Reconstruction method and system for cavity shielded by shallow reinforcing mesh

Abstract

The invention discloses a method and a system for reconstructing a cavity shielded by a shallow reinforcing mesh. The method comprises the following steps: constructing a layered medium scene containing the shallow reinforcing mesh and the cavity; acquiring image data information with different shielding states and preprocessing the image data information to construct a reconstruction data set; constructing a reconstruction initial model of the cavity shielded by the shallow reinforcing mesh, and training to obtain a reconstruction model of the cavity shielded by the shallow reinforcing mesh; and adopting the obtained reconstruction model of the cavity shielded by the shallow reinforcing mesh to reconstruct the actual cavity shielded by the shallow reinforcing mesh. According to the method, multiple types of training data are acquired through a constructed layered medium scene containing a shallow reinforcing mesh and a cavity, and a reconstruction model of the cavity shielded by the shallow reinforcing mesh, which comprises a U-shaped network, a Sobel operator, an attention mechanism, a residual structure, an average mechanism and a jump connection scheme, is trained; therefore, the reconstruction of the cavity under the shielding of the shallow reinforcing mesh can be realized, the reliability is higher, the accuracy is better, and the reconstruction efficiency is higher.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of ground penetrating radar detection, and particularly relates to a reconstruction method and system for a cavity under shielding of a shallow steel bar mesh. BACKGROUND

[0002] Ground penetrating radar (GPR) is a non-destructive testing technology based on the principle of high-frequency electromagnetic wave reflection, which is suitable for detecting medium changes and abnormal reflectors in the underground or internal structure. The working principle of GPR is to lay an antenna system on the ground, transmit high-frequency pulse electromagnetic waves, and receive the signals reflected or scattered at different medium boundaries, thereby obtaining the spatial structure information of the underground or internal medium. GPR technology has the characteristics of moderate penetration depth, high resolution, non-destructive, etc., and is widely used in many fields.

[0003] However, in typical scenarios such as municipal roads and underground engineering, there is often a steel bar mesh in the shallow layer. As a strong reflector, the scattering signal of the steel bar mesh will significantly mask the weak target echo under the steel bar mesh, especially the reflection signal of the cavity disease. At present, researchers have proposed corresponding schemes for reconstructing underground structure images from the obtained B-scan images of GPR, mainly including offset algorithm schemes, tomographic imaging schemes, full waveform inversion schemes, and deep learning-based schemes. For example, Ding Yilin et al. proposed a YOLO network detection algorithm, which combines simulation data sets and real data sets to detect various diseases under the shielding of the steel bar mesh; however, this scheme ignores the position and size of the disease. Zhao Huimin et al. based on the pix2pix parameter inversion network directly inverts the defects on the three-dimensional tunnel similar model containing the shallow steel bar mesh, and the steel bar mesh interference affects the inversion result to some extent.

[0004] In summary, the existing schemes do not consider the problem of interference suppression of steel bar mesh echo on underground cavity inversion in the layered background medium scene, and the accuracy and reliability of cavity inversion are relatively low. SUMMARY

[0005] One of the purposes of the present application is to provide a reconstruction method for a cavity under shielding of a shallow steel bar mesh, which has high reliability, good accuracy and high efficiency.

[0006] The second purpose of the present application is to provide a system for implementing the reconstruction method for a cavity under shielding of a shallow steel bar mesh.

[0007] The reconstruction method for a cavity under shielding of a shallow steel bar mesh provided by the present application includes the following steps:

[0008] S1. Construct a layered medium scene containing a shallow steel bar mesh and a cavity;

[0009] S2. Based on the scene constructed in step S1, image data information with different shielding states is obtained;

[0010] S3. The image data information obtained in step S2 is preprocessed to construct a reconstruction data set;

[0011] S4. Constructing a reconstruction initial model of the cavity under the shielding of the shallow reinforcement mesh;

[0012] The reconstruction initial model of the cavity under the shielding of the shallow reinforcement mesh includes a first stage model and a second stage model;

[0013] Based on the U-shaped network, the Sobel operator, the residual structure and the skip connection scheme, the first stage model is constructed; the first stage model is used to suppress the echo of the reinforcement mesh;

[0014] Based on the U-shaped network, the residual structure, the attention mechanism, the average mechanism, the multi-scale edge fusion and the skip connection scheme, the second stage model is constructed; the second stage model is used to reconstruct the cavity according to the output of the first stage model;

[0015] S5. The reconstruction data set obtained in step S3 is used to train the reconstruction initial model of the cavity under the shielding of the shallow reinforcement mesh constructed in step S4, to obtain a reconstruction model of the cavity under the shielding of the shallow reinforcement mesh;

[0016] S6. The reconstruction model of the cavity under the shielding of the shallow reinforcement mesh obtained in step S5 is used to reconstruct the actual cavity under the shielding of the shallow reinforcement mesh.

[0017] The step S1 specifically includes the following steps:

[0018] The constructed layered medium scene includes four layers, from top to bottom, air layer, surface layer, base layer containing reinforcement mesh echo and soil base layer;

[0019] The material of the surface layer is set to be asphalt concrete, and at the same time, the surface layer is divided into three layers according to the different particle sizes of the aggregate, the different shapes of the aggregate and the random distribution of the aggregate position;

[0020] The material of the base layer is set to be reinforced concrete, and the spacing and number of steel bars are set, and the base layer is divided into two layers according to the different particle sizes of the aggregate and the different shapes of the aggregate;

[0021] The soil base layer is set to be a random mixture of soil, and the soil surface is set to be an uneven undulating surface.

[0022] The step S2 specifically includes the following steps:

[0023] Based on the scene constructed in step S1, a plurality of forward model files containing cavity targets are generated;

[0024] The type of the cavity target includes a rectangular-shaped non-water cavity, a rectangular-shaped water cavity, an irregular-shaped non-water cavity, and an irregular-shaped water cavity; the cavity target is arranged in a soil base layer; and the size and depth of the cavity are randomly generated.

[0025] The obtained forward model file includes a B-scan image of the underground cavity when the shallow reinforcement net is shielded, a B-scan image of the underground cavity when the shallow reinforcement net is not shielded, and a black-and-white mask image of the cavity target.

[0026] The step S3 specifically includes the following steps.

[0027] A maximum reference threshold of the B-scan image is set.

[0028] Signal suppression processing: all data higher than the maximum reference threshold in the image data information obtained in step S2 is replaced by the maximum reference threshold to complete the suppression of strong signals.

[0029] Signal enhancement processing: an exponential gain sequence that increases with the increase of depth is constructed, and the data after the signal suppression processing is multiplied by the constructed exponential gain sequence to complete the signal enhancement processing.

[0030] Image scaling processing: the image data after the signal enhancement processing is uniformly scaled to a set size.

[0031] The preprocessing of the image data information obtained in step S2 is completed.

[0032] The first-stage model includes a first-stage encoder, a first-stage decoder, and a first-stage filtering module.

[0033] The first-stage encoder includes a first coding first residual module, a first coding first edge module, a first coding first max-pooling module, a first coding second residual module, a first coding second edge module, a first coding second max-pooling module, a first coding third residual module, a first coding third edge module, a first coding third max-pooling module, a first coding fourth residual module, a first coding fourth edge module, a first coding fourth max-pooling module, and a first coding fifth residual module, which are sequentially connected in series.

[0034] The first-stage decoder comprises, in sequence, a first-decoding first-transposed-convolution module, a first-decoding first-residual module, a first-decoding first-edge module, a first-decoding second-transposed-convolution module, a first-decoding second-residual module, a first-decoding second-edge module, a first-decoding third-transposed-convolution module, a first-decoding third-residual module, a first-decoding third-edge module, a first-decoding fourth-transposed-convolution module, a first-decoding fourth-residual module, and a first-decoding fourth-edge module; meanwhile, the output of the first-decoding fourth-edge module and the output of the first-decoding first-residual module are spliced to serve as the input of the first-decoding first-edge module; the output of the first-decoding third-edge module and the output of the first-decoding second-residual module are spliced to serve as the input of the first-decoding second-edge module; the output of the first-decoding second-edge module and the output of the first-decoding third-residual module are spliced to serve as the input of the first-decoding third-edge module; and the output of the first-decoding first-edge module and the output of the first-decoding fourth-residual module are spliced to serve as the input of the first-decoding fourth-edge module;

[0035] The first-stage filter module comprises, in sequence, a first-stage convolution layer and a first-stage activation function layer, and the output of the first-stage decoder is the input of the first-stage filter module;

[0036] The first-stage model is used for suppressing the echo of the steel bar mesh and outputs a GPR image without the echo of the steel bar mesh.

[0037] The first-stage convolution layer is a 2d convolution layer; The first-decoding first-transposed-convolution module to the first-decoding fourth-transposed-convolution module are all transposed convolution layers and are used for enlarging the resolution; The first-decoding first-transposed-convolution module to the first-decoding fourth-transposed-convolution module are all transposed convolution layers and are used for enlarging the resolution; The first-decoding first-transposed-convolution module to the first-decoding fourth-transposed-convolution module are all transposed convolution layers and are used for enlarging the resolution;

[0038] The first-decoding first-edge module to the first-decoding fourth-edge module have the same structure;

[0039] The processing procedure of the edge module comprises the following steps:

[0040] The horizontal edge information and the vertical edge information of the input image are detected by using a Sobel operator, and the absolute horizontal edge information and the absolute vertical edge information are obtained by taking the absolute values, respectively;

[0041] The obtained absolute horizontal edge information and absolute vertical edge information are spliced with the input image information, and then the spliced image information is processed by a The output of the edge module is obtained by sequentially passing the output of the 2D convolution layer through a normalization layer and a ReLU activation function layer.

[0042] The second-stage model includes a second-stage encoder, a second-stage decoder, a second-stage average module, and a second-stage reconstruction module.

[0043] The second-stage encoder includes a second-encoding first residual module, a second-encoding first max-pooling module, a second-encoding second residual module, a second-encoding second max-pooling module, a second-encoding third residual module, a second-encoding third max-pooling module, a second-encoding fourth residual module, a second-encoding fourth max-pooling module, and a second-encoding fifth residual module connected in sequence.

[0044] The second-stage encoder includes a second-decoding first transpose convolution module, a second-decoding first residual module, a second-decoding second transpose convolution module, a second-decoding second residual module, a second-decoding third transpose convolution module, a second-decoding third residual module, a second-decoding fourth transpose convolution module, and a second-decoding fourth residual module connected in sequence. Meanwhile, the output of the second-encoding fourth residual module is processed by a fourth attention module and spliced with the output of the second-encoding fifth residual module to serve as the input of the second-decoding first transpose convolution module. The output of the second-encoding third residual module is processed by a third attention module and spliced with the output of the second-decoding first residual module to serve as the input of the second-decoding second transpose convolution module. The output of the second-encoding second residual module is processed by a second attention module and spliced with the output of the second-decoding second residual module to serve as the input of the second-decoding third transpose convolution module. The output of the second-encoding first residual module is processed by a first attention module and spliced with the output of the second-decoding third residual module to serve as the input of the second-decoding fourth transpose convolution module.

[0045] The second-stage averaging module includes a second-averaging first convolutional layer, a second-averaging second convolutional layer, a second-averaging third convolutional layer, and a second-averaging fourth convolutional layer. The output of the second encoding first residual module is processed by the second-averaging first convolutional layer to obtain a second-averaged first signal. The output of the second encoding second residual module is processed by the second-averaging second convolutional layer to obtain a second-averaged second signal. The output of the second encoding third residual module is processed by the second-averaging third convolutional layer to obtain a second-averaged third signal. The output of the second encoding fourth residual module is processed by the second-averaging fourth convolutional layer to obtain a second-averaged fourth signal. The average of the obtained second-averaged first signal to second-averaged fourth signal is calculated to obtain the output of the second-stage averaging module. Here, the second-averaged first signal to second-averaged fourth signal are all B-scan two-dimensional images of equal size. The average of these four images is used to obtain the output of the second-stage averaging module. The output of the second-stage averaging module is also a B-scan two-dimensional image of the same size.

[0046] The second-stage reconstruction module includes reconstructing the convolutional layer; the output of the second-stage decoder is processed by the reconstructed convolutional layer and then summed with the output of the second-stage averaging module to obtain the final dilated reconstruction result.

[0047] Reconstruct the convolutional layer as The 2D convolutional layer; the reconstructed activation function layer is a Sigmoid activation function layer; the second encoding first max pooling module to the second encoding fourth max pooling module are all... The first module is a max pooling module with a stride of 2; the second decoding first transposed convolution module and the second decoding first transposed convolution module are both... The transposed convolutional layer; the second average first convolutional layer to the second average fourth convolutional layer are all... 2D convolutional layers.

[0048] The structures of the first attention module through the fourth attention module are all the same;

[0049] The attention module's processing steps include the following:

[0050] Obtain the gated feature map of the input image and jump feature map Gated feature map The feature map output by the second decoding first residual module, the skip feature map The feature map output by the second encoding and third residual modules;

[0051] Will Passing in sequence The image is processed by 2D convolutional layers and normalization layers to obtain the first image component; Passing in sequence a 2D convolution layer and a normalization layer, to obtain a second image component;

[0052] a ReLU activation function layer, a 2D convolution layer, a normalization layer and a Sigmoid activation function layer, to obtain a third image component;

[0053] a pixel multiplication operation, to obtain the output of the attention module.

[0054] The structures of the first encoding first residual module to the first encoding fifth residual module, the first decoding first residual module to the first decoding fourth residual module, the second encoding first residual module to the second encoding fifth residual module, and the second decoding first residual module to the second decoding fourth residual module are the same;

[0055] The processing procedure of the residual module includes the following steps:

[0056] The input image data is divided into two paths: the first path sequentially passes through a 2D convolution layer, a normalization layer, a ReLU activation function layer, a 2D convolution layer and a normalization layer, to obtain first path features; and the second path sequentially passes through a 2D convolution layer and a normalization layer, to obtain second path features;

[0057] The first path features and the second path features are added and then processed by a ReLU activation function layer, to obtain the output of the residual module.

[0058] The training in step S5 specifically includes the following steps:

[0059] During the training, the preprocessed B-scan image of the underground cavity with the shallow layer of reinforcement shielding is taken as the input of the first stage model, the preprocessed B-scan image of the underground cavity without the shallow layer of reinforcement shielding is taken as the label of the input of the first stage model, and the first stage model is trained;

[0060] Then, the output of the first stage model is taken as the input of the second stage model, and the preprocessed black and white mask image of the cavity target is taken as the label of the input of the second stage model, and the second stage model is trained;

[0061] The loss function of the first stage model is a weighted sum of an L1 loss function and an SSIM loss function;

[0062] The loss function of the second stage model is a weighted sum of a BCE loss function, a Dice loss function, a Boundary loss function and a Focal loss function.

[0063] During training, the weighted sum of the loss function of the first stage model and the loss function of the second stage model is used as the total loss function.

[0064] During training, the Adam optimizer is used for training, and the learning rate adaptive mechanism and the early stopping strategy are used to improve the stability and efficiency of the training process.

[0065] The application also provides a system for implementing the reconstruction method of the cavity under the shielding of the shallow reinforcement mesh, comprising a scene construction module, a data acquisition module, a data processing module, a model construction module, a model training module and a cavity reconstruction module; the scene construction module, the data acquisition module, the data processing module, the model construction module, the model training module and the cavity reconstruction module are sequentially connected; the scene construction module is used to construct a layered medium scene containing a shallow reinforcement mesh and a cavity, and upload data information to the data acquisition module; the data acquisition module is used to acquire image data information with different shielding states based on the received data information and the constructed scene, and upload data information to the data processing module; the data processing module is used to preprocess the acquired image data information according to the received data information, to construct a reconstruction data set, and upload data information to the model construction module; the model construction module is used to construct a reconstruction initial model of the cavity under the shielding of the shallow reinforcement mesh according to the received data information, and upload data information to the model training module; wherein the reconstruction initial model of the cavity under the shielding of the shallow reinforcement mesh comprises a first stage model and a second stage model; the first stage model is constructed based on a U-shaped network, a Sobel operator, a residual structure and a skip connection scheme; the first stage model is used to suppress the echo of the reinforcement mesh; the second stage model is constructed based on a U-shaped network, a residual structure, an attention mechanism, an average mechanism, a multi-scale edge fusion and a skip connection scheme; the second stage model is used to reconstruct the cavity according to the output of the first stage model; the model training module is used to train the constructed reconstruction initial model of the cavity under the shielding of the shallow reinforcement mesh using the obtained reconstruction data set according to the received data information, to obtain a reconstruction model of the cavity under the shielding of the shallow reinforcement mesh, and upload data information to the cavity reconstruction module; the cavity reconstruction module is used to perform actual reconstruction of the cavity under the shielding of the shallow reinforcement mesh using the obtained reconstruction model of the cavity under the shielding of the shallow reinforcement mesh according to the received data information.

[0066] The reconstruction method and system of the cavity under the shallow reinforcement net shielding provided by the application can obtain multiple types of training data by constructing a layered medium scene containing a shallow reinforcement net and a cavity, and train a reconstruction model of the cavity under the shallow reinforcement net shielding including a U-shaped network, a Sobel operator, an attention mechanism, a residual structure, an average mechanism and a skip connection scheme, so that the application can not only realize the reconstruction of the cavity under the shallow reinforcement net shielding, but also has higher reliability, better accuracy and higher reconstruction efficiency. BRIEF DESCRIPTION OF DRAWINGS DETAILED DESCRIPTION OF THE INVENTION BRIEF DESCRIPTION OF DRAWINGS

[0067] Figure 1 The method flowchart of the method of the application.

[0068] Figure 2 The structure diagram of the first stage model of the method of the application.

[0069] Figure 3 The structure diagram of the edge module of the method of the application.

[0070] Figure 4 The structure diagram of the second stage model of the method of the application.

[0071] Figure 5 The structure diagram of the attention module of the method of the application.

[0072] Figure 6 The structure diagram of the residual module of the method of the application.

[0073] Figure 7 The reconstruction result diagram of the rectangular cavity of the embodiment of the method of the application.

[0074] Figure 8 The comparison diagram of the first stage network result and the real data of the embodiment of the method of the application; wherein, Figure 8 (a) is the output result diagram of the first stage network, Figure 8 (b) is the real first stage network label diagram obtained by simulation.

[0075] Figure 9 The comparison diagram of the second stage network result and the real data of the embodiment of the method of the application; wherein, Figure 9 (a) is the output result diagram of the second stage network, Figure 9 (b) is the real second stage network label diagram obtained by simulation.

[0076] Figure 10 The function module diagram of the system of the application. DETAILED DESCRIPTION

[0077] As Figure 1The method flowchart of the method of the application is shown: the reconstruction method of the cavity under the shallow reinforcement net cover disclosed by the application comprises the following steps:

[0078] S1. Construct a layered medium scene containing a shallow reinforcement net and a cavity; specifically comprising the following steps:

[0079] The constructed layered medium scene comprises four layers, from top to bottom, an air layer, a surface layer, a base layer containing reinforcement net echoes and a soil base layer;

[0080] The material of the surface layer is set as asphalt concrete, and the surface layer is divided into three layers according to different particle sizes of the aggregate, different shapes of the aggregate and random distribution of the aggregate position;

[0081] The material of the base layer is set as reinforced concrete, and the spacing (preferably 70 cm, 40 cm, 28 cm and 24 cm) and the number (preferably 3, 5, 7 and 8) of the reinforcement are set, and the base layer is divided into two layers according to different particle sizes of the aggregate and different shapes of the aggregate;

[0082] The soil base layer is set as randomly mixed soil, and the soil surface is set as an uneven undulating surface.

[0083] In a preferred scheme, the constructed scene has a horizontal length of 2.0 meters and a vertical depth of 1.6 meters, and the overall medium is divided into four layers: the first layer is an air layer with a thickness of 0.1 meters; the second layer is a surface layer with a thickness of 0.18 meters; the third layer is a base layer with a thickness of 0.52 meters; and the fourth layer is a soil base layer with a thickness of 0.80 meters; and an irregular undulating surface is provided above the roadbed, with a maximum undulating height of 0.06m; the surface layer is divided into three small layers according to different particle sizes of the aggregate, different shapes of the aggregate and random distribution of the aggregate position, and each layer has a thickness of 0.04m, 0.06m and 0.08m; the base layer is divided into two small layers according to different particle sizes of the aggregate, different shapes of the aggregate and random distribution of the aggregate position, and each layer has a thickness of 0.36m and 0.18m;

[0084] The reinforcement is arranged in a horizontal direction at the junction of the asphalt and the base layer, i.e. in the surface layer, with a fixed radius of 0.005m, which does not change during the sample generation process; different complexity radar reflection background modeling is achieved by controlling the number and spacing of the reinforcement in the reinforcement net;

[0085] In the model, the horizontal range is 0.5-1.5 meters and the vertical range is 0.9-1.08 meters, and the cavity center is randomly embedded in the soil base layer without physical overlap with the reinforcement net component; ensuring a certain diversity and randomness in different positions, sizes and forms;

[0086] In the aspect of reconstructing the position and size of the cavity, considering the complexity of the dielectric constant of the scene and the singleness of the dielectric constant of the cavity target, the layered medium scene is set to 1 and the cavity is set to 0, and the corresponding black and white mask image of the reconstructed position and size of the cavity is obtained through the corresponding forward model file;

[0087] In the setting of the forward parameter, the radar emission waveform is selected as Ricker wavelet, the center frequency is set to 800MHz, and the time window is set to 27ns to ensure effective detection of deep targets, and the antenna type is selected as a real aperture antenna.

[0088] S2. Based on the scene constructed in step S1, obtain image data information with different shielding states; specifically including the following steps:

[0089] Based on the scene constructed in step S1, generate a plurality of forward model files containing cavity targets;

[0090] The type of the cavity target includes a rectangular-shaped non-water cavity, a rectangular-shaped water-containing cavity, an irregular-shaped non-water cavity, and an irregular-shaped water-containing cavity; the cavity target is arranged in the soil layer; the size and depth of the cavity are randomly generated;

[0091] The obtained forward model file includes a B-scan image of the underground cavity when the shallow steel mesh is shielded, a B-scan image of the underground cavity when the shallow steel mesh is not shielded, and a black and white mask image of the cavity target.

[0092] S3. Preprocess the image data information obtained in step S2 to construct a reconstruction data set; specifically including the following steps:

[0093] Set the maximum reference threshold of the B-scan image;

[0094] Signal suppression processing: due to the strong reflection of the steel mesh, the echo of the weak target signal under the steel mesh is not obvious; therefore, in the image data information obtained in step S2, all data higher than the maximum reference threshold are replaced by the maximum reference threshold to complete the suppression of strong signals;

[0095] Signal enhancement processing: the cavity is usually located at a relatively deep position, has a low relative dielectric constant, and has a weak target echo signal; therefore, an exponential gain sequence that increases with depth is constructed, and the data after signal suppression processing is multiplied by the constructed exponential gain sequence to complete signal enhancement processing;

[0096] Image scaling processing: the image data after signal enhancement processing is uniformly scaled to a set size;

[0097] The preprocessing of the image data information obtained in step S2 is completed.

[0098] S4. Constructing a reconstruction initial model of the cavity under the shielding of the shallow reinforcement mesh;

[0099] Since the strong reflection of the reinforcement mesh has a great influence on the reconstruction accuracy of the cavity, a two-stage model is constructed, the task of the first stage is to suppress the reinforcement mesh echo; the task of the second stage is to reconstruct the cavity target in the GPR B-scan image of the first stage which suppresses the reinforcement mesh echo.

[0100] The reconstruction initial model of the cavity under the shielding of the shallow reinforcement mesh includes a first stage model and a second stage model;

[0101] Based on the U-shaped network, the Sobel operator, the residual structure and the skip connection scheme, the first stage model is constructed; the first stage model is used to suppress the echo of the reinforcement mesh;

[0102] Based on the U-shaped network, the residual structure, the attention mechanism, the average mechanism, the multi-scale edge fusion and the skip connection scheme, the second stage model is constructed; the second stage model is used to reconstruct the cavity according to the output of the first stage model.

[0103] In specific implementation, the constructed first stage model includes a first stage encoder, a first stage decoder and a first stage filtering module, and the specific structure is as shown in Figure 2 ;

[0104] The first stage encoder includes a first encoding first residual module, a first encoding first edge module, a first encoding first max-pooling module, a first encoding second residual module, a first encoding second edge module, a first encoding second max-pooling module, a first encoding third residual module, a first encoding third edge module, a first encoding third max-pooling module, a first encoding fourth residual module, a first encoding fourth edge module, a first encoding fourth max-pooling module and a first encoding fifth residual module which are sequentially connected in series;

[0105] The first-stage decoder comprises, in sequence, a first decoding first transposed convolutional module, a first decoding first residual module, a first decoding first edge module, a first decoding second transposed convolutional module, a first decoding second residual module, a first decoding second edge module, a first decoding third transposed convolutional module, a first decoding third residual module, a first decoding third edge module, a first decoding fourth transposed convolutional module, a first decoding fourth residual module, and a first decoding fourth edge module. Simultaneously, the output of the first encoding fourth edge module and the output of the first decoding first residual module are concatenated and used as the input of the first decoding first edge module; the output of the first encoding third edge module and the output of the first decoding second residual module are concatenated and used as the input of the first decoding second edge module; the output of the first encoding second edge module and the output of the first decoding third residual module are concatenated and used as the input of the first decoding third edge module; and the output of the first encoding first edge module and the output of the first decoding fourth residual module are concatenated and used as the input of the first decoding fourth edge module.

[0106] The first-stage filtering module includes a first-stage convolutional layer and a first-stage activation function layer connected in series, and the output of the first-stage decoder is the input of the first-stage filtering module;

[0107] The first-stage model is used to suppress the echo of the rebar mesh and output a GPR image without the echo of the rebar mesh.

[0108] In the first-stage model, shallow edge features of the encoder channel are spliced ​​with features of the same scale of the decoder through skip connections, thereby enhancing the preservation of weak hyperbolic signals after the reinforcement mesh suppression.

[0109] The first stage of the convolutional layer is... The 2D convolutional layer; the first-stage activation function layer is a ReLU activation function layer; the first decoding first transposed convolutional module to the first decoding fourth transposed convolutional module are all... The transposed convolutional layer is used to amplify the resolution; the first encoding first max pooling module to the first encoding fourth max pooling module are all... A max-pooling layer with a stride of 2 is used to halve the size of the input feature map.

[0110] The first encoding first edge module to the first encoding fourth edge module, and the first decoding first edge module to the first decoding fourth edge module all have the same structure;

[0111] The processing of the edge module includes the following steps (structure as follows): Figure 3 (as shown)

[0112] The Sobel operator is used to detect horizontal edge information and vertical edge information of the input image, and absolute horizontal edge information and absolute vertical edge information are obtained by taking absolute values, respectively. The horizontal direction edge detection and the vertical direction edge detection are performed by the convolution kernels, wherein the horizontal direction convolution kernel weight matrix is fixed as , the vertical direction convolution kernel weight matrix is fixed as , and all convolution kernel weights are set to be untrainable state;

[0113] After the obtained absolute horizontal edge information and absolute vertical edge information are spliced with the input image information, fusion processing is performed by a 2d convolution layer, and then processing is sequentially performed by a normalization layer and a ReLU activation function layer to obtain the output of the edge module.

[0114] The edge module retains the classical edge detection characteristics by fixed convolution kernels, and enhances the edge information in the original features by the feature fusion mechanism, thereby effectively improving the feature representation capability.

[0115] In specific implementation, the constructed second stage model includes a second stage encoder, a second stage decoder, a second stage average module and a second stage reconstruction module, and the specific structure is shown in Figure 4 .

[0116] The second stage encoder includes a second encoding first residual module, a second encoding first max pooling module, a second encoding second residual module, a second encoding second max pooling module, a second encoding third residual module, a second encoding third max pooling module, a second encoding fourth residual module, a second encoding fourth max pooling module and a second encoding fifth residual module which are sequentially connected.

[0117] ​The second stage encoder comprises a second decoding first transposed convolution module, a second decoding first residual module, a second decoding second transposed convolution module, a second decoding second residual module, a second decoding third transposed convolution module, a second decoding third residual module, a second decoding fourth transposed convolution module and a second decoding fourth residual module connected in sequence; meanwhile, the output of the second encoding fourth residual module is processed through a fourth attention module and spliced with the output of the second encoding fifth residual module to serve as the input of the second decoding first transposed convolution module; the output of the second encoding third residual module is processed through a third attention module and spliced with the output of the second decoding first residual module to serve as the input of the second decoding second transposed convolution module; the output of the second encoding second residual module is processed through a second attention module and spliced with the output of the second decoding second residual module to serve as the input of the second decoding third transposed convolution module; the output of the second encoding first residual module is processed through a first attention module and spliced with the output of the second decoding third residual module to serve as the input of the second decoding fourth transposed convolution module;

[0118] The second stage average module comprises a second average first convolution layer, a second average second convolution layer, a second average third convolution layer and a second average fourth convolution layer; the output of the second encoding first residual module is processed through the second average first convolution layer to obtain a second average first signal; the output of the second encoding second residual module is processed through the second average second convolution layer to obtain a second average second signal; the output of the second encoding third residual module is processed through the second average third convolution layer to obtain a second average third signal; the output of the second encoding fourth residual module is processed through the second average fourth convolution layer to obtain a second average fourth signal; the second average first signal to the second average fourth signal are averaged to obtain the output of the second stage average module; wherein the second average first signal to the second average fourth signal are B-scan two-dimensional images with equal sizes, and here, four images are averaged to obtain the output of the second stage average module; the output of the second stage average module is also a B-scan two-dimensional image with the same size;

[0119] The second stage reconstruction module comprises a reconstruction convolution layer; the output of the second stage decoder is processed through the reconstruction convolution layer and summed with the output of the second stage average module to obtain the final result of the cavity reconstruction.

[0120] In the second stage model, the attention gate jump connection is introduced at the channel to fuse the texture of the encoder shallow layer and the semantics of the decoder high layer, so that the network has fine boundaries and global context when reconstructing the cavity, and the ability to capture and locate the weak reflection signal is significantly enhanced.

[0121] The reconstruction convolution layer is The 2D convolutional layer; the reconstructed activation function layer is a Sigmoid activation function layer; the second encoding first max pooling module to the second encoding fourth max pooling module are all... The first module is a max pooling module with a stride of 2; the second decoding first transposed convolution module and the second decoding first transposed convolution module are both... The transposed convolutional layer; the second average first convolutional layer to the second average fourth convolutional layer are all... 2D convolutional layers.

[0122] The structures of the first attention module through the fourth attention module are all the same;

[0123] The processing procedure of the attention module is structured as follows: Figure 5 As shown, it includes the following steps:

[0124] Obtain the gated feature map of the input image and jump feature map Gated feature map The feature map output by the second decoding first residual module, the skip feature map The feature map output by the second encoding and third residual modules;

[0125] Will Passing in sequence The image is processed by 2D convolutional layers and normalization layers to obtain the first image component; Passing in sequence The second image component is obtained by processing 2D convolutional layers and normalization layers;

[0126] After summing the first and second image components, the image is then passed sequentially through a ReLU activation function layer. The image is processed by 2D convolutional layers, normalization layers, and sigmoid activation function layers to obtain the third image component;

[0127] Combine the third image component with After multiplying by pixels, the output of the attention module is obtained.

[0128] The attention module effectively identifies and enhances spatial feature regions related to the current decoding stage in skip connections through dual-path feature compression and co-activation mechanisms, significantly improving the accuracy of feature fusion.

[0129] The structures of the first encoding first residual module to the first encoding fifth residual module, the first decoding first residual module to the first decoding fourth residual module, the second encoding first residual module to the second encoding fifth residual module, and the second decoding first residual module to the second decoding fourth residual module are all the same;

[0130] The processing procedure of the residual module is structured as follows: Figure 6As shown, comprising the following steps:

[0131] The input image data is divided into two paths: the first path sequentially passes through 2d convolutional layer (expanding input channels to target output dimension and maintaining spatial size), normalization layer, ReLU activation function layer, 2d convolutional layer (deep feature extraction and maintaining channel dimension) and normalization layer of Residual Block, to obtain the first path feature; the second path sequentially passes through 2d convolutional layer (adjusting the number of channels) and normalization layer, to obtain the second path feature;

[0132] The first path feature and the second path feature are added and then processed by the ReLU activation function layer to obtain the output of the residual module;

[0133] The residual module maintains spatial dimension invariability through equal-width padding, realizes adaptive adjustment of channel dimension through conditional shortcut connection, and preserves original feature information using the identity mapping mechanism, effectively solving the gradient disappearance problem in deep neural network training.

[0134] S5. Using the reconstructed data set obtained in step S3, the reconstructed initial model of the shallow reinforcement mesh-shielded cavity is trained to obtain the reconstructed model of the shallow reinforcement mesh-shielded cavity.

[0135] During training, the preprocessed B-scan image of the underground cavity with shallow reinforcement mesh shielding is used as the input of the first stage model, the preprocessed B-scan image of the underground cavity without shallow reinforcement mesh shielding is used as the label of the input of the first stage model, and the first stage model is trained.

[0136] Then, the output of the first stage model is used as the input of the second stage model, and the preprocessed black and white mask image of the cavity target is used as the label of the input of the second stage model, and the second stage model is trained.

[0137] During training, the loss function of the first stage model is the weighted sum of the L1 loss function and the SSIM loss function.

[0138] The loss function of the second stage model is the weighted sum of the BCE loss function, the Dice loss function, the Boundary loss function, and the Focal loss function.

[0139] During training, the weighted sum of the loss function of the first stage model and the loss function of the second stage model is used as the total loss function.

[0140] During training, the Adam optimizer is used for training, and the learning rate adaptive mechanism and the early stopping strategy are used to improve the stability and efficiency of the training process.

[0141] S6. Adopting the reconstruction model of the cavity under the shallow reinforcement net shielding obtained in step S5, the actual reconstruction of the cavity under the shallow reinforcement net shielding is carried out.

[0142] The application can automatically output high-precision underground cavity distribution map, has fast processing capacity and good robustness, and is suitable for underground cavity detection and safety evaluation tasks in various layered geological or urban infrastructure scenes.

[0143] The effect of the method of the application is described below in combination with an embodiment:

[0144] Figure 7 The reconstruction result diagram of the rectangular cavity with a steel bar spacing of 70 cm is shown. Figure 7 Input into the two-stage network trained by the scheme of the application.

[0145] The comparison diagram of the result output by the first-stage network and the real data is shown in Figure 8 Through Figure 8 It can be seen that the first-stage network steel reinforcement net clutter is well suppressed, while the cavity target echo is well preserved, the target boundary profile is clear, and the weak echo structure can be distinguished. This shows that the proposed first-stage network can weaken the strong scattering steel reinforcement net interference while preserving as much effective information of the cavity as possible, providing reliable input data for the target position and size reconstruction of the second stage.

[0146] The comparison diagram of the result output by the second-stage network and the real data is shown in Figure 9 Through Figure 9 It can be seen that the second-stage network can accurately identify the position of the cavity target, and better restore its geometric shape and size features. For the rectangular cavity, the reconstruction result is highly consistent with the real mask boundary, the target area is complete and the boundary is clear.

[0147] As Figure 10The system for implementing the reconstruction method of the cavity under the shielding of the shallow reinforcement mesh comprises a scene construction module, a data acquisition module, a data processing module, a model construction module, a model training module and a cavity reconstruction module; the scene construction module, the data acquisition module, the data processing module, the model construction module, the model training module and the cavity reconstruction module are sequentially connected; the scene construction module is used for constructing a layered medium scene containing a shallow reinforcement mesh and a cavity, and uploading data information to the data acquisition module; the data acquisition module is used for acquiring image data information with different shielding states based on the received data information and the constructed scene, and uploading data information to the data processing module; the data processing module is used for preprocessing the acquired image data information according to the received data information, constructing a reconstruction data set, and uploading data information to the model construction module; the model construction module is used for constructing a reconstruction initial model of the cavity under the shielding of the shallow reinforcement mesh according to the received data information, and uploading data information to the model training module; wherein the reconstruction initial model of the cavity under the shielding of the shallow reinforcement mesh comprises a first-stage model and a second-stage model; the first-stage model is constructed based on a U-shaped network, a Sobel operator, a residual structure and a skip connection scheme; the first-stage model is used for suppressing the echo of the reinforcement mesh; the second-stage model is constructed based on a U-shaped network, a residual structure, an attention mechanism and a skip connection scheme; the second-stage model is used for reconstructing the cavity according to the output of the first-stage model; the model training module is used for training the constructed reconstruction initial model of the cavity under the shielding of the shallow reinforcement mesh by using the obtained reconstruction data set according to the received data information, obtaining a reconstruction model of the cavity under the shielding of the shallow reinforcement mesh, and uploading data information to the cavity reconstruction module; and the cavity reconstruction module is used for reconstructing the actual cavity under the shielding of the shallow reinforcement mesh by using the obtained reconstruction model of the cavity under the shielding of the shallow reinforcement mesh according to the received data information.

Claims

1. A method for reconstructing voids under shallow steel mesh concealment, characterized in that... Includes the following steps: S1. Construct a layered medium scenario containing shallow steel mesh and voids; S2. Based on the scene constructed in step S1, obtain image data information with different occlusion states; S3. Preprocess the image data information obtained in step S2 to construct a reconstructed dataset; S4. Construct an initial reconstruction model of the voids concealed by shallow steel mesh; The initial model for reconstructing cavities under shallow steel mesh includes a first-stage model and a second-stage model; A first-stage model is constructed based on a U-shaped network, the Sobel operator, residual structures, and a skip connection scheme; the first-stage model is used to suppress the echo of the steel mesh. A second-stage model is constructed based on U-shaped networks, residual structures, attention mechanisms, averaging mechanisms, multi-scale edge fusion, and skip connection schemes. The second-stage model is used to reconstruct holes based on the output of the first-stage model. S5. Using the reconstructed dataset obtained in step S3, train the initial reconstruction model of the void under the shallow steel mesh constructed in step S4 to obtain the reconstruction model of the void under the shallow steel mesh. S6. Using the reconstruction model of the void under the shallow steel mesh obtained in step S5, the actual reconstruction of the void under the shallow steel mesh is carried out.

2. The method for reconstructing voids under shallow steel mesh as described in claim 1, characterized in that... Step S1 specifically includes the following steps: The constructed layered medium scenario consists of four layers, from top to bottom: an air layer, a surface layer, a base layer containing steel mesh echoes, and a soil base layer; The surface layer is made of asphalt concrete, and is divided into three layers according to the different aggregate sizes, shapes, and random distribution of aggregate positions. The base material is set to reinforced concrete, and the spacing and number of steel bars are set. The base is divided into two layers according to the different aggregate sizes and shapes. The subgrade is set to a randomly mixed soil, and the soil surface is set to an uneven, undulating surface.

3. The method for reconstructing voids under shallow steel mesh as described in claim 2, characterized in that... Step S2 specifically includes the following steps: Based on the scenario constructed in step S1, several forward model files containing hollow targets are generated. The types of cavities include rectangular cavities without water, rectangular cavities with water, irregularly shaped cavities without water, and irregularly shaped cavities with water; the cavities are set in the soil base layer; the size and depth of the cavities are randomly generated. The obtained forward model files include B-scan images of underground cavities with shallow steel mesh covering them, B-scan images of underground cavities without shallow steel mesh covering them, and black-and-white mask images of the cavity targets. Step S3 specifically includes the following steps: Set the maximum reference threshold for the B-scan image; Signal suppression processing: In the image data information obtained in step S2, all data higher than the maximum reference threshold are replaced with the maximum reference threshold to complete the suppression of strong signals; Signal enhancement processing: Construct an exponential gain sequence that increases with depth, and multiply the data after signal suppression by the constructed exponential gain sequence to complete the signal enhancement processing; Image scaling: The image data after signal enhancement is uniformly scaled to a set size; Complete the preprocessing of the image data information obtained in step S2.

4. The method for reconstructing voids under shallow steel mesh as described in claim 3, characterized in that... The first-stage model is constructed, including the first-stage encoder, the first-stage decoder, and the first-stage filtering module; The first-stage encoder includes a first encoding first residual module, a first encoding first edge module, a first encoding first max pooling module, a first encoding second residual module, a first encoding second edge module, a first encoding second max pooling module, a first encoding third residual module, a first encoding third edge module, a first encoding third max pooling module, a first encoding fourth residual module, a first encoding fourth edge module, a first encoding fourth max pooling module, and a first encoding fifth residual module, which are connected in series. The first-stage decoder comprises, in sequence, a first decoding first transposed convolutional module, a first decoding first residual module, a first decoding first edge module, a first decoding second transposed convolutional module, a first decoding second residual module, a first decoding second edge module, a first decoding third transposed convolutional module, a first decoding third residual module, a first decoding third edge module, a first decoding fourth transposed convolutional module, a first decoding fourth residual module, and a first decoding fourth edge module. Simultaneously, the output of the first encoding fourth edge module and the output of the first decoding first residual module are concatenated and used as the input of the first decoding first edge module; the output of the first encoding third edge module and the output of the first decoding second residual module are concatenated and used as the input of the first decoding second edge module; the output of the first encoding second edge module and the output of the first decoding third residual module are concatenated and used as the input of the first decoding third edge module; and the output of the first encoding first edge module and the output of the first decoding fourth residual module are concatenated and used as the input of the first decoding fourth edge module. The first-stage filtering module includes a first-stage convolutional layer and a first-stage activation function layer connected in series, and the output of the first-stage decoder is the input of the first-stage filtering module; The first-stage model is used to suppress the echo of the rebar mesh and output a GPR image without the echo of the rebar mesh.

5. The method for reconstructing voids under shallow steel mesh as described in claim 4, characterized in that... The first encoding first edge module to the first encoding fourth edge module, and the first decoding first edge module to the first decoding fourth edge module all have the same structure; The processing of the edge module includes the following steps: The Sobel operator is used to detect the horizontal and vertical edge information of the input image, and the absolute values ​​are taken to obtain the absolute horizontal and absolute vertical edge information respectively. After concatenating the obtained absolute horizontal edge information and absolute vertical edge information with the input image information, ... The 2D convolutional layers are fused together, and then processed sequentially through a normalization layer and a ReLU activation function layer to obtain the output of the edge module.

6. The method for reconstructing voids under shallow steel mesh as described in claim 5, characterized in that... The constructed second-stage model includes a second-stage encoder, a second-stage decoder, a second-stage averaging module, and a second-stage reconstruction module; The second-stage encoder includes a second encoding first residual module, a second encoding first max pooling module, a second encoding second residual module, a second encoding second max pooling module, a second encoding third residual module, a second encoding third max pooling module, a second encoding fourth residual module, a second encoding fourth max pooling module, and a second encoding fifth residual module, which are connected in series. The second-stage encoder includes a second decoding first transposed convolutional module, a second decoding first residual module, a second decoding second transposed convolutional module, a second decoding second residual module, a second decoding third transposed convolutional module, a second decoding third residual module, a second decoding fourth transposed convolutional module, and a second decoding fourth residual module, all connected in series. Simultaneously, the output of the second decoding fourth residual module is processed by a fourth attention module and then concatenated with the output of the second decoding fifth residual module, serving as the input to the second decoding first transposed convolutional module; the output of the second decoding third residual module is processed by a third attention module and then concatenated with the output of the second decoding first residual module, serving as the input to the second decoding second transposed convolutional module. The output of the second encoding second residual module is processed by the second attention module and then concatenated with the output of the second decoding second residual module as the input of the second decoding third transposed convolution module; the output of the second encoding first residual module is processed by the first attention module and then concatenated with the output of the second decoding third residual module as the input of the second decoding fourth transposed convolution module. The second-stage averaging module includes a second-averaging first convolutional layer, a second-averaging second convolutional layer, a second-averaging third convolutional layer, and a second-averaging fourth convolutional layer; the output of the second encoding first residual module is processed by the second-averaging first convolutional layer to obtain the second-averaged first signal; The output of the second encoding second residual module is processed by the second averaging second convolutional layer to obtain the second average second signal; The output of the second encoding third residual module is processed by the second averaging third convolutional layer to obtain the second average third signal; the output of the second encoding fourth residual module is processed by the second averaging fourth convolutional layer to obtain the second average fourth signal; the average of the obtained second average first signal to second average fourth signal is calculated to obtain the output of the second stage averaging module; The second-stage reconstruction module includes reconstructing the convolutional layer; the output of the second-stage decoder is processed by the reconstructed convolutional layer and then summed with the output of the second-stage averaging module to obtain the final dilated reconstruction result.

7. The method for reconstructing voids under shallow steel mesh as described in claim 6, characterized in that... The structures of the first attention module through the fourth attention module are all the same; The attention module's processing steps include the following: Obtain the gated feature map of the input image and jump feature map Gated feature map The feature map output by the second decoding first residual module, the skip feature map The feature map output by the second encoding and third residual modules; Will Passing in sequence The image is processed by 2D convolutional layers and normalization layers to obtain the first image component; Passing in sequence The second image component is obtained by processing 2D convolutional layers and normalization layers; After summing the first and second image components, the image is then passed sequentially through a ReLU activation function layer. The image is processed by 2D convolutional layers, normalization layers, and sigmoid activation function layers to obtain the third image component; Combine the third image component with After multiplying by pixels, the output of the attention module is obtained.

8. The method for reconstructing voids under shallow steel mesh as described in claim 7, characterized in that... The structures of the first encoding first residual module to the first encoding fifth residual module, the first decoding first residual module to the first decoding fourth residual module, the second encoding first residual module to the second encoding fifth residual module, and the second decoding first residual module to the second decoding fourth residual module are all the same; The residual module processing procedure includes the following steps: The input image data is divided into two paths: the first path passes through sequentially... 2D convolutional layers, normalization layers, ReLU activation function layers, The first path features are obtained by processing the 2D convolutional layer and normalization layer; The second route passes through in sequence. The second path features are obtained by processing the 2D convolutional layer and normalization layer; The first and second features are added together, and then processed through a ReLU activation function layer to obtain the output of the residual module.

9. The method for reconstructing voids under shallow steel mesh as described in claim 8, characterized in that... The first stage of the convolutional layer is... The 2D convolutional layer; the first-stage activation function layer is a ReLU activation function layer; the first decoding first transposed convolutional module to the first decoding fourth transposed convolutional module are all... The transposed convolutional layer is used to amplify the resolution; the first encoding first max pooling module to the first encoding fourth max pooling module are all... A max-pooling layer with a stride of 2 is used to halve the size of the input feature map; Reconstruct the convolutional layer as The 2D convolutional layer; the reconstructed activation function layer is a Sigmoid activation function layer; the second encoding first max pooling module to the second encoding fourth max pooling module are all... The first module is a max pooling module with a stride of 2; the second decoding first transposed convolution module and the second decoding first transposed convolution module are both... The transposed convolutional layer; the second average first convolutional layer to the second average fourth convolutional layer are all... 2D convolutional layers; The training described in step S5 specifically includes the following steps: During training, the pre-processed B-scan image of underground cavities with shallow steel mesh covering them was used as the input of the first-stage model, and the pre-processed B-scan image of underground cavities without shallow steel mesh covering them was used as the label for the input of the first-stage model. The first-stage model was then trained. Then, the output of the first-stage model is used as the input of the second-stage model, and the preprocessed black-and-white mask image of the hole target is used as the label of the input of the second-stage model to train the second-stage model. The loss function of the first-stage model is a weighted sum of the L1 loss function and the SSIM loss function; The loss function of the second-stage model is a weighted sum of the BCE loss function, Dice loss function, Boundary loss function, and Focal loss function; During training, the weighted sum of the loss functions of the first-stage model and the second-stage model is used as the total loss function. During training, the Adam optimizer is used, and an adaptive learning rate mechanism and an early stopping strategy are employed to improve the stability and efficiency of the training process.

10. A system for reconstructing a shallow steel mesh concealing a void as described in any one of claims 1 to 9, characterized in that... It includes a scene construction module, a data acquisition module, a data processing module, a model construction module, a model training module, and a void reconstruction module; these modules are connected in series. The scene construction module is used to construct a layered medium scene containing shallow steel mesh and voids, and uploads the data information to the data acquisition module. The data acquisition module is used to acquire image data information with different occlusion states based on the constructed scene and the received data information, and uploads the data information to the data processing module. The data processing module is used to preprocess the acquired image data based on the received data information to construct a reconstructed dataset and upload the data information to the model building module; The model building module is used to construct an initial reconstruction model of voids under shallow rebar mesh shading based on the received data information, and upload the data information to the model training module. The initial reconstruction model of voids under shallow rebar mesh shading includes a first-stage model and a second-stage model. The first-stage model is constructed based on a U-shaped network, Sobel operator, residual structure, and skip connection scheme; the first-stage model is used to suppress the echo of the rebar mesh. The second-stage model is constructed based on a U-shaped network, residual structure, attention mechanism, averaging mechanism, multi-scale edge fusion, and skip connection scheme; the second-stage model is used to reconstruct voids based on the output of the first-stage model. The model training module is used to train the initial reconstruction model of the voids under the shallow steel mesh based on the received data information and the obtained reconstruction dataset, so as to obtain the reconstruction model of the voids under the shallow steel mesh and upload the data information to the void reconstruction module; the void reconstruction module is used to reconstruct the actual voids under the shallow steel mesh based on the received data information and the obtained reconstruction model of the voids under the shallow steel mesh.

Images