Tunnel crack identification method, system, electronic device and storage medium based on laser radar and drone photography

Abstract

The present invention discloses a tunnel crack identification method, system, electronic device and storage medium based on laser radar and drone photography. The method includes: acquiring spatial geometric information, photographing surface image information inside the tunnel, and constructing a three-dimensional texture point cloud model; extracting geometric morphological features and surface texture features of the three-dimensional texture point cloud model to obtain a preliminary crack feature set; classifying and preliminarily screening the texture features inside the tunnel based on the preliminary crack feature set to obtain a preliminary screening crack region set; expanding the crack region based on the preliminary screening crack region to generate an expanded crack region set; analyzing the point cloud density within the expanded crack region set through a Gaussian mixture model to determine the depth and width characteristics of the crack and obtain a refined screening crack feature set; and using a probability optimization algorithm based on a Markov random field for the refined screening crack feature set to perform joint probability distribution modeling on the geometric and texture features of the crack to obtain a crack identification result.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel crack detection, and in particular to a tunnel crack identification method, system, electronic device and storage medium based on laser radar and drone photography. Background Art

[0002] The safety and durability of tunnel engineering are crucial research areas in the infrastructure field, directly related to the smooth flow of transportation and the safety of public life and property. During the service life of tunnels, cracks, as a common disease, not only affect structural stability but may also induce larger-scale damage. Therefore, the timely identification and screening of cracks have become an indispensable technical link to ensure the long-term operation of tunnels. Currently, methods based on traditional manual inspections or single-sensor detection have exposed obvious shortcomings in practical applications. For example, manual inspections are inefficient and easily interfered with by subjective factors, while single sensors such as optical cameras or laser equipment have difficulty in fully capturing the geometric and texture characteristics of cracks due to their single data dimension, resulting in limited recognition accuracy or a high misjudgment rate.

[0003] Despite the continuous advancement of technological means in recent years, tunnel crack identification still faces significant challenges. The most core technical factors focus on the effective fusion of multi-source data and the accuracy of the screening mechanism. Since lidar is good at acquiring spatial geometric information, and high-definition images taken by drones are more suitable for providing surface texture details, how to achieve efficient matching and fusion of these two heterogeneous data in the spatiotemporal dimension has become a key problem in improving recognition capabilities. In addition, the diversity and complexity of crack characteristics make it difficult for a single threshold screening strategy to adapt to different scenarios, which can easily lead to missed or false alarms. How to design a hierarchical and progressive screening mechanism to balance efficiency and accuracy further exacerbates the difficulty of technical implementation. These unresolved technical factors directly limit the reliability and practicality of crack identification, and a systematic solution is urgently needed.

[0004] Therefore, how to design a multi-level screening mechanism based on the fusion of lidar spatial data and texture information captured by drones, and gradually optimize the accuracy and robustness of crack identification through the collaborative work of initial screening, fine screening and manual confirmation, has become a key issue that needs to be overcome in this study. Summary of the Invention

[0005] In order to solve the above technical problems, the present invention provides a tunnel crack identification method based on laser radar and drone photography, the method comprising:

[0006] Using a laser radar to obtain spatial geometric information, using a drone to capture surface image information of the interior of the tunnel, and constructing a three-dimensional textured point cloud model based on the spatial geometric information and the surface image information;

[0007] Extracting geometric features and surface texture features of the three-dimensional texture point cloud model to obtain a preliminary crack feature set;

[0008] Classifying the internal texture features of the tunnel based on the preliminary crack feature set to obtain a classified crack feature subset, and using a threshold segmentation method to preliminarily screen high-risk crack areas in the classified crack feature subset to obtain a preliminarily screened crack area set;

[0009] Acquiring the curvature and texture contrast of boundary features from the initially screened crack region set, and expanding the crack region to generate an expanded crack region set;

[0010] If the curvature change trend of the expanded fracture area set matches the preset fracture morphology template, the point cloud density within the expanded fracture area set is analyzed by a Gaussian mixture model to determine the depth and width characteristics of the fractures and obtain a fine-screened fracture feature set;

[0011] For the finely screened crack feature set, a probability optimization algorithm based on Markov random fields is used to perform joint probability distribution modeling on the geometric and texture features of the cracks to obtain crack recognition results.

[0012] Preferably, the method for constructing a three-dimensional texture point cloud model includes:

[0013] Acquire point cloud data using a laser radar, extract spatial geometric features from the point cloud data using a volume geometry algorithm, and obtain a three-dimensional point cloud coordinate set;

[0014] The two-dimensional texture data of the tunnel interior is captured by a drone, and the surface details are separated using an image segmentation algorithm to obtain a texture feature set.

[0015] Using a preset coordinate system mapping algorithm, the three-dimensional point cloud coordinate set and the texture feature set are aligned in time and space dimensions, and the coordinate mapping is adjusted by an iterative closest point algorithm to obtain an optimized matching data set;

[0016] Based on the optimized matching data set, the aligned point cloud coordinate set and the texture feature set are fused to generate a three-dimensional texture point cloud model.

[0017] Preferably, the method for obtaining the preliminary crack feature set includes:

[0018] Acquire crack geometry data based on the three-dimensional texture point cloud model to obtain an initial point cloud data set;

[0019] extracting crack geometry and surface texture from the initial point cloud dataset using stereo microscopy technology to obtain a separated feature set;

[0020] Determining the degree of overlap between crack binding and morphological features based on the separated feature set to determine the geometric morphological features;

[0021] Separating the surface texture from the geometric features to obtain the surface texture features;

[0022] The preliminary crack feature set is constructed based on the geometric features and the surface texture features.

[0023] Preferably, the method for obtaining the primary screening crack region set includes:

[0024] If the number of geometric feature points in the preliminary crack feature set exceeds a first preset threshold, a deep convolutional neural network is used to classify the internal texture features of the tunnel to determine the potential risk level of the cracks, thereby obtaining the classified crack feature subset;

[0025] Processing the fissure features in the classified fissure feature subset using an adaptive threshold segmentation algorithm to obtain a preliminary screening result set of high-risk areas, and extracting a boundary feature description set from the preliminary screening result set;

[0026] If the characteristic value in the boundary feature description set exceeds a second preset threshold, the adaptive threshold is optimized by a threshold adjustment technique to obtain the primary screening crack region set.

[0027] Preferably, the method for generating the expanded fissure region set includes:

[0028] Acquiring the curvature and texture contrast of boundary features from the pre-screened crack region, extracting curvature features and texture features using an edge detection algorithm, and obtaining an initial feature set;

[0029] Based on the initial feature set, the boundary curvature is analyzed by a region growing algorithm to determine the expansion direction of the crack region; if the boundary curvature exceeds a third preset threshold, the crack region is expanded along the curvature feature to obtain a preliminary expansion region;

[0030] Determining the edges of the preliminary expansion area based on texture contrast, filtering non-crack areas using the texture features, optimizing the non-crack areas using a growth algorithm, obtaining connectivity of crack extensions, and generating a connected expansion set;

[0031] Boundary features of the expanded region are extracted from the connected expanded set to obtain the expanded crack region set.

[0032] Preferably, the method for obtaining the fine screening crack feature set includes:

[0033] If the curvature change trend of the expanded crack region set matches the preset crack morphology template, the curvature distribution is calculated to extract the significant points of curvature change in the region to obtain a curvature feature point set;

[0034] Based on the curvature feature point set, point cloud subsets within the crack area are divided using point cloud segmentation technology to determine point cloud density distribution, and the point cloud density distribution is clustered and analyzed using a Gaussian mixture model to obtain preliminary characteristics of the crack depth and width;

[0035] If the proportion of high-density areas of the preliminary features exceeds that of low-density areas, the point cloud data of the high-density areas are processed by morphological filtering to obtain a smoothed point cloud subset, and the spatial distance of the crack boundary points is calculated based on the smoothed point cloud subset to determine the precise value of the crack width;

[0036] Based on the preliminary features and the precise values, a three-dimensional feature vector of the crack is constructed to obtain the fine-screened crack feature set.

[0037] Preferably, the method for obtaining the crack identification result includes:

[0038] Using the Markov random field algorithm, the initial joint probability is constructed for the geometric features and texture features in the fine screening crack feature set to obtain a preliminary probability model;

[0039] Adjusting the preliminary probability model using a random optimization algorithm to obtain an optimized probability model, and calculating the joint probability value of crack characteristics based on the optimized probability model to obtain a crack probability distribution;

[0040] A crack extraction result is generated based on the crack probability distribution, and a spatial clustering algorithm is used to group the crack regions in the crack extraction result to obtain the crack identification result.

[0041] The present invention also provides a tunnel crack identification system based on laser radar and drone photography, wherein the identification system applies any of the above-mentioned identification methods and comprises: a point cloud model construction module, a feature extraction module, a primary screening module, a crack expansion module, a fine screening module, and an identification module;

[0042] The point cloud model construction module uses a laser radar to obtain spatial geometric information, uses a drone to capture surface image information inside the tunnel, and constructs a three-dimensional textured point cloud model based on the spatial geometric information and the surface image information;

[0043] The feature extraction module is used to extract the geometric features and surface texture features of the three-dimensional texture point cloud model to obtain a preliminary crack feature set;

[0044] The preliminary screening module classifies the internal texture features of the tunnel based on the preliminary crack feature set to obtain a classified crack feature subset, and uses a threshold segmentation method to preliminarily screen high-risk crack areas in the classified crack feature subset to obtain a preliminarily screened crack area set;

[0045] The crack expansion module is used to obtain the curvature and texture contrast of the boundary features from the initial screening crack region set, and expand the crack region to generate an expanded crack region set;

[0046] If the curvature change trend of the expanded crack region set matches the preset crack morphology template, the fine screening module analyzes the point cloud density within the expanded crack region set through a Gaussian mixture model to determine the depth and width characteristics of the cracks and obtain a fine screening crack feature set;

[0047] The identification module is used to perform joint probability distribution modeling on the geometric and texture features of the cracks based on the Markov random field probability optimization algorithm for the finely screened crack feature set to obtain a crack identification result.

[0048] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the program, the tunnel crack identification method based on laser radar and drone photography is implemented.

[0049] The present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed, it implements the tunnel crack identification method based on laser radar and drone photography.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] This method generates a three-dimensional textured point cloud model by fusing point cloud data with two-dimensional image data. It then uses stereo microscopy to extract crack features and a deep convolutional neural network to classify these texture features and determine the crack risk level. Subsequently, it uses adaptive threshold segmentation and region growing algorithms to initially screen and expand high-risk crack areas. A Gaussian mixture model is then used to analyze point cloud density and determine crack depth and width characteristics. Finally, a probabilistic optimization algorithm based on Markov random fields is used to model crack features and generate the final identification results. This method enables accurate crack identification and risk assessment in complex environments, providing strong support for geological hazard warning and engineering safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0053] Figure 1 Schematic diagram of a method flow in an embodiment of the present invention;

[0054] Figure 2 Schematic diagram of the structure of an electronic device according to an embodiment of the present invention.

[0055] Description of reference numerals:

[0056] 1010 , processor; 1020 , memory; 1030 , input / output interface; 1040 , communication interface; 1050 , bus. DETAILED DESCRIPTION

[0057] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0058] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present disclosure should have the usual meanings understood by people with ordinary skills in the field to which the present disclosure belongs. The "first", "second" and similar words used in the embodiments of the present disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative position relationships. When the absolute position of the described object changes, the relative position relationship may also change accordingly.

[0059] Example 1

[0060] In this embodiment, if Figure 1 As shown, a tunnel crack identification method based on laser radar and drone photography includes:

[0061] S1. Use LiDAR to obtain spatial geometric information, use a drone to capture surface image information inside the tunnel, and construct a three-dimensional textured point cloud model based on the spatial geometric information and surface image information.

[0062] The method for constructing a three-dimensional textured point cloud model includes: using a lidar to acquire point cloud data, extracting spatial geometric features from the point cloud data through a stereo geometry algorithm, and obtaining a three-dimensional point cloud coordinate set; using a drone to photograph the two-dimensional texture data inside the tunnel, separating surface details through an image segmentation algorithm, and obtaining a texture feature set; using a preset coordinate system mapping algorithm to align the three-dimensional point cloud coordinate set and the texture feature set in the spatiotemporal dimensions, and adjusting the coordinate mapping through an iterative nearest point algorithm to obtain an optimized matching data set; based on the optimized matching data set, fusing the aligned point cloud coordinate set and texture feature set to generate a three-dimensional textured point cloud model.

[0063] In this example, a robot equipped with a lidar (lidar) scans the tunnel, capturing distance information about the surrounding environment and generating point cloud data containing millions of points. This point cloud data is recorded as 3D coordinates, representing positions in space. A volumetric geometry algorithm is then used to extract spatial geometric features, resulting in a 3D point cloud coordinate set. A drone equipped with a high-definition camera then captures images within the tunnel at a resolution of 4000 × 3000 pixels. An image segmentation algorithm is then used to separate surface details and extract the texture of the tunnel wall, forming a texture feature set. Then the coordinate system of the lidar is unified with the coordinate system of the drone camera, and the preset coordinate system mapping algorithm is used to align the three-dimensional point cloud coordinate set and the texture feature set in the time and space dimensions. The global coordinate origin is set as the starting point of the street, and the pixel points in the point cloud coordinates and texture data are aligned through rotation and translation transformations to generate a preliminary matching data set. If there is an offset, for example, the corner position in the point cloud deviates by 2 cm from the edge in the texture, the coordinate mapping is adjusted by iterative nearest point algorithm to calculate the distance between the nearest points in the two data sets, and the coordinate mapping is gradually optimized, eventually reducing the deviation to the millimeter level to obtain an optimized matching data set; based on the optimized matching data set, the aligned point cloud coordinate set and texture feature set are fused to generate a three-dimensional texture point cloud model.

[0064] S2. Extract the geometric features and surface texture features of the three-dimensional texture point cloud model to obtain a preliminary crack feature set.

[0065] The method for obtaining a preliminary crack feature set includes: obtaining crack geometry data based on a three-dimensional texture point cloud model to obtain an initial point cloud data set; extracting crack geometry and surface texture from the initial point cloud data set using stereo microscopy technology to obtain a separated feature set; judging the degree of overlap between crack binding and morphological features based on the separated feature set to determine geometric morphological features; separating surface texture from geometric morphological features to obtain surface texture features; and constructing a preliminary crack feature set based on geometric morphological features and surface texture features.

[0066] In this embodiment, a point cloud model is generated using 3D textures to obtain fracture geometry data, resulting in an initial point cloud dataset. Stereoscopic microscopy is then used to process the initial point cloud dataset, extracting fracture geometry and surface texture to obtain a separated feature set. The degree of overlap between fracture geometry and morphological features in this separated feature set is determined to determine a geometric morphological feature set. Surface texture is then separated from the geometric morphological feature set to obtain a texture feature subset, yielding an independent texture description. If the texture feature subset overlaps with fracture features, a clustering algorithm is used to process the overlap to obtain a preliminary fracture feature set.

[0067] S3. Based on the preliminary crack feature set, the internal texture features of the tunnel are classified to obtain a classified crack feature subset. The threshold segmentation method is then used to preliminarily screen the high-risk crack areas in the classified crack feature subset to obtain a preliminarily screened crack area set.

[0068] The method for obtaining a preliminary screening crack area set includes: if the number of geometric morphological feature points in the preliminary crack feature set exceeds a first preset threshold, using a deep convolutional neural network to classify the internal texture features of the tunnel, determine the potential risk level of the cracks, and obtain a classified crack feature subset; using an adaptive threshold segmentation algorithm to process the crack features in the classified crack feature subset to obtain a preliminary screening result set of high-risk areas, and extracting a boundary feature description set in the preliminary screening result set; if the feature value in the boundary feature description set exceeds a second preset threshold, optimizing the adaptive threshold through a threshold adjustment technique to obtain a preliminary screening crack area set.

[0069] In this embodiment, if the number of geometric morphological feature points of the initial screening crack area set exceeds a first preset threshold, the distribution characteristics of the number of feature points are calculated by a statistical tool to obtain a feature set after the initial screening, texture feature data are obtained from the feature set after the initial screening, and the texture features are classified by a deep convolutional neural network to obtain a classified texture feature set; based on the classified texture feature set, a clustering algorithm is used to group the crack features to determine the distribution pattern of each group of cracks; if the distribution pattern of each group of cracks meets the preset conditions, a logistic regression algorithm is used to analyze the relationship between texture features and potential risks to determine the risk level of the cracks; high-risk cracks are extracted from the determined risk levels features, generate a classified fissure feature subset; use an adaptive threshold segmentation algorithm to process the fissure features in the classified fissure feature subset to obtain a preliminary screening result set for high-risk areas; use boundary feature extraction technology to obtain regional boundary data from the preliminary screening result set and determine the preliminary screening area set; for the regional boundaries in the preliminary screening area set, use a feature extraction method to obtain a boundary feature description set; if the feature value in the boundary feature description set exceeds the preset threshold range, the adaptive threshold is optimized through the threshold adjustment technology to obtain an adjusted segmentation result set; based on the adjusted segmentation result set, the updated boundary of the high-risk area is obtained, the range of the risk area is judged, and the preliminary screening fissure area set is obtained.

[0070] S4. Obtain the curvature and texture contrast of the boundary features from the initial screening crack region set, and expand the crack region to generate an expanded crack region set.

[0071] The method for generating an expanded crack region set includes: obtaining the curvature and texture contrast of boundary features from the initially screened crack region set, extracting curvature features and texture features using an edge detection algorithm to obtain an initial feature set; based on the initial feature set, analyzing the boundary curvature using a region growing algorithm to determine the expansion direction of the crack region; if the boundary curvature exceeds a third preset threshold, expanding the crack region along the curvature feature to obtain a preliminary expanded region; judging the edges of the preliminary expanded region based on texture contrast, filtering non-crack regions using texture features, and optimizing the non-crack regions using a growing algorithm to obtain the connectivity of the crack extension and generate a connected expanded set; extracting the boundary features of the expanded region from the connected expanded set to obtain the expanded crack region set.

[0072] In this embodiment, boundary curvature can be characterized by detecting the degree of curvature at the crack edge. For example, in a 50-centimeter-long crack, if the radius of curvature of a certain section is less than 5 centimeters, this area can be considered significantly curved and potentially a high-risk point. When using an edge detection algorithm to extract curvature features, image processing techniques can be used to scan the rate of change of boundary pixels. Texture contrast is determined by comparing the grayscale difference between the crack region and its surroundings. Assuming the mean grayscale value of the crack region is 80 and that of the surrounding area is 120, a higher contrast indicates that the crack is easier to identify. When extracting texture features using the edge detection algorithm, local window analysis can be used. When analyzing boundary curvature, the region growing algorithm expands based on a seed point. Assuming a point with a curvature radius less than 3 centimeters is selected as a seed, if the curvature continuity of the adjacent region remains above 80%, expansion is carried out in this direction. If the boundary curvature exceeds a preset threshold, the crack region is expanded along the curvature feature to obtain a preliminary expansion area. The edges of the initial expansion area are determined based on texture contrast, and the non-crack area is filtered using texture features. The non-crack area is optimized through a growth algorithm to obtain the connectivity of the crack extension and generate a connected expansion set. The boundary features of the expansion area are extracted from the connected expansion set to obtain the expanded crack area set.

[0073] S5. If the curvature change trend of the expanded fracture area set matches the preset fracture morphology template, the point cloud density within the expanded fracture area set is analyzed using a Gaussian mixture model to determine the depth and width characteristics of the fractures and obtain a finely screened fracture feature set.

[0074] The method for obtaining a fine-screened crack feature set includes: if the curvature change trend of the expanded crack area set matches the preset crack morphology template, then through curvature distribution calculation, the significant points of curvature change in the area are extracted to obtain a curvature feature point set; based on the curvature feature point set, point cloud segmentation technology is used to divide the point cloud subsets in the crack area, the point cloud density distribution is determined, and the point cloud density distribution is clustered and analyzed by a Gaussian mixture model to obtain preliminary features of the crack depth and width; if the proportion of high-density areas of the preliminary features exceeds that of low-density areas, the point cloud data of the high-density area is processed by morphological filtering to obtain a smoothed point cloud subset, the spatial distance of the crack boundary points is calculated based on the smoothed point cloud subset, and the precise value of the crack width is determined; based on the preliminary features and the precise value, a three-dimensional feature vector of the crack is constructed to obtain a fine-screened crack feature set.

[0075] In this embodiment, when the curvature change trend of the expanded crack area set matches the preset crack morphology template, significant points can be extracted by analyzing the curvature distribution: assuming that the curvature value of a crack area shows an obvious mutation at the boundary, jumping from a gentle 0.2 to 0.8, it indicates that this may be the key point of the crack edge. By recording the positions of these significant points, a curvature feature point set is formed. Based on the curvature feature point set, the point cloud segmentation technology is used to divide the point cloud subsets in the crack area and determine the point cloud density distribution. The point cloud density distribution is clustered through the Gaussian mixture model to distinguish high-density and low-density areas. Assume that after clustering, 60% of the point cloud data in a certain crack area belong to the high-density area and 40% belong to the low-density area. If the proportion of high-density areas in the preliminary features exceeds that of low-density areas, the point cloud data in the high-density area is processed by morphological filtering to obtain a smoothed point cloud subset. The spatial distance of the crack boundary points is calculated based on the smoothed point cloud subset to determine the precise value of the crack width. Based on the preliminary features and the precise value, the three-dimensional feature vector of the crack is constructed to obtain a fine-screened crack feature set.

[0076] S6. For the fine-screened crack feature set, a probability optimization algorithm based on Markov random fields is used to perform joint probability distribution modeling on the geometric and texture features of the cracks to obtain the crack identification results.

[0077] The method for obtaining the crack identification result includes: using a Markov random field algorithm to construct an initial joint probability for the geometric features and texture features in the fine-screened crack feature set to obtain a preliminary probability model; using a random optimization algorithm to adjust the preliminary probability model to obtain an optimized probability model, and calculating the joint probability value of the crack features based on the optimized probability model to obtain a crack probability distribution; generating a crack extraction result based on the crack probability distribution, and using a spatial clustering algorithm to group the crack areas in the crack extraction result to obtain a crack identification result.

[0078] It should be noted that the method of the embodiments of the present disclosure can be performed by a single device, such as a computer or server. The method of the embodiments of the present disclosure can also be applied in a distributed scenario, where multiple devices cooperate to perform the method. In such a distributed scenario, one of the multiple devices may only perform one or more steps of the method of the embodiments of the present disclosure, and the multiple devices will interact with each other to complete the method.

[0079] It should be noted that the above describes some embodiments of the present disclosure. Other embodiments are within the scope of the appended claims. In some cases, it should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution, and the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention. The actions or steps recorded in the claims can be performed in an order different from that in the above embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-tasking and parallel processing are also possible or may be advantageous.

[0080] Example 2

[0081] In this embodiment, the tunnel crack identification system based on laser radar and drone photography includes: a point cloud model construction module, a feature extraction module, a primary screening module, a crack expansion module, a fine screening module and an identification module.

[0082] The point cloud model construction module uses lidar to obtain spatial geometric information and uses drones to capture surface image information inside the tunnel. Based on the spatial geometric information and surface image information, a three-dimensional textured point cloud model is constructed.

[0083] The feature extraction module is used to extract the geometric features and surface texture features of the three-dimensional texture point cloud model to obtain a preliminary crack feature set.

[0084] The initial screening module classifies the internal texture features of the tunnel based on the preliminary crack feature set to obtain the classified crack feature subset, and uses the threshold segmentation method to perform initial screening of high-risk crack areas in the classified crack feature subset to obtain the initial screening crack area set.

[0085] The crack expansion module is used to obtain the curvature and texture contrast of the boundary features from the initial screening crack region set, and expand the crack region to generate the expanded crack region set.

[0086] If the curvature change trend of the expanded fracture area set matches the preset fracture morphology template, the fine screening module analyzes the point cloud density within the expanded fracture area set through a Gaussian mixture model to determine the depth and width characteristics of the fracture and obtain a fine screening fracture feature set.

[0087] The recognition module is used to finely screen the crack feature set and adopt a probability optimization algorithm based on Markov random fields to perform joint probability distribution modeling on the geometric and texture features of the cracks to obtain the crack recognition results.

[0088] The system of the above embodiment is used to implement the corresponding tunnel crack identification method based on laser radar and drone photography in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0089] It should be noted that the above-mentioned tunnel crack identification system based on laser radar and drone photography is embodied in the form of functional units. The term "module" here can be implemented in software and / or hardware form, and is not specifically limited to this.

[0090] For example, a "module" may be a software program, a hardware circuit, or a combination of the two that implements the aforementioned functionality. The hardware circuit may include an application specific integrated circuit (ASIC), an electronic circuit, a processor (e.g., a shared processor, a dedicated processor, or a group of processors) and memory for executing one or more software or firmware programs, combined logic circuits, and / or other suitable components that support the described functionality.

[0091] Example 3

[0092] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the program, it implements the tunnel crack identification method based on laser radar and drone photography as described in any of the above embodiments.

[0093] Figure 2 10 is a schematic diagram showing a more specific hardware structure of an electronic device provided in this embodiment. The device may include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040 are communicatively connected to each other within the device via the bus 1050.

[0094] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0095] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage devices, dynamic storage devices, etc. The memory 1020 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.

[0096] The input / output interface 1030 is used to connect input / output modules to implement information input and output. The input / output modules can be configured as components within the device (not shown in the figure) or can be externally connected to the device to provide corresponding functions. Input devices may include a keyboard, mouse, touch screen, microphone, various sensors, etc., and output devices may include a display, speaker, vibrator, indicator light, etc.

[0097] The communication interface 1040 is used to connect to a communication module (not shown) to enable communication between the device and other devices. The communication module can communicate via a wired method (e.g., USB (Universal Serial Bus), network cable, etc.) or a wireless method (e.g., mobile network, WIFI (Wireless Fidelity), Bluetooth, etc.).

[0098] The bus 1050 comprises a path for transmitting information between the various components of the device (eg, the processor 1010 , the memory 1020 , the input / output interface 1030 , and the communication interface 1040 ).

[0099] It should be noted that although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040, and the bus 1050, in a specific implementation, the device may also include other components necessary for normal operation. In addition, it will be understood by those skilled in the art that the above device may only include the components necessary to implement the embodiments of this specification, and does not necessarily include all the components shown in the figure.

[0100] The system of the above embodiment is used to implement the corresponding tunnel crack identification method based on laser radar and drone photography in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0101] Example 4

[0102] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present invention also provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable the computer to execute the tunnel crack identification method based on laser radar and drone photography as described in any of the above embodiments.

[0103] The computer-readable media of this embodiment include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0104] The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the tunnel crack identification method based on laser radar and drone photography as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0105] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present disclosure (including the claims) is limited to these examples. Within the scope of the present disclosure, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present disclosure as described above, which are not provided in detail for the sake of simplicity.

[0106] In addition, to simplify the description and discussion, and so as not to obscure the embodiments of the present disclosure, known power / ground connections to integrated circuit (IC) chips and other components may or may not be shown in the provided figures. In addition, devices may be shown in the form of block diagrams to avoid obscuring the embodiments of the present disclosure, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform on which the embodiments of the present disclosure are to be implemented (i.e., these details should be fully within the purview of those skilled in the art). Where specific details (e.g., circuits) are set forth to describe exemplary embodiments of the present disclosure, it will be apparent to those skilled in the art that the embodiments of the present disclosure may be implemented without these specific details or with variations in these specific details. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0107] Although the present disclosure has been described in conjunction with specific embodiments thereof, many alternatives, modifications, and variations of these embodiments will be apparent to those skilled in the art based on the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the embodiments discussed.

[0108] Therefore, the units of each example described in the embodiments of this application can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0109] The embodiments of the present disclosure are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present disclosure should be included in the scope of protection of the present disclosure.

Claims

1. A tunnel crack identification method based on laser radar and drone photography is characterized by: The method comprises: Using a laser radar to obtain spatial geometric information, using a drone to capture surface image information of the interior of the tunnel, and constructing a three-dimensional textured point cloud model based on the spatial geometric information and the surface image information; Extracting geometric features and surface texture features of the three-dimensional texture point cloud model to obtain a preliminary crack feature set; Classifying the internal texture features of the tunnel based on the preliminary crack feature set to obtain a classified crack feature subset, and using a threshold segmentation method to preliminarily screen high-risk crack areas in the classified crack feature subset to obtain a preliminarily screened crack area set; Acquiring the curvature and texture contrast of boundary features from the initially screened crack region set, and expanding the crack region to generate an expanded crack region set; If the curvature change trend of the expanded fracture area set matches the preset fracture morphology template, the point cloud density within the expanded fracture area set is analyzed by a Gaussian mixture model to determine the depth and width characteristics of the fractures and obtain a fine-screened fracture feature set; For the finely screened crack feature set, a probability optimization algorithm based on Markov random fields is used to perform joint probability distribution modeling on the geometric and texture features of the cracks to obtain crack identification results; The method for obtaining the primary screening crack region set includes: If the number of geometric feature points in the preliminary crack feature set exceeds a first preset threshold, a deep convolutional neural network is used to classify the internal texture features of the tunnel to determine the potential risk level of the cracks, thereby obtaining the classified crack feature subset; Processing the fissure features in the classified fissure feature subset using an adaptive threshold segmentation algorithm to obtain a preliminary screening result set of high-risk areas, and extracting a boundary feature description set from the preliminary screening result set; If the feature value in the boundary feature description set exceeds a second preset threshold, the adaptive threshold is optimized by a threshold adjustment technique to obtain the primary screening crack region set; The method for generating the expanded crack region set includes: Acquiring the curvature and texture contrast of boundary features from the pre-screened crack region, extracting curvature features and texture features using an edge detection algorithm, and obtaining an initial feature set; Based on the initial feature set, the boundary curvature is analyzed by a region growing algorithm to determine the expansion direction of the crack region; if the boundary curvature exceeds a third preset threshold, the crack region is expanded along the curvature feature to obtain a preliminary expansion region; Determining the edges of the preliminary expansion area based on texture contrast, filtering non-crack areas using the texture features, optimizing the non-crack areas using a growth algorithm, obtaining connectivity of crack extensions, and generating a connected expansion set; Extracting boundary features of the extended region from the connected extended set to obtain the expanded crack region set; The method for obtaining the fine screening crack feature set includes: If the curvature change trend of the expanded crack region set matches the preset crack morphology template, the curvature distribution is calculated to extract the significant points of curvature change in the region to obtain a curvature feature point set; Based on the curvature feature point set, point cloud subsets within the crack area are divided using point cloud segmentation technology to determine point cloud density distribution, and the point cloud density distribution is clustered and analyzed using a Gaussian mixture model to obtain preliminary characteristics of the crack depth and width; If the proportion of high-density areas of the preliminary features exceeds that of low-density areas, the point cloud data of the high-density areas are processed by morphological filtering to obtain a smoothed point cloud subset, and the spatial distance of the crack boundary points is calculated based on the smoothed point cloud subset to determine the precise value of the crack width; Based on the preliminary features and the precise values, a three-dimensional feature vector of the crack is constructed to obtain the fine-screened crack feature set.

2. The tunnel crack identification method based on laser radar and drone photography according to claim 1 is characterized in that: Methods for constructing a 3D textured point cloud model include: Acquire point cloud data using a laser radar, extract spatial geometric features from the point cloud data using a volume geometry algorithm, and obtain a three-dimensional point cloud coordinate set; The two-dimensional texture data of the tunnel interior is captured by a drone, and the surface details are separated using an image segmentation algorithm to obtain a texture feature set. Using a preset coordinate system mapping algorithm, the three-dimensional point cloud coordinate set and the texture feature set are aligned in time and space dimensions, and the coordinate mapping is adjusted by an iterative closest point algorithm to obtain an optimized matching data set; Based on the optimized matching data set, the aligned point cloud coordinate set and the texture feature set are fused to generate a three-dimensional texture point cloud model.

3. The tunnel crack identification method based on laser radar and drone photography according to claim 1 is characterized in that: The method for obtaining the preliminary crack feature set includes: Acquire crack geometry data based on the three-dimensional texture point cloud model to obtain an initial point cloud data set; extracting crack geometry and surface texture from the initial point cloud dataset using stereo microscopy technology to obtain a separated feature set; Determining the degree of overlap between crack binding and morphological features based on the separated feature set to determine the geometric morphological features; Separating the surface texture from the geometric features to obtain the surface texture features; The preliminary crack feature set is constructed based on the geometric features and the surface texture features.

4. The tunnel crack identification method based on laser radar and drone photography according to claim 1 is characterized in that: The method for obtaining the crack identification result includes: Using the Markov random field algorithm, the initial joint probability is constructed for the geometric features and texture features in the fine screening crack feature set to obtain a preliminary probability model; Adjusting the preliminary probability model using a random optimization algorithm to obtain an optimized probability model, and calculating the joint probability value of crack characteristics based on the optimized probability model to obtain a crack probability distribution; A crack extraction result is generated based on the crack probability distribution, and a spatial clustering algorithm is used to group the crack regions in the crack extraction result to obtain the crack identification result.

5. A tunnel crack identification system based on laser radar and drone photography, wherein the identification system applies the identification method according to any one of claims 1 to 4, characterized in that: include: Point cloud model building module, feature extraction module, primary screening module, crack expansion module, fine screening module and recognition module; The point cloud model construction module uses a laser radar to obtain spatial geometric information, uses a drone to capture surface image information inside the tunnel, and constructs a three-dimensional textured point cloud model based on the spatial geometric information and the surface image information; The feature extraction module is used to extract the geometric features and surface texture features of the three-dimensional texture point cloud model to obtain a preliminary crack feature set; The preliminary screening module classifies the internal texture features of the tunnel based on the preliminary crack feature set to obtain a classified crack feature subset, and uses a threshold segmentation method to preliminarily screen high-risk crack areas in the classified crack feature subset to obtain a preliminarily screened crack area set; The crack expansion module is used to obtain the curvature and texture contrast of the boundary features from the initial screening crack region set, and expand the crack region to generate an expanded crack region set; If the curvature change trend of the expanded crack region set matches the preset crack morphology template, the fine screening module analyzes the point cloud density within the expanded crack region set through a Gaussian mixture model to determine the depth and width characteristics of the cracks and obtain a fine screening crack feature set; The identification module is used to perform joint probability distribution modeling on the geometric and texture features of the cracks based on the Markov random field probability optimization algorithm for the finely screened crack feature set to obtain a crack identification result.

6. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the program, the method for identifying tunnel cracks based on laser radar and drone photography as described in any one of claims 1 to 4 is implemented.

7. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed, the tunnel crack identification method based on laser radar and drone photography as described in any one of claims 1 to 4 is implemented.

Images