inspection data to analyze the crack expansion trend.

[0022] In one embodiment, a data management platform is also included; The data management platform stores, queries, and visualizes crack area masks, geometric parameters, risk levels, and global crack distribution maps.

[0023] In specific implementation, this invention includes the following steps: The inspection area is divided according to the bridge structural parameters and the optimal flight route is generated. The UAV inspection platform is controlled to collect full-domain images of the bridge structure along the route. The bridge structural parameters include bridge type, span arrangement, component division, structural dimensions and clearance restrictions. The inspection area is divided into independent areas such as the main beam area, pier area, abutment area, bearing area and bridge deck system. Different image acquisition resolution and overlap rate requirements are set for each area according to the geometric complexity and the degree of crack occurrence. The optimal flight path generation is based on a three-dimensional trajectory planning algorithm, which comprehensively considers the UAV's endurance constraints, wind resistance, shooting distance, image overlap rate and obstacle avoidance requirements to generate the shortest flight path covering all inspection areas. The flight path height is set according to the height of the bridge components and the shooting resolution requirements. The RTK positioning unit records the acquisition coordinates of each image in real time, and the gimbal attitude stabilization unit compensates for flight attitude fluctuations in real time to ensure the integrity and clarity of the full-domain image acquisition. The acquired multi-source image data is registered, fused, and preprocessed. Crack detection and segmentation are performed based on a deep learning model, and pixel-level crack region masks are output. The multi-source image data includes visible light images and multispectral images. The multispectral images and visible light images are spatially aligned and fused to generate a fused image with both high spatial resolution and high spectral resolution. Preprocessing includes denoising, enhancement, and geometric correction. The deep learning model uses a trained crack segmentation neural network. The preprocessed image is input, and the crack region mask is output at the pixel level. The classification confidence of each pixel in the mask is output synchronously. Pixels with confidence scores lower than a preset threshold are marked as uncertain regions and left for manual review. Topological connectivity analysis is performed on the crack region mask to extract the geometric parameters and spatial distribution characteristics of the crack network. The topological connectivity analysis divides the crack region mask into independent crack connected domains. Each connected domain represents an independent crack or a branch of the crack network. The skeleton of each connected domain is extracted to obtain the crack centerline network. Based on the centerline network, the total length, number of branches, number of nodes and network complexity index of the crack network are calculated. Spatial distribution characteristics include the location distribution of cracks on bridge components, crack density distribution, and the angle distribution between crack direction and the direction of principal stress of the structure. Geometric parameters include crack length, width, area, and direction, which are calculated and output by crack quantization unit. A dynamic assessment model for crack risk is constructed. Based on geometric parameters, spatial distribution characteristics and bridge structural parameters, the crack risk level is assessed and a crack distribution map of the entire bridge area is generated. The dynamic assessment model for crack risk adopts a multi-factor assessment model that combines the analytic hierarchy process (AHP) and the fuzzy comprehensive evaluation method. The AHP determines the weight vectors of geometric parameters, spatial distribution characteristics and structural parameters. The weights of geometric parameters include crack length weight, width weight, area weight and depth weight. Spatial distribution characteristic weights include crack density weight, network complexity weight, and orientation danger angle weight; structural parameter weights include component importance weight and material sensitivity weight; the fuzzy comprehensive evaluation method maps the measured values ​​of each parameter to a preset risk level membership function, calculates the comprehensive risk score, and determines the crack risk level based on the score interval. The overall crack distribution map uses the bridge plan view or BIM model as the base map, and overlays the spatial location, risk level and geometric parameter annotations of each crack to form an overview map of the overall crack distribution of the bridge.

[0024] In one embodiment, the crack risk dynamic assessment model also incorporates historical inspection data. Through time-series comparison analysis, it analyzes the crack propagation rate and evolution trend. The historical inspection data includes crack area masks, geometric parameters, risk levels, and collection timestamps from each inspection. When calculating the current risk level, the crack risk dynamic assessment model incorporates the crack propagation rate and evolution trend as dynamic correction factors.

[0025] All standard parts used in this invention can be purchased from the market. The UAV platform, image sensor, LiDAR, RTK module and computing device are all conventional models in the prior art. In addition, the deep learning model training, point cloud processing and 3D visualization all adopt mature algorithm frameworks and software tools in the prior art, which will not be described in detail here. The contents not described in detail in this specification are prior art known to those skilled in the art.

[0026] The contents not described in detail in this specification are existing technologies known to those skilled in the art.

[0027] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0028] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A bridge crack comprehensive intelligent inspection and evaluation system based on image recognition, characterized in that: Includes a drone inspection platform, image acquisition module, crack recognition module, and quantitative analysis and evaluation module; The image acquisition module is integrated into the UAV inspection platform, which performs full-area image acquisition of the bridge structure along a preset route. The image acquisition module acquires multispectral image data and visible light image data of the bridge surface, and the crack identification module performs crack detection and segmentation on the acquired image data based on a deep learning model, and outputs crack region mask. The quantitative analysis and evaluation module extracts the geometric parameters of the cracks based on the crack area mask, assesses the crack risk level based on the geometric parameters and bridge structural parameters, and generates a crack distribution map of the entire bridge area.

2. The intelligent inspection and evaluation system for bridge cracks based on image recognition according to claim 1, characterized in that: The UAV inspection platform also includes an RTK positioning unit and a gimbal attitude stabilization unit. The RTK positioning unit acquires centimeter-level positioning coordinates of the image acquisition location, and the gimbal attitude stabilization unit is integrated into the image acquisition module to compensate for the impact of UAV flight attitude fluctuations on image acquisition.

3. The intelligent inspection and evaluation system for bridge cracks based on image recognition according to claim 1, characterized in that: The crack identification module includes a preprocessing unit, a feature extraction unit, and a crack segmentation unit; The preprocessing unit performs noise reduction, enhancement, and geometric correction on the image data, while the feature extraction unit extracts multi-scale features of cracks in the processed image data. The crack segmentation unit outputs a pixel-level crack region mask.

4. The intelligent inspection and evaluation system for bridge cracks based on image recognition according to claim 1, characterized in that: The quantitative analysis and evaluation module includes a crack quantification unit and a global evaluation unit; The crack quantization unit extracts crack length, width, area, and orientation parameters based on the crack region mask; The global assessment unit determines the crack risk level based on a comparison of geometric parameters and preset thresholds.

5. The intelligent inspection and evaluation system for bridge cracks based on image recognition according to claim 4, characterized in that: The image acquisition module also includes a lidar scanning unit to acquire three-dimensional point cloud data of the bridge structure, and the crack quantization unit also includes a three-dimensional reconstruction unit. The three-dimensional reconstruction unit calculates the crack depth parameters based on the acquired three-dimensional point cloud data of the crack region.

6. The intelligent inspection and evaluation system for bridge cracks based on image recognition according to claim 4, characterized in that: The global assessment unit also includes a time-series evolution analysis unit; The time-series evolution analysis unit compares the current inspection data with historical inspection data to analyze the crack expansion trend.

7. The intelligent inspection and evaluation system for bridge cracks based on image recognition according to claim 1, characterized in that: It also includes a data management platform; The data management platform stores, queries, and visualizes crack area masks, geometric parameters, risk levels, and global crack distribution maps.

8. A method for intelligent inspection and evaluation of bridge cracks based on image recognition, using the system as described in any one of claims 1-7, characterized in that: Includes the following steps: S1: Divide the inspection area according to the bridge structure parameters and generate the optimal flight path, and control the UAV inspection platform to collect full-area images of the bridge structure along the flight path; S2: Register, fuse, and preprocess the acquired multi-source image data, perform crack detection and segmentation based on a deep learning model, and output pixel-level crack region masks; S3: Perform topological connectivity analysis on the crack region mask to extract the geometric parameters and spatial distribution characteristics of the crack network; S4: Construct a dynamic assessment model for crack risk, evaluate the crack risk level based on geometric parameters, spatial distribution characteristics and bridge structural parameters, and generate a crack distribution map of the entire bridge area.

9. The method for intelligent inspection and evaluation of bridge cracks based on image recognition according to claim 8, characterized in that: The crack risk dynamic assessment model in S4 also incorporates historical inspection data, and analyzes the crack propagation rate and evolution trend through time-series comparison.

Images