Method, system and equipment for detecting cracks of box girder based on perfect colors and medium

Abstract

The invention discloses a method, a system and equipment for detecting a crack of a box girder based on a perfect color, and a medium, and relates to the technical field of image processing, and the method comprises the steps: carrying out the image data collection of the interior of a to-be-detected variable cross-section box girder through a perfect color collection device, recording the brightness of a collection position in real time, and generating an image set, the image set comprises at least two images and brightness corresponding to each image; image splicing processing is carried out on the image data in the image set to obtain a to-be-processed target image, disease data analysis processing is carried out on the to-be-processed target image, and the crack type and area of the to-be-detected variable cross-section box girder are determined; and performing three-dimensional display on the to-be-detected variable cross-section box girder according to the crack type and the area. According to the method, the disease condition of the variable cross-section box girder can be intuitively presented, so that technicians and decision makers can more clearly know the distribution and severity of the diseases, and a more reasonable maintenance strategy is formulated.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a method, system, equipment and medium for detecting cracks in a box girder based on Zhenquancai. Background Art

[0002] The industry currently relies on manual inspection for internal and external cracks in continuously variable-section box girders, using flashlights for supplemental lighting and cameras to record information. This primarily involves inspecting the two sub-slabs and the top slab. This relies entirely on manual fieldwork, resulting in inaccurate data and low efficiency. Furthermore, internal data processing is labor-intensive and presents data in a non-intuitive manner. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to address the deficiencies of the existing technology and specifically provide a method, system, equipment and medium for detecting cracks in box girders based on Zhenquancai, as follows:

[0004] 1) In the first aspect, the present invention provides a method for detecting cracks in a box girder based on Zhenquancai, and the specific technical solution is as follows:

[0005] The image data of the interior of the variable-section box girder to be inspected is collected by using the Zhenquancai acquisition equipment, and the brightness of the acquisition position is recorded in real time to generate an image set, which includes at least two images and the brightness corresponding to each image;

[0006] Performing image stitching processing on the image data in the image set to obtain a target image to be processed, and performing disease data analysis processing on the target image to be processed to determine the crack type and area of the variable-section box girder to be detected;

[0007] According to the crack type and area, the variable-section box girder to be inspected is displayed in three dimensions.

[0008] The beneficial effects of the method for detecting cracks in box beams based on Zhenquancai provided by the present invention are as follows:

[0009] First, the image data of the interior of the variable-section box girder is collected through the Zhenquancai acquisition equipment, and the brightness of the acquisition position is recorded in real time to generate an image set containing multiple images and corresponding brightness. This ensures that the collected images have clarity and details under different lighting conditions. Even in low-light environments, it can obtain high-quality image data by virtue of its hardware advantages such as the F1.0 large aperture and high-sensitivity sensor, as well as advanced ISP image processing technologies such as dynamic color gamut technology and 3D Color enhancement engine, providing a reliable foundation for subsequent analysis. Secondly, the image data in the image set is spliced to obtain the target image to be processed, and the disease data analysis is performed to accurately determine the type and area of cracks. This method based on image stitching and data analysis can fully cover the detection area of the variable-section box girder to avoid omissions, and the analysis results are more accurate, which helps to timely discover and locate diseases and provide a scientific basis for subsequent maintenance and reinforcement. Finally, a three-dimensional display based on crack type and area can intuitively present the damage status of the variable-section box girder, allowing technicians and decision makers to more clearly understand the distribution and severity of the damage, thereby formulating more reasonable maintenance strategies, improving work efficiency and the scientific nature of decision-making.

[0010] Based on the above solution, the present invention can also be improved as follows.

[0011] Furthermore, the process of collecting image data from the interior of the variable-section box girder to be inspected using the Zhenquan color acquisition equipment is as follows:

[0012] The Zhenquan color acquisition device is carried on a vehicle-carrying platform, and the vehicle-carrying platform is controlled to move along a preset track;

[0013] The moving distance of the vehicle-carrying platform is collected in real time, and when the moving distance exceeds a preset distance, a shooting instruction is generated. Through the shooting instruction, the Zhenquan color acquisition device is controlled to collect image data of the interior of the variable-section box girder to be inspected.

[0014] Furthermore, before controlling the Zhenquan color acquisition device to acquire image data of the interior of the variable-section box girder to be inspected through the shooting instruction, the method further includes:

[0015] When the shooting instruction is generated, the target brightness corresponding to the current position is collected, and it is determined whether the target brightness exceeds the preset brightness. If the determination result is no, fill light processing is performed.

[0016] Furthermore, the process of performing image stitching processing on the image data in the image set to obtain a target image to be processed is specifically as follows:

[0017] Performing grayscale processing on any image data to obtain a first image corresponding to the image data;

[0018] Extract feature points from any first image using a feature detection algorithm, generate feature descriptors corresponding to the first image using a feature descriptor extractor, and determine matching point pairs corresponding to the first image based on a matcher;

[0019] Through the projection mapping matrix and the matching point pairs, the target image to be processed is obtained after splicing and alignment.

[0020] Furthermore, the process of three-dimensionally displaying the variable-section box girder to be inspected is specifically as follows:

[0021] Build a scene, and use a renderer and a shader to draw the variable-section box girder to be tested;

[0022] Set the crack type, area, and brightness to complete the three-dimensional display.

[0023] 2) In the second aspect, the present invention also provides a box girder crack detection system based on Zhenquancai, and the specific technical solution is as follows:

[0024] The acquisition module is used to: acquire image data of the interior of the variable-section box girder to be inspected through the Zhenquancai acquisition device, and record the brightness of the acquisition position in real time to generate an image set, which includes at least two images and the brightness corresponding to each image;

[0025] The processing module is used to: perform image stitching processing on the image data in the image set to obtain a target image to be processed, and perform disease data analysis processing on the target image to be processed to determine the crack type and area of the variable-section box girder to be detected;

[0026] The display module is used to: perform three-dimensional display of the variable-section box girder to be inspected according to the crack type and area.

[0027] Based on the above solution, the present invention can also be improved as follows.

[0028] Furthermore, the process of collecting image data from the interior of the variable-section box girder to be inspected using the Zhenquan color acquisition equipment is as follows:

[0029] The Zhenquan color acquisition device is carried on a vehicle-carrying platform, and the vehicle-carrying platform is controlled to move along a preset track;

[0030] The moving distance of the vehicle-carrying platform is collected in real time, and when the moving distance exceeds a preset distance, a shooting instruction is generated. Through the shooting instruction, the Zhenquan color acquisition device is controlled to collect image data of the interior of the variable-section box girder to be inspected.

[0031] Furthermore, before controlling the Zhenquan color acquisition device to acquire image data of the interior of the variable-section box girder to be inspected through the shooting instruction, the method further includes:

[0032] When the shooting instruction is generated, the target brightness corresponding to the current position is collected, and it is determined whether the target brightness exceeds the preset brightness. If the determination result is no, fill light processing is performed.

[0033] Furthermore, the process of performing image stitching processing on the image data in the image set to obtain a target image to be processed is specifically as follows:

[0034] Performing grayscale processing on any image data to obtain a first image corresponding to the image data;

[0035] Extract feature points from any first image using a feature detection algorithm, generate feature descriptors corresponding to the first image using a feature descriptor extractor, and determine matching point pairs corresponding to the first image based on a matcher;

[0036] Through the projection mapping matrix and the matching point pairs, the target image to be processed is obtained after splicing and alignment.

[0037] Furthermore, the process of three-dimensionally displaying the variable-section box girder to be inspected is specifically as follows:

[0038] Build a scene, and use a renderer and a shader to draw the variable-section box girder to be tested;

[0039] Set the crack type, area, and brightness to complete the three-dimensional display.

[0040] 3) In a third aspect, the present invention further provides an electronic device, comprising a processor, wherein the processor is coupled to a memory, wherein at least one computer program is stored in the memory, and the at least one computer program is loaded and executed by the processor so that the electronic device implements any of the above methods.

[0041] 4) In a fourth aspect, the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores at least one computer program, and the at least one computer program is loaded and executed by a processor to enable a computer to implement any of the above methods.

[0042] It should be noted that the beneficial effects achieved by the technical solutions of the second to fourth aspects of the present invention and the corresponding possible implementation methods can be found in the above-mentioned technical effects of the first aspect and its corresponding possible implementation methods, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0044] Figure 1 This is one of the flow charts of a method for detecting cracks in a box girder based on Zhenquancai in an embodiment of the present invention;

[0045] Figure 2 This is a second flow chart of a method for detecting cracks in a box girder based on Zhenquancai according to an embodiment of the present invention;

[0046] Figure 3 Schematic diagram of the structure of a method for detecting cracks in a box girder based on Zhenquancai according to an embodiment of the present invention;

[0047] Figure 4 This is a structural framework diagram of an electronic device of the present invention. DETAILED DESCRIPTION

[0048] To make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0049] like Figure 1 As shown, a method for detecting cracks in a box beam based on Zhenquancai according to an embodiment of the present invention includes the following steps:

[0050] S1, using the Zhenquan color acquisition equipment, collect image data of the interior of the variable-section box girder to be inspected, and record the brightness of the acquisition position in real time to generate an image set, which includes at least two images and the brightness corresponding to each image;

[0051] S2, performing image stitching processing on the image data in the image set to obtain a target image to be processed, and performing disease data analysis processing on the target image to be processed to determine the crack type and area of the variable-section box girder to be detected;

[0052] S3, performing a three-dimensional display of the variable-section box girder to be inspected according to the crack type and area.

[0053] The beneficial effects of the method for detecting cracks in box beams based on Zhenquancai provided by the present invention are as follows:

[0054] First, the image data of the interior of the variable-section box girder is collected through the Zhenquancai acquisition equipment, and the brightness of the acquisition position is recorded in real time to generate an image set containing multiple images and corresponding brightness. This ensures that the collected images have clarity and details under different lighting conditions. Even in low-light environments, it can obtain high-quality image data by virtue of its hardware advantages such as the F1.0 large aperture and high-sensitivity sensor, as well as advanced ISP image processing technologies such as dynamic color gamut technology and 3D Color enhancement engine, providing a reliable foundation for subsequent analysis. Secondly, the image data in the image set is spliced to obtain the target image to be processed, and the disease data analysis is performed to accurately determine the type and area of cracks. This method based on image stitching and data analysis can fully cover the detection area of the variable-section box girder to avoid omissions, and the analysis results are more accurate, which helps to timely discover and locate diseases and provide a scientific basis for subsequent maintenance and reinforcement. Finally, a three-dimensional display based on crack type and area can intuitively present the damage status of the variable-section box girder, allowing technicians and decision makers to more clearly understand the distribution and severity of the damage, thereby formulating more reasonable maintenance strategies, improving work efficiency and the scientific nature of decision-making.

[0055] It should be noted that when collecting internal image data in this solution, in order to ensure the clarity of the collected image data, it is necessary to confirm the surrounding brightness during collection in real time, and when the brightness is lower than the preset brightness, fill light processing is performed in time to make the features in the image data clear.

[0056] The specific process of fill light processing is as follows:

[0057] The target light intensity and target color temperature corresponding to the current position are determined by a photosensitive sensor and a CMOS or CCD, and the target distance between the current position and the end section of the variable-section box girder to be detected is determined. The fill light method includes at least:

[0058] The first method is to calculate the color temperature weight and light intensity weight through a preset weight function, and calculate the optimal fill light parameters based on the calculated weight values in combination with a preset model or algorithm.

[0059] The specific process of calculating the color temperature weight and light intensity weight using the preset weight function is as follows:

[0060] Determine the ratio of the target distance to the length of the variable cross-section box girder to be tested.

[0061] 1. Define the weight function

[0062] Color temperature weight function: The expression of color temperature weight function g(c) is:

[0063] g(c)=d|c-c1|+e

[0064] Where: c is the color temperature of the current environment. c1 is the neutral color temperature (e.g., 5500K, typically used for daylight environments). d and e are constants that control the speed and initial value of the weight change when the color temperature deviates from the neutral value.

[0065] Light intensity weight function: The expression of light intensity weight function f(l) is:

[0066] f(l)=a·(ll t )+b

[0067] Where: l is the light intensity value of the current environment. t is the preset light intensity threshold. a and b are constants that control the weight growth rate and initial value respectively when the light intensity is less than the threshold.

[0068] 2. Input parameters to calculate weights

[0069] Calculate the color temperature weight: Input the detected color temperature value c into the color temperature weight function g(c) to calculate the color temperature weight ω c :

[0070] ω c =g(c)=d·|c-c1|+e

[0071] Calculate the light intensity weight: Input the detected light intensity value l into the light intensity weight function f(l) to calculate the light intensity weight ω l :

[0072] ω l =f(l))=a·(ll t )+b

[0073] 3. Adjust weight parameters

[0074] According to the actual application scenario, the constants d, e, a, b and other parameters can be adjusted through experiments or experience to ensure that the weight function can reasonably reflect the influence of color temperature and light intensity on the fill light effect.

[0075] 4. Apply weights

[0076] The calculated color temperature weight ω c and light intensity weight ω l This is used in subsequent fill light parameter adjustments, such as calculating the compensation area size or selecting the optimal fill light image through a linear model.

[0077] The second method is to use a combination of natural light and equipment-based lighting to perform fill-in lighting. That is, a first reflector is set at each end of the variable-section box girder to be inspected, and the first reflector is adjusted in real time so that it can refract sunlight into the interior of the variable-section box girder to be inspected or onto a second reflector placed on the vehicle platform. The second reflector is used to fill in the interior of the variable-section box girder to be inspected. In real time, it is determined whether the brightness of the sunlight reflected by the first reflector or the second reflector meets the preset brightness requirement. If so, the fill-in lighting process is completed. If not, the fill-in lighting process is continued through the equipment-based fill-in lighting method. The equipment-based fill-in lighting method is the first fill-in lighting method.

[0078] The second method makes full use of natural light for fill lighting, which can effectively reduce dependence on equipment fill lighting and reduce energy consumption. When natural light fill lighting still cannot meet the requirements, it can be further supplemented by combining it with equipment fill lighting. This combination method not only makes full use of the advantages of natural light, but also makes up for the shortcomings of natural light through equipment fill lighting, and can respond more flexibly to different lighting conditions.

[0079] The image captured after fill light processing may have over-fill light (brightness exceeds the maximum value of the preset brightness range) due to the difference between the fill light area and the actual position, or the fill light requirements may not be met due to the fill light meeting the requirements in some areas but some areas are still blocked, or the shooting area is completely blocked, that is, fill light is not achieved.

[0080] When any target image data has any of the above three situations, the target image is first preprocessed through the preset model to obtain the initial crack area of the target image and the initial probability corresponding to each initial crack area, and the above data is substituted into the subsequent image stitching and analysis process. That is, if the area corresponding to the target image in the stitched target image to be processed cannot be analyzed and identified, the output result of the preset model is directly output as the final analysis and identification result.

[0081] Because the images the preset model recognizes are quite specialized, its final output must be combined with the output of the large language model. To ensure the preset model achieves the required accuracy, the training process requires careful attention to the diversity of the training set and a sufficient number of iterations to ensure the preset model meets the required accuracy.

[0082] The specific process of judging whether the training set meets the diversity requirement is as follows:

[0083] According to the above three situations, it is determined whether the number of training samples in each situation reaches the preset number. If the number of training samples corresponding to any situation does not reach the preset number, the number of training samples that does not reach the preset number is expanded by expansion processing.

[0084] The specific process of expansion processing is as follows:

[0085] The training sample in this case is defined as the first sample. The feature frame in each first sample is identified, and the feature frame is moved to any position in the first sample by random transposition to generate a new first sample.

[0086] After the target image is input into the preset model, the preset model will give priority to judging the target image and determining whether a compensation strategy needs to be executed. That is, when the target image meets the fill-in requirements for some areas but some areas are still blocked and fail to meet the fill-in requirements, and the area of the area that meets the requirements is less than 10% of the area of the entire target image, it is determined that a compensation strategy needs to be executed. The compensation strategy refers to adjusting the contrast, sharpness, brightness and saturation corresponding to the target image, and after one adjustment, continuing to judge whether the target image can identify valid features. If so, the valid features are used as the output results of the preset model. Valid features refer to obtaining the sound wave signal graph of the position corresponding to the target image through sound wave detection, processing the sound wave signal graph through a large language model, and obtaining a processing result. If the processing result shows that there is a crack at the position, and the crack feature can be identified in the target image after one adjustment, it means that the crack feature is a valid feature. If no valid features can be identified, the target distance is used to determine whether the location is a high-incidence area for cracks. If not, and the processing results of the large language model also indicate that there are no cracks at this location, the target image is not processed further. If the processing results of the large language model indicate that there are cracks at this location, the target image is annotated, and when the target image to be processed is obtained by splicing, the target image is combined with its adjacent images in the target image to determine whether there are cracks at the corresponding location of the target image. This is then output as the final result after combining the output of the large language model.

[0087] The specific process of adjusting the contrast, sharpness, brightness and saturation of the target image is as follows:

[0088] Brightness adjustment:

[0089] Calculate Current Brightness: Evaluates the brightness of the current image by calculating the average pixel value of the image or using histogram analysis.

[0090] Adjust Luminance: Adjusts the luminance of the image using a linear or nonlinear transformation based on a target luminance value.

[0091] Linear Transformation: Changes the overall brightness of the image by adjusting the brightness offset.

[0092] Histogram equalization: By adjusting the histogram distribution of the image, the brightness distribution of the image is made more uniform, thereby improving the contrast and brightness of the image.

[0093] Contrast adjustment:

[0094] Calculate current contrast: Evaluate the contrast of the current image by calculating the pixel value range (maximum and minimum values) of the image or using contrast indicators such as standard deviation.

[0095] Adjust contrast: Use linear or nonlinear transformation to adjust the contrast of an image based on the target contrast value. For example, linear transformation: Increases the contrast of an image by adjusting the pixel value range.

[0096] Non-linear transformation: Use methods such as gamma correction or logarithmic transformation to non-linearly adjust the contrast of the image to enhance the details of the image.

[0097] Sharpness adjustment:

[0098] Calculate current sharpness: Evaluate the sharpness of the current image by calculating the edge strength of the image or using a sharpness indicator (such as the Laplacian operator).

[0099] Adjust sharpness: Use edge enhancement algorithms (such as high-pass filtering, Laplace filtering, or bilateral filtering) to enhance the edge details of the image, thereby improving the sharpness of the image. For example:

[0100] Laplacian filtering: Use the Laplacian operator to detect the edges of the image and enhance the sharpness of the image by superimposing edge information.

[0101] Bilateral filtering: While enhancing sharpness, it preserves the texture details of the image and avoids excessive blurring.

[0102] Saturation adjustment:

[0103] Calculate the current saturation: Calculate the saturation of the current image by converting the image from the RGB color space to the HSV (Hue, Saturation, Value) color space.

[0104] Adjust saturation: According to the target saturation value, the saturation component in the HSV color space is adjusted to enhance the color saturation of the image. For example:

[0105] Linear adjustment: Enhances the color saturation of the image by increasing the value of the saturation component.

[0106] Non-linear adjustment: Use curve adjustment tools (such as the S-curve) to make non-linear adjustments to saturation to enhance the color contrast of the image.

[0107] The large language model in this application can output a conclusion on whether there are cracks or internal damage corresponding to the input sound wave signal graph through semantic analysis and other methods. The large language model can directly process the input sound wave signal graph to generate a first processing result, and output the first processing result as the final result; or after generating the first processing result, it can search the historical database for the historical sound wave signal graph corresponding to the target historical box girder with the highest similarity to the target material, target length, and target cross-sectional shape and target cross-sectional area of the chamber (the hollow in the middle of the box girder) of the variable-section box girder to be tested as a reference, and fine-tune the first processing result according to the corresponding weight to obtain a second processing result, and output the second processing result as the final result.

[0108] Based on the target material, multiple first target historical box girders with the same target material are searched. Among all the first target historical box girders, multiple second target historical box girders with the same length as the target or within a preset length range based on the target length are searched. Among the multiple second target historical box girders, a third target historical box girder with the same or similar target cross-sectional shape or target cross-sectional area is searched and used as the target historical box girder.

[0109] Furthermore, the process of collecting image data from the interior of the variable-section box girder to be inspected using the Zhenquan color acquisition equipment is as follows:

[0110] The Zhenquan color acquisition device is carried on a vehicle-carrying platform, and the vehicle-carrying platform is controlled to move along a preset track;

[0111] The moving distance of the vehicle-carrying platform is collected in real time, and when the moving distance exceeds a preset distance, a shooting instruction is generated. Through the shooting instruction, the Zhenquan color acquisition device is controlled to collect image data of the interior of the variable-section box girder to be inspected.

[0112] Furthermore, before controlling the Zhenquan color acquisition device to acquire image data of the interior of the variable-section box girder to be inspected through the shooting instruction, the method further includes:

[0113] When the shooting instruction is generated, the target brightness corresponding to the current position is collected, and it is determined whether the target brightness exceeds the preset brightness. If the determination result is no, fill light processing is performed.

[0114] Furthermore, the process of performing image stitching processing on the image data in the image set to obtain a target image to be processed is specifically as follows:

[0115] Performing grayscale processing on any image data to obtain a first image corresponding to the image data;

[0116] Extract feature points from any first image using a feature detection algorithm, generate feature descriptors corresponding to the first image using a feature descriptor extractor, and determine matching point pairs corresponding to the first image based on a matcher;

[0117] Through the projection mapping matrix and the matching point pairs, the target image to be processed is obtained after splicing and alignment.

[0118] Furthermore, the process of three-dimensionally displaying the variable-section box girder to be inspected is specifically as follows:

[0119] Build a scene, and use a renderer and a shader to draw the variable-section box girder to be tested;

[0120] Set the crack type, area, and brightness to complete the three-dimensional display.

[0121] Example 1, as Figure 2 As shown, after the inspection personnel quickly assemble the collection equipment, they push the cart platform to open the collection equipment to collect data, and the data is stored in the control storage device. The specific implementation steps are as follows:

[0122] 1) After the inspector arrives at the inspection location, the collection equipment is powered on, and the new collection project is completed, the inspector manually pushes the vehicle platform forward;

[0123] 2) As the vehicle platform moves forward, the encoder sends a pulse signal to the control storage device, which is the mileage information;

[0124] 3) After the control storage device receives the mileage information, it sends the photo information to the image acquisition device; it should be noted that the hardware-controlled photo taking is controlled by the control storage device sending high and low level trigger signals to the connected camera to control its photo taking. After the control storage device receives the specified mileage information, it sends a hardware trigger signal to the camera, and the camera takes a photo after receiving the hardware trigger signal. The control storage device can send trigger signals to multiple cameras at the same time, so each camera can take a photo at the same time and under the same mileage conditions. Sending multiple trigger signals can complete the continuous photo taking function. Cameras that meet different mileage requirements need to send hardware trigger signals intermittently according to the specified mileage information through two interfaces.

[0125] 4) After receiving the photo information, the image acquisition device takes a photo. If the current ambient illumination is lower than 0.0005 lux, the camera automatically turns on the fill light to improve the image quality. Regardless of the illumination level, the acquired image is a color image.

[0126] 5) Finally, the control storage device stores the mileage data, color image data, and project-related information. The project-related information includes the road section number, bridge name, collection time, starting and ending pile number information, bridge width, and detection direction, which need to be filled in before data collection.

[0127] like Figure 3 As shown in the figure, the image acquisition device is composed of multiple full-color cameras. The control and storage device is composed of key components such as the acquisition mainboard, power module, and synchronization control board. The vehicle carrier is composed of a push-type vehicle carrier, encoder, and battery. The image acquisition device is responsible for acquiring the apparent high-definition color images of the interior of the continuously variable cross-section box girder. The control and storage device is responsible for controlling the cameras to take photos, storing images and mileage data, and recording relevant engineering data. The vehicle carrier is responsible for powering the entire system, acting as a carrier for the acquisition equipment, and emitting encoder pulse signals.

[0128] The image acquisition system features multiple full-color cameras, each with its lenses facing outward, arranged in a semicircular configuration. The captured images cover nearly 270 degrees of the box girder's internal cross-section. This integrated system avoids the fragmented and inaccurate defect location associated with manual inspections, while also improving image quality and facilitating post-processing.

[0129] The camera's automatic soft fill light feature alleviates the operational issues previously encountered during manual inspections. Previously, manual inspections required holding a flashlight and a camera in one hand while taking photos of the internal surface of the box girder. Now, inspectors can simply push the platform forward to continuously capture photos, eliminating the need for additional fill light operations.

[0130] 1) The camera automatically recognizes the brightness of the current environment

[0131] The camera automatically identifies the brightness of the current environment. The specific implementation steps are as follows:

[0132] (1) Equipped with a light sensor: The light sensor can sense the light intensity in the environment. When the ambient brightness drops to a certain level, it will send a signal to the camera's control system. This light sensor uses a photodiode or other light-sensitive element to convert light intensity into an electrical signal. For example, in a bright environment, the light sensor receives a lot of light, and the electrical signal generated indicates that the ambient brightness is high; in a dark environment, the light received is small, and the electrical signal generated reflects that the ambient brightness is low.

[0133] (2) Preset illumination threshold: When the ambient illumination is lower than 0.0005 Lux, the fill light will turn on. During operation, the ambient illumination detected by the light sensor is compared with this preset threshold. If the detected illumination value is lower than 0.0005 Lux, the control system determines that the fill light needs to be turned on to ensure the presentation of the color image; if the ambient illumination is greater than or equal to 0.0005 Lux, the fill light does not need to be turned on, and the camera's own photosensitivity and imaging technology can maintain the color image.

[0134] (3) Filter time setting: In order to prevent the fill light from turning on and off frequently due to the instantaneous changes in the light in the environment, a filter time can be set. During this time, the camera will not be disturbed by the ambient light. That is, the camera can have a built-in light intensity sensor that samples the ambient light intensity at a fixed time interval (i.e., the "filter time"). By calculating the average light intensity over several consecutive sampling cycles, the impact of instantaneous light changes on the fill light control can be reduced; the collected light intensity data can be filtered through digital signal processing technology, such as using a low-pass filter to reduce the impact of high-frequency noise and instantaneous changes, thereby avoiding the frequent switching of the fill light; a light intensity threshold is set in the camera's firmware or software. The fill light will only be activated when the average light intensity remains below this threshold for a certain period of time (filter time).

[0135] 2) Image stitching

[0136] In order to better process and utilize the stored image data, image stitching is required. The images of the interior of the continuously variable cross-section box girder with a depth of nearly 270 degrees captured by four cameras are stitched together into a complete image at intervals of 10 meters. The specific steps for image stitching are as follows:

[0137] (1) Image reading and preprocessing: First, read each image and then convert each image into grayscale to extract feature points.

[0138] (2) Feature point extraction and matching: Use feature detection algorithms (such as SurfFeatureDetector and OrbFeatureDetector) to extract key points in the image, and then generate feature descriptors through feature descriptor extractors (such as SurfDescriptorExtractor and OrbDescriptorExtractor). Next, use a matcher (such as FlannBasedMatcher) to match feature points and find matching point pairs.

[0139] (3) Matrix calculation and image transformation: After matching is complete, the homography matrix needs to be calculated. Use the findHomography function to calculate the projection mapping matrix between images. This matrix describes the perspective transformation relationship from one image to another. Use the warpPerspective function to transform the images according to the projection mapping matrix so that they can be aligned when stitching.

[0140] (4) Image fusion and output: Use weighted averaging or other image fusion algorithms to eliminate obvious gaps at the splicing points, making the spliced image more natural. Finally, output the spliced image as a JPEG file.

[0141] 3) Data Generation

[0142] After the data is processed, it is necessary to export the corresponding test report, which includes information such as the location, type, and size of the disease. This requires the software to have an automatic generation function, which is implemented as follows:

[0143] (1) After data processing is completed, click the "Export Report" button;

[0144] (2) Pop up the storage path, that is, the target address disk, and select the drive letter such as F drive;

[0145] (3) The stored data includes: disease pictures, inspection reports, etc.

[0146] Import the collected data images into the processing software. The processing software will splice the images into a long image in the order of collection. Starting from the first collected photo, the entire long image is marked with the total mileage, and the mileage information with the image as the background can be obtained. When processing the software, manually check the entire image, select the disease, and mark the disease type after the selection. Combined with the mileage information on the image, all information about the disease can be obtained in the exported data.

[0147] 4) 3D display

[0148] The three-dimensional display is to help inspectors understand the location of the disease intuitively and clearly, and to facilitate subsequent maintenance by combining it with the inspection report. The specific implementation of the three-dimensional display is as follows:

[0149] (1) Rendering based on Three.js technology requires first defining the scene, defining lighting and camera effects within the scene, then creating a mesh for the model object, selecting the corresponding material information, and extracting the corresponding vertex shader and fragment shader. Finally, the Three.js renderer draws the final graphics on the screen based on the relevant data in the scene and the selected shader program.

[0150] (2) The accuracy of the disease data information requires the reflection of the details of the panoramic map. The role of lighting cannot be ignored. Without a lighting model, the details of the map will be covered. Therefore, the lighting settings can be made in the modeling software in advance to provide data reference for the lighting settings in the web page model display scene. The specific process of setting the lighting is as follows:

[0151] ① Preset lighting model: In 3D modeling software, preset different types and intensities of lights according to scene requirements to simulate the lighting effects in the actual environment. This can better show the map details when rendering the model;

[0152] ② Lighting texture materials: Use specialized lighting texture materials that can help simulate light reflection and shadow effects in the real world, enhancing the realism of the textures.

[0153] ③ Intelligent lighting control algorithm: Apply intelligent lighting control algorithms, such as PID control algorithm, fuzzy control algorithm, etc., to achieve precise adjustment and intelligent control of lighting, and improve the energy-saving effect and intelligence level of the lighting control system.

[0154] (3) By uploading the model to the database, the continuous variable cross-section box girder model can be loaded on the web page. However, the display effect of the model without the map is not ideal. The model has poor realism and cannot be interactive. It is necessary to continue adding MTL files to improve the display scene. The process of improving the display scene is as follows:

[0155] ① Ensure file matching: Place the model file (OBJ) and MTL file in the same folder, and ensure that the file names of the MTL file match those of the OBJ file so that the model file can correctly reference the MTL file;

[0156] ② Import the model and material: In 3D modeling software, such as Blender, Maya or 3ds Max, after importing the OBJ file, find the material option in the software's material editor. There is usually a button called "Import Material" or "Associate MTL File". Click it to associate the MTL file.

[0157] ③ Check material application: Check whether the material of the MTL file is correctly applied to the model in the 3D view, and ensure that each surface shows the expected material effect;

[0158] ④ Adjust and edit materials: Further adjust and edit the model's materials as needed, and change the material's color, texture, lighting, and reflection properties.

[0159] ⑤Use 3D modeling software to import MTL: In 3D MAX, you can import OBJ files through the "File" menu, and then import the MTL material file in the material editor. 3D MAX will automatically apply the MTL material file to the corresponding model;

[0160] ⑥ Import MTL into Maya: In Maya, after importing the OBJ file, enter the "Attribute Editor" window, create a new material in the "Material" tab, and select the MTL file. Maya will automatically apply the material to the OBJ model.

[0161] (4) After the scene of the continuously variable cross-section box girder model is loaded, the visualization and three-dimensional interaction of the continuously variable cross-section box girder model on the web page are finally realized through the data attribute association method and the model object coordinate matching method. Clicking the component object of the model with the mouse will display the object's attribute information on the left side of the window.

[0162] In the above embodiments, although the steps are numbered S1, S2, etc., these are only specific embodiments given by the present invention. Those skilled in the art may adjust the execution order of S1, S2, etc. according to actual conditions, which is also within the scope of protection of the present invention. It can be understood that in some embodiments, some or all of the above embodiments may be included.

[0163] The present invention also provides a box girder crack detection system based on Zhenquancai, and the specific technical solution is as follows:

[0164] The acquisition module is used to: acquire image data of the interior of the variable-section box girder to be inspected through the Zhenquancai acquisition device, and record the brightness of the acquisition position in real time to generate an image set, which includes at least two images and the brightness corresponding to each image;

[0165] The processing module is used to: perform image stitching processing on the image data in the image set to obtain a target image to be processed, and perform disease data analysis processing on the target image to be processed to determine the crack type and area of the variable-section box girder to be detected;

[0166] The display module is used to: perform three-dimensional display of the variable-section box girder to be inspected according to the crack type and area.

[0167] It should be noted that the beneficial effects of the crack detection system for box girders based on Zhenquancai provided by the above embodiment are the same as the beneficial effects of the crack detection method for box girders based on Zhenquancai, which will not be repeated here. In addition, when the system provided by the above embodiment realizes its functions, it only uses the division of the above functional modules as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the system can be divided into different functional modules according to actual conditions to complete all or part of the functions described above. In addition, the system and method embodiments provided by the above embodiment belong to the same concept, and their specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0168] like Figure 4 As shown, an electronic device 300 according to an embodiment of the present invention includes a processor 320, which is coupled to a memory 310. The memory 310 stores at least one computer program 330. The at least one computer program 330 is loaded and executed by the processor 320 to enable the electronic device 300 to implement any of the above methods. Specifically:

[0169] The electronic device 300 may have relatively large differences due to different configurations or performances, and may include one or more processors 320 (Central Processing Units, CPU) and one or more memories 310, wherein the one or more memories 310 store at least one computer program 330, and the at least one computer program 330 is loaded and executed by the one or more processors 320 to enable the electronic device 300 to implement a method for detecting cracks in a box girder based on Zhenquancai provided in the above embodiment. Of course, the electronic device 300 may also have components such as a wired or wireless network interface, a keyboard, and an input / output interface for input and output. The electronic device 300 may also include other components for implementing device functions, which will not be described in detail here.

[0170] A computer-readable storage medium according to an embodiment of the present invention stores at least one computer program, and the at least one computer program is loaded and executed by a processor to enable a computer to implement any of the above methods.

[0171] Alternatively, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc (CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, or the like.

[0172] In an exemplary embodiment, a computer program product or computer program is also provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform any of the above methods.

[0173] It should be noted that the terms "first," "second," and the like in the specification and claims of this application are used to distinguish similar objects and to define a specific order or precedence. Where appropriate, the order used for similar objects may be interchanged, such that the embodiments of the present application described herein can be implemented in an order other than the order shown or described.

[0174] Those skilled in the art will appreciate that the present invention may be implemented as a system, method, or computer program product. Therefore, the present disclosure may be implemented in the following forms: entirely in hardware, entirely in software (including firmware, resident software, microcode, etc.), or in a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, the present invention may be implemented in the form of a computer program product embodied in one or more computer-readable media containing computer-readable program code.

[0175] Any combination of one or more computer-readable media can be used. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or device.

[0176] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A method for detecting cracks in box beams based on Zhenquancai, characterized in that: include: The image data of the interior of the variable-section box girder to be inspected is collected by using the Zhenquancai acquisition equipment, and the brightness of the acquisition position is recorded in real time to generate an image set, which includes at least two images and the brightness corresponding to each image; Performing image stitching processing on the image data in the image set to obtain a target image to be processed, and performing disease data analysis processing on the target image to be processed to determine the crack type and area of the variable-section box girder to be detected; According to the crack type and area, the variable-section box girder to be inspected is displayed in three dimensions.

2. The method for detecting cracks in box beams based on Zhenquancai according to claim 1, characterized in that: The process of collecting image data from the interior of the variable-section box girder to be inspected using the Zhenquan color acquisition equipment is as follows: The Zhenquan color acquisition device is carried on a vehicle-carrying platform, and the vehicle-carrying platform is controlled to move along a preset track; The moving distance of the vehicle-carrying platform is collected in real time, and when the moving distance exceeds a preset distance, a shooting instruction is generated. Through the shooting instruction, the Zhenquan color acquisition device is controlled to collect image data of the interior of the variable-section box girder to be inspected.

3. The method for detecting cracks in box beams based on Zhenquancai according to claim 2, characterized in that: Before controlling the Zhenquancai acquisition device to acquire image data of the interior of the variable-section box girder to be inspected through the shooting instruction, the method further includes: When the shooting instruction is generated, the target brightness corresponding to the current position is collected, and it is determined whether the target brightness exceeds the preset brightness. If the determination result is no, fill light processing is performed.

4. The method for detecting cracks in box beams based on Zhenquancai according to claim 1, characterized in that: The specific process of performing image stitching processing on the image data in the image set to obtain a target image to be processed is as follows: Performing grayscale processing on any image data to obtain a first image corresponding to the image data; Extract feature points from any first image using a feature detection algorithm, generate feature descriptors corresponding to the first image using a feature descriptor extractor, and determine matching point pairs corresponding to the first image based on a matcher; Through the projection mapping matrix and the matching point pairs, the target image to be processed is obtained after splicing and alignment.

5. The method for detecting cracks in box beams based on Zhenquancai according to claim 1, characterized in that: The specific process of three-dimensionally displaying the variable-section box girder to be inspected is as follows: Build a scene, and use a renderer and a shader to draw the variable-section box girder to be tested; Set the crack type, area, and brightness to complete the three-dimensional display.

6. A box girder crack detection system based on Zhenquancai, characterized in that: include: The acquisition module is used to: acquire image data of the interior of the variable-section box girder to be inspected through the Zhenquancai acquisition device, and record the brightness of the acquisition position in real time to generate an image set, which includes at least two images and the brightness corresponding to each image; The processing module is used to: perform image stitching processing on the image data in the image set to obtain a target image to be processed, and perform disease data analysis processing on the target image to be processed to determine the crack type and area of the variable-section box girder to be detected; The display module is used to: perform three-dimensional display of the variable-section box girder to be inspected according to the crack type and area.

7. The box girder crack detection system based on Zhenquancai according to claim 6 is characterized in that: The process of collecting image data from the interior of the variable-section box girder to be inspected using the Zhenquan color acquisition equipment is as follows: The Zhenquan color acquisition device is carried on a vehicle-carrying platform, and the vehicle-carrying platform is controlled to move along a preset track; The moving distance of the vehicle-carrying platform is collected in real time, and when the moving distance exceeds a preset distance, a shooting instruction is generated. Through the shooting instruction, the Zhenquan color acquisition device is controlled to collect image data of the interior of the variable-section box girder to be inspected.

8. The box girder crack detection system based on Zhenquancai according to claim 7 is characterized in that: Before controlling the Zhenquancai acquisition device to acquire image data of the interior of the variable-section box girder to be inspected through the shooting instruction, the method further includes: When the shooting instruction is generated, the target brightness corresponding to the current position is collected, and it is determined whether the target brightness exceeds the preset brightness. If the determination result is no, fill light processing is performed.

9. An electronic device, characterized in that: The electronic device includes a processor coupled to a memory, wherein the memory stores at least one computer program, and the at least one computer program is loaded and executed by the processor so that the electronic device implements the method according to any one of claims 1 to 5.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores at least one computer program, which is loaded and executed by a processor to enable a computer to implement the method according to any one of claims 1 to 5.

Images