Underground object three-dimensional reconstruction method and device

Abstract

The invention relates to the technical field of infrastructure nondestructive testing, in particular to an underground object three-dimensional reconstruction method and device, and the method comprises the steps: constructing a target data set, a neural network architecture and a target loss function; using the target data set, the neural network architecture and the target loss function to construct a target neural network model for three-dimensional reconstruction of the underground target body; and inputting the three-dimensional ground penetrating radar signal data into the target neural network model so as to output a three-dimensional reconstruction result of the underground target body corresponding to the three-dimensional ground penetrating radar signal data through the target neural network model. The method can achieve the end-to-end accurate recognition of the underground pipeline and the cavity, achieves the precise classification of the types of the underground target body, can reconstruct the geometric shape of the underground target body, remarkably improves the efficiency and accuracy of underground space detection, can be widely applied to the post-processing stage of ground penetrating radar data and various engineering scenes, and is high in practicality. And the safety and the management efficiency of the urban infrastructure can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of nondestructive testing of infrastructure, and in particular to a method and device for three-dimensional reconstruction of underground objects. Background Art

[0002] Road underground space is an important part of the city. In urban operation and maintenance, there are two types of road underground targets that need to be monitored: one is the various pipelines that ensure the operation of the city; the other is various diseases such as voids and cavities caused by construction quality, geological conditions, environmental factors, and abnormal loads.

[0003] Among related technologies, traditional detection methods mostly rely on manual analysis and two-dimensional signal interpretation, which makes it difficult to accurately restore the three-dimensional geometric shape of underground targets. Especially in complex geological environments, traditional methods have problems such as low accuracy and low efficiency. Ground penetrating radar (GPR), as a non-destructive testing technology, has the characteristics of strong penetration ability and sensitive signal response.

[0004] However, ground penetrating radar signal processing technology in related technologies still faces many challenges in three-dimensional reconstruction, such as insufficient target shape modeling, classification bias caused by data imbalance, and poor geometric consistency. Most of them require manual operation, and the detection efficiency and accuracy have certain shortcomings, which need to be solved urgently. Summary of the invention

[0005] The present application provides a method and device for three-dimensional reconstruction of underground objects to solve the problems that ground penetrating radar signal processing technology in related technologies still faces many challenges in three-dimensional reconstruction, such as insufficient target shape modeling, classification bias caused by data imbalance, and poor geometric consistency. In addition, most of them require manual operation, and the detection efficiency and accuracy are both insufficient.

[0006] The first aspect of the present application provides a method for three-dimensional reconstruction of underground objects, including the following steps: constructing a target data set, a neural network architecture and a target loss function; using the target data set, the neural network architecture and the target loss function to construct a target neural network model for the purpose of three-dimensional reconstruction of the underground target body; inputting three-dimensional ground penetrating radar signal data into the target neural network model to output the three-dimensional reconstruction result of the underground target body corresponding to the three-dimensional ground penetrating radar signal data through the target neural network model.

[0007] Optionally, in one embodiment of the present application, the construction of the target data set, the neural network architecture and the target loss function includes: obtaining simulated ground penetrating radar signal data of a preset typical underground target body and performing data preprocessing to obtain preprocessed data; constructing a three-dimensional matrix corresponding to the three-dimensional model of the typical underground target body, and extracting the material spatial distribution label of the typical underground target body; and constructing the target data set based on the preprocessed data, the three-dimensional matrix and the material spatial distribution label.

[0008] Optionally, in one embodiment of the present application, the target neural network model for the purpose of three-dimensional reconstruction of the underground target body is constructed using the target data set, the neural network architecture and the target loss function, including: inputting the preprocessed data into the initial neural network model to output the material category of each voxel in the underground space corresponding to the simulated ground penetrating radar signal data to construct the target neural network model.

[0009] Optionally, in one embodiment of the present application, the three-dimensional reconstruction result of the underground target body corresponding to the three-dimensional ground penetrating radar signal data is output through the target neural network model, including: identifying the type information and geometric information of the underground target body corresponding to the three-dimensional ground penetrating radar signal data through the mapping model in the neural network architecture; and determining the three-dimensional reconstruction result based on the type information and the geometric information.

[0010] Optionally, in one embodiment of the present application, the target neural network model for the purpose of three-dimensional reconstruction of an underground target body is constructed using the target data set, the neural network architecture and the target loss function, including: based on a target encoding-decoding architecture, constructing a target three-dimensional convolutional neural network architecture; based on the target three-dimensional convolutional neural network architecture, determining the neural network architecture to construct the target neural network model.

[0011] Optionally, in one embodiment of the present application, the objective loss function is: , in, represents the loss value, Indicates The loss value of each voxel, is the number of categories, is the scale factor, is the total variation regularization term.

[0012] The second aspect of the present application provides a three-dimensional reconstruction device for underground objects, including: a first construction module, used to construct a target data set, a neural network architecture and a target loss function; a second construction module, used to use the target data set, the neural network architecture and the target loss function to construct a target neural network model for the purpose of three-dimensional reconstruction of the underground target body; a reconstruction module, used to input three-dimensional ground penetrating radar signal data into the target neural network model, so as to output the three-dimensional reconstruction result of the underground target body corresponding to the three-dimensional ground penetrating radar signal data through the target neural network model.

[0013] Optionally, in one embodiment of the present application, the first construction module includes: an acquisition unit, used to acquire simulated ground penetrating radar signal data of a preset typical underground target body and perform data preprocessing to obtain preprocessed data; an extraction unit, used to construct a three-dimensional matrix corresponding to the three-dimensional model of the typical underground target body, and extract the material spatial distribution label of the typical underground target body; a first construction unit, used to construct the target data set based on the preprocessed data, the three-dimensional matrix and the material spatial distribution label.

[0014] Optionally, in one embodiment of the present application, the second construction module includes: a second construction unit, used to input the preprocessed data into the initial neural network model to output the material category of each voxel in the underground space corresponding to the simulated ground penetrating radar signal data to construct the target neural network model.

[0015] Optionally, in one embodiment of the present application, the reconstruction module includes: an identification unit for identifying type information and geometric information of an underground target body corresponding to the three-dimensional ground penetrating radar signal data through a mapping model in the neural network architecture; and a determination unit for determining the three-dimensional reconstruction result based on the type information and the geometric information.

[0016] Optionally, in one embodiment of the present application, the second construction module includes: a third construction unit, used to construct a target three-dimensional convolutional neural network architecture based on the target encoding-decoding architecture; and a fourth construction unit, used to determine the neural network architecture based on the target three-dimensional convolutional neural network architecture to construct the target neural network model.

[0017] Optionally, in one embodiment of the present application, the objective loss function is: , in, represents the loss value, Indicates The loss value of each voxel, is the number of categories, is the scale factor, is the total variation regularization term.

[0018] The third aspect of the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method for three-dimensional reconstruction of underground objects as described in the above embodiment.

[0019] The fourth aspect of the present application provides a computer-readable storage medium, which stores a computer program. When the program is executed by a processor, it implements the above-mentioned method for three-dimensional reconstruction of underground objects.

[0020] The fifth aspect of the present application provides a computer program product, including a computer program, which, when executed, is used to implement the above-mentioned method for three-dimensional reconstruction of underground objects.

[0021] The embodiment of the present application can construct a target neural network model based on a target data set, a neural network architecture, and a target loss function, so as to input the three-dimensional ground penetrating radar signal data into the target neural network model, thereby outputting the three-dimensional reconstruction result of the underground target body corresponding to the three-dimensional ground penetrating radar signal data. Thus, it is realized that through engineering scene investigation, the initial neural network model is trained by using simulation tools to generate ground penetrating radar simulation signal data with high accuracy, thereby improving the practical application ability of the target neural network model; and combined with the three-dimensional deep convolutional neural network architecture, the three-dimensional reconstruction performance of the target neural network is effectively improved through residual connection and jump connection. Further, the present application can also further enhance the geometric consistency of the three-dimensional reconstruction result by combining the composite loss function of Dice loss and total variation regularization term; finally, the present application can realize end-to-end accurate recognition of underground pipelines and cavities, and while realizing accurate classification of underground target body categories, it can also reconstruct the geometric shape of underground target bodies, significantly improve the efficiency and accuracy of underground space detection, and can be widely used in the data post-processing stage of two-dimensional or three-dimensional ground penetrating radars, as well as various engineering scenarios such as urban infrastructure management, road monitoring, and underground disease assessment, which helps to improve the safety and management efficiency of urban infrastructure. This solves the problem that ground penetrating radar signal processing technology in related technologies still faces many challenges in three-dimensional reconstruction, such as insufficient target shape modeling, classification bias caused by data imbalance, and poor geometric consistency. In addition, most of them require manual operation, and have certain deficiencies in detection efficiency and accuracy.

[0022] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which: Figure 1 A flowchart of a method for three-dimensional reconstruction of underground objects provided according to an embodiment of the present application; Figure 2 A flowchart of 3D reconstruction of underground pipelines and cavities according to one embodiment of the present application; Figure 3 A schematic diagram of a neural network architecture according to an embodiment of the present application; Figure 4 A schematic diagram of the structure of a three-dimensional reconstruction device for underground objects provided according to an embodiment of the present application; Figure 5 It is a schematic diagram of the structure of an electronic device provided according to an embodiment of the present application.

[0024] Reference numerals: 10-underground object three-dimensional reconstruction device: 100-first construction module, 200-second construction module and 300-reconstruction module; 501-memory, 502-processor and 503-communication interface. DETAILED DESCRIPTION

[0025] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0026] The following describes the method and device for three-dimensional reconstruction of underground objects in the embodiments of the present application with reference to the accompanying drawings. In view of the fact that the ground-penetrating radar signal processing technology in the related technologies mentioned in the above background technology still faces many challenges in three-dimensional reconstruction, such as insufficient target body shape modeling, classification bias caused by data imbalance, and poor geometric consistency, and most of them require manual operation, and the detection efficiency and accuracy have certain defects, the present application provides a method for three-dimensional reconstruction of underground objects, in which a target neural network model can be constructed based on a target data set, a neural network architecture, and a target loss function, so as to input the three-dimensional ground-penetrating radar signal data into the target neural network model, thereby outputting the three-dimensional reconstruction result of the underground target body corresponding to the three-dimensional ground-penetrating radar signal data. Thus, it is realized that through the investigation of engineering scenes, the simulation tool is used to generate ground-penetrating radar simulation signal data with high accuracy to train the initial neural network model, and the practical application ability of the target neural network model is improved; and combined with the three-dimensional deep convolutional neural network architecture, the three-dimensional reconstruction performance of the target neural network is effectively improved through residual connection and jump connection. Furthermore, the present application can also further enhance the geometric consistency of the three-dimensional reconstruction result by combining the composite loss function of Dice loss and total variation regularization term; finally, the present application can realize the end-to-end accurate recognition of underground pipelines and cavities, and while realizing the accurate classification of underground target body categories, it can also reconstruct the geometric shape of underground target bodies, significantly improve the efficiency and accuracy of underground space detection, and can be widely used in the data post-processing stage of two-dimensional or three-dimensional ground-penetrating radar and various engineering scenes such as urban infrastructure management, road monitoring and underground disease assessment, which is helpful to improve the safety and management efficiency of urban infrastructure. Thus, it solves the problem that the ground-penetrating radar signal processing technology in the related technology still faces many challenges in three-dimensional reconstruction, such as insufficient target body shape modeling, classification bias caused by data imbalance, and poor geometric consistency, and most of them require manual operation, and the detection efficiency and accuracy are all insufficient.

[0027] Specifically, Figure 1 A flowchart of a method for three-dimensional reconstruction of underground objects provided in an embodiment of the present application.

[0028] like Figure 1 As shown, the underground object three-dimensional reconstruction method includes the following steps: In step S101, a target data set, a neural network architecture, and a target loss function are constructed.

[0029] In some embodiments, when the present application performs three-dimensional reconstruction of underground objects, in order to facilitate practical application, the implementation process can be modeled as a model, but is not limited to. When constructing the model, it can be constructed based on a certain deep learning neural network architecture, trained and tested using a target data set, and a certain target loss function is designed to improve the accuracy of the final three-dimensional reconstruction result.

[0030] Here, the target data set can be understood as a collection of various data used in modeling, which can be used to train and test the model.

[0031] Next, the process of constructing the target data set in the embodiment of the present application is further explained.

[0032] Optionally, in one embodiment of the present application, a target data set, a neural network architecture and a target loss function are constructed, including: obtaining simulated ground penetrating radar signal data of a preset typical underground target body and performing data preprocessing to obtain preprocessed data; constructing a three-dimensional matrix corresponding to the three-dimensional model of the typical underground target body, and extracting material spatial distribution labels of the typical underground target body; and constructing a target data set based on the preprocessed data, the three-dimensional matrix and the material spatial distribution labels.

[0033] In the actual execution process, when constructing the target data set, the present application can first obtain the simulated ground penetrating radar signal data of a preset typical underground target body, and then preprocess the data to obtain the preprocessed data as part of the target data set.

[0034] The preset typical underground target bodies here can be understood as some relatively typical underground target bodies determined in advance, such as cavities and underground pipes. In actual application, underground objects in different places may be different, and underground objects that need to be 3D reconstructed may also be different. Therefore, the preset typical underground target bodies can be determined by professional and technical personnel in this field according to actual conditions, thereby improving the practical application ability of the model and the accuracy of the 3D reconstruction results.

[0035] For example, Figure 2 This is a flowchart of 3D reconstruction of underground pipelines and cavities according to an embodiment of the present application. Figure 2 As shown, the present application can first combine the engineering scenario investigation, summarize the rules of pipeline and disease target modeling, use gprMax to carry out typical target radar signal simulation of multiple survey lines, obtain the ground penetrating radar signals of underground pipelines and voids and the corresponding model size information, obtain the two-dimensional B-scan signals and combine them to obtain the three-dimensional simulated ground penetrating radar signal data C-scan data, and then perform normalization preprocessing on them, unify them into the [0,1] interval, and obtain the preprocessed data.

[0036] It should be noted that the ground penetrating radar signal data in the embodiments of the present application can be generated based on but not limited to open source simulation software and actual experiments, and the specific acquisition method can be selected or adjusted according to actual conditions. The embodiments of the present application are only illustrative and not specifically limited.

[0037] Furthermore, the embodiment of the present application can simultaneously extract the spatial distribution of materials such as soil, metal pipes, plastic pipes, concrete pipes and cavities in the input file as different labels.

[0038] Finally, the present application can convert the three-dimensional models of these typical underground target bodies into corresponding three-dimensional matrices, and unify the preprocessed data, material spatial distribution labels, and three-dimensional matrices into data of specified dimensions to form a target data set, wherein the specific dimensions can be selected or adjusted by professionals and technicians in this field, which is only an exemplary description here without specific restrictions. Also, it should be noted that the preprocessed data, material spatial distribution labels, and three-dimensional matrices that constitute the target data set are different, and the target data sets they constitute are also different, which is only an exemplary description in the embodiments of the present application without specific restrictions.

[0039] Step S102, using the target data set, the neural network architecture and the target loss function to construct a target neural network model for the purpose of three-dimensional reconstruction of the underground target body. The target loss function can be, but is not limited to, expressed as: , in, represents the loss value, Indicates The loss value of each voxel, is the number of categories, is the scale factor, is the total variation regularization term.

[0040] Based on the relevant descriptions of other embodiments, it can be understood that, to facilitate practical applications, the embodiments of the present application can use the target data set, neural network architecture and target loss function to model the process of three-dimensional reconstruction of underground objects into a certain model.

[0041] In the actual implementation process, the present application can use the target data set, neural network architecture and target loss function to construct a target neural network model for the purpose of three-dimensional reconstruction of underground target bodies. Among them, the target neural network model can be different according to the target data set, neural network architecture and target loss function.

[0042] Among them, the target loss function in the embodiment of the present application can be constructed by using, but not limited to, an improved Dice loss function. Specifically, the embodiment of the present application can use an improved Dice loss function and combine it with a total variation regularization term to optimize the target neural network model.

[0043] For example, the Dice loss calculation formula can be, but is not limited to, expressed as:

[0044] in, It is The prediction results of the voxels, It is The actual result of the voxels, is the total number of voxels. In order to enhance the geometric continuity of the prediction result, the embodiment of the present application may also add a total variation regularization term to the loss function. Considering that the prediction result is a discrete matrix, the total variation regularization term in the embodiment of the present application may be, but is not limited to, expressed as:

[0045] in, Refers to the location Voxel prediction value. The final objective loss function can be, but is not limited to, expressed as:

[0046] in, represents the loss value, Indicates The loss value of each voxel, is the number of categories. In actual use, the label matrix can be uniquely encoded, and then the loss value can be calculated for each category. For example, in the embodiment of the present application, it can be taken as 5; Is a scale factor. In order to avoid the adverse effect of too large regularization term on training, its value can be set to the number of all elements involved in the calculation. The formula can be but is not limited to the following:

[0047] in, , and Represent the sizes of the three dimensions of the label matrix respectively.

[0048] It should be noted that the prediction results in the embodiments of the present application refer to the distribution of materials in three-dimensional space corresponding to the three-dimensional simulated ground penetrating radar signal C-scan data in the process of building the target neural network model.

[0049] The target loss function of the target neural network model in the embodiment of the present application adopts the Dice loss function with an added total variation regularization term constraint, and the weight of the regularization term is designed to be the inverse of the number of elements of the output matrix, thereby optimizing the target neural network model to enhance the accuracy and geometric consistency of the three-dimensional reconstruction of the target neural network model.

[0050] Optionally, in one embodiment of the present application, a target neural network model for the purpose of three-dimensional reconstruction of an underground target body is constructed using a target data set, a neural network architecture, and a target loss function, including: inputting the preprocessed data into an initial neural network model to output the material category of each voxel in the underground space corresponding to the simulated ground penetrating radar signal data to construct a target neural network model.

[0051] Based on the relevant descriptions of other embodiments, it can be understood that the embodiments of the present application can use the target data set, neural network architecture and target loss function to construct a target neural network model. In the process of training the target neural network model, certain prediction results will be obtained, that is, the distribution of materials in three-dimensional space corresponding to the three-dimensional simulated ground penetrating radar signal C-scan data.

[0052] In the actual implementation process, still refer to Figure 2 ,like Figure 2 As shown, the present application can input the preprocessed data, i.e., the normalized three-dimensional simulated ground penetrating radar signal C-scan data, into the initial neural network model, and then output the material category of each voxel in the underground space corresponding to the simulated ground penetrating radar signal data through the initial neural network model, i.e., the distribution of the material corresponding to the three-dimensional simulated ground penetrating radar signal C-scan data in the three-dimensional space, thereby completing the training of the initial neural network model to construct the target neural network model.

[0053] Optionally, in one embodiment of the present application, a target neural network model for the purpose of three-dimensional reconstruction of an underground target body is constructed using a target data set, a neural network architecture, and a target loss function, including: based on a target encoding-decoding architecture, constructing a target three-dimensional convolutional neural network architecture; based on the target three-dimensional convolutional neural network architecture, determining a neural network architecture to construct a target neural network model.

[0054] As a possible implementation method, the neural network architecture for constructing the target neural network model in the embodiments of the present application can be, but is not limited to, a target "U"-shaped three-dimensional convolutional neural network architecture based on a target encoding-decoding architecture. Among them, the target encoding-decoding architecture can be understood as a related encoding-decoding architecture that can meet the construction of the target neural network model, for example, the U-Net encoding-decoding structure or the SegNet encoding-decoding structure in the convolutional neural network (CNN); the embodiments of the present application are only for illustrative purposes and are not specifically limited.

[0055] For example, Figure 3 This is a schematic diagram of a neural network architecture of an embodiment of the present application, such as Figure 3 As shown, the target "U"-shaped three-dimensional convolutional neural network architecture based on the target encoding-decoding architecture in the embodiment of the present application can be composed of, but not limited to, "conv", "n", "lrelu", "drop", "upscale" and "concat". Among them, "conv", "n", "lrelu", "drop", "upscale" and "concat" represent a three-dimensional convolution layer, an instance normalization layer, a leaky corrected linear unit (Leaky ReLU) activation layer, a random drop layer, an upsampling layer and a cascade operation, respectively. And, the "+" in the legend indicates a continuous combination of each operation.

[0056] Furthermore, jump connections and long-distance parallel connections are added to the neural network architecture in the embodiment of the present application. Among them, the jump connection refers to a short-circuit structure that directly adds the transformed tensor to the input tensor. Two convolution modules are used in the embodiment of the present application. The convolution kernel size is 3×3×3, padded with 1×1×1, and the strides of "conv" and "conv" are 1×1×1 and 2×2×2, respectively. Therefore, the first one will not change the size of the input tensor, but the second one will reduce the size by half.

[0057] The number of channels of the input tensor is 1, and the number of channels is changed to , and then use 5 encoders to change the number of tensor channels to , and the size is also reduced to half of the input layer by layer. The decoder restores the feature map to twice the original input size layer by layer, while reducing the number of channels. After the output is finally converted to the original size, the softmax activation function can be used for normalization to obtain the prediction result and complete the construction of the target neural network model.

[0058] Step S103, inputting the three-dimensional ground penetrating radar signal data into the target neural network model, so as to output the three-dimensional reconstruction result of the underground target body corresponding to the three-dimensional ground penetrating radar signal data through the target neural network model.

[0059] In other embodiments, after the target neural network model is constructed, the embodiments of the present application can use the target neural network model to complete the three-dimensional reconstruction of the actual underground target body.

[0060] Among them, the underground target body here can be understood as the underground object that needs to be reconstructed in three dimensions. For example, in urban operation and maintenance work, it is necessary to focus on monitoring various pipelines that ensure the operation of the city and various diseases such as voids and cavities caused by construction quality, geological conditions, environmental factors, and abnormal loads, that is, underground pipelines and cavities are two types of underground road target bodies.

[0061] Specifically, the embodiment of the present application can input the three-dimensional ground penetrating radar signal data corresponding to the underground target body to be three-dimensionally reconstructed into the target neural network model, and then output the three-dimensional reconstruction result of the underground target body corresponding to the three-dimensional ground penetrating radar signal data through the target neural network model.

[0062] The 3D reconstruction result includes, but is not limited to, the material category of each location in the underground space (similar to the output result of 3D segmentation). Furthermore, through the material category of each location, the embodiment of the present application can identify the type of underground target body at each location and determine the geometric shape of the underground target body.

[0063] Optionally, in one embodiment of the present application, a three-dimensional reconstruction result of an underground target body corresponding to the three-dimensional ground penetrating radar signal data is output through a target neural network model, including: identifying type information and geometric information of the underground target body corresponding to the three-dimensional ground penetrating radar signal data through a mapping model in a neural network architecture; and determining the three-dimensional reconstruction result based on the type information and geometric information.

[0064] In some embodiments, in order to improve the accuracy of the three-dimensional reconstruction results, the present application can also build a mapping model between radar signals and underground space target category distribution in the neural network architecture used to construct the target neural model, so as to use the mapping model to identify the type information and geometric information of the underground target body corresponding to the three-dimensional ground penetrating radar signal data, thereby achieving accurate reconstruction of the type and geometric information of the underground space target body.

[0065] Therefore, the embodiment of the present application can perform geometric three-dimensional reconstruction of the underground target body while identifying the category of each voxel of the underground target body. Therefore, the three-dimensional reconstruction of the underground target body in the embodiment of the present application can essentially be understood as voxel reconstruction.

[0066] Here, voxel refers to the smallest unit used to represent and quantify a 3D object or space in the context of 3D reconstruction. Voxel has its own position and usually has one or more attributes related to the reconstruction task, such as density, color, material, etc.

[0067] According to the method for three-dimensional reconstruction of underground objects proposed in the embodiment of the present application, a target neural network model can be constructed based on a target data set, a neural network architecture, and a target loss function, so as to input three-dimensional ground penetrating radar signal data into the target neural network model, thereby outputting the three-dimensional reconstruction result of the underground target body corresponding to the three-dimensional ground penetrating radar signal data. Thus, it is realized that through engineering scene investigation, the initial neural network model is trained by using simulation tools to generate ground penetrating radar simulation signal data with high accuracy, thereby improving the practical application ability of the target neural network model; and combined with the three-dimensional deep convolutional neural network architecture, the three-dimensional reconstruction performance of the target neural network is effectively improved through residual connection and jump connection. Furthermore, the present application can also further enhance the geometric consistency of the three-dimensional reconstruction result by combining the composite loss function of Dice loss and total variation regularization term; finally, the present application can realize end-to-end accurate recognition of underground pipelines and cavities, and while realizing accurate classification of underground target body categories, it can also reconstruct the geometric shape of underground target bodies, significantly improve the efficiency and accuracy of underground space detection, and can be widely used in the data post-processing stage of two-dimensional or three-dimensional ground penetrating radars, as well as various engineering scenarios such as urban infrastructure management, road monitoring, and underground disease assessment, which helps to improve the safety and management efficiency of urban infrastructure. This solves the problem that ground penetrating radar signal processing technology in related technologies still faces many challenges in three-dimensional reconstruction, such as insufficient target shape modeling, classification bias caused by data imbalance, and poor geometric consistency. In addition, most of them require manual operation, and have certain deficiencies in detection efficiency and accuracy.

[0068] Next, the three-dimensional reconstruction device for underground objects proposed in accordance with an embodiment of the present application will be described with reference to the accompanying drawings.

[0069] Figure 4 It is a schematic diagram of the structure of the underground object three-dimensional reconstruction device according to an embodiment of the present application.

[0070] like Figure 4 As shown, the underground object three-dimensional reconstruction device 10 includes: a first construction module 100, a second construction module 200 and a reconstruction module 300.

[0071] Among them, the first construction module 100 is used to construct a target data set, a neural network architecture and a target loss function.

[0072] The second construction module 200 is used to construct a target neural network model for the purpose of three-dimensional reconstruction of underground target bodies by using the target data set, the neural network architecture and the target loss function.

[0073] The reconstruction module 300 is used to input the three-dimensional ground penetrating radar signal data into the target neural network model to output the three-dimensional reconstruction result of the underground target body corresponding to the three-dimensional ground penetrating radar signal data through the target neural network model.

[0074] Optionally, in one embodiment of the present application, the first construction module 100 includes: an acquisition unit, an extraction unit and a first construction unit.

[0075] The acquisition unit is used to acquire simulated ground penetrating radar signal data of a preset typical underground target body and perform data preprocessing to obtain preprocessed data.

[0076] The extraction unit is used to construct a three-dimensional matrix corresponding to the three-dimensional model of a typical underground target body and extract the material spatial distribution label of the typical underground target body.

[0077] The first construction unit is used to construct a target data set based on the preprocessed data, the three-dimensional matrix and the material spatial distribution label.

[0078] Optionally, in one embodiment of the present application, the second construction module 200 includes: a second construction unit, used to input the preprocessed data into the initial neural network model to output the material category of each voxel in the underground space corresponding to the simulated ground penetrating radar signal data to construct a target neural network model.

[0079] Optionally, in one embodiment of the present application, the reconstruction module 300 includes: an identification unit and a determination unit.

[0080] Among them, the identification unit is used to identify the type information and geometric information of the underground target body corresponding to the three-dimensional ground penetrating radar signal data through the mapping model in the neural network architecture.

[0081] The determination unit is used to determine the three-dimensional reconstruction result according to the type information and the geometric information.

[0082] Optionally, in one embodiment of the present application, the second building module 200 includes: a third building unit and a fourth building unit.

[0083] Among them, the third construction unit is used to construct a target three-dimensional convolutional neural network architecture based on the target encoding-decoding architecture.

[0084] The fourth construction unit is used to determine the neural network architecture based on the target three-dimensional convolutional neural network architecture to construct the target neural network model.

[0085] Optionally, in one embodiment of the present application, the target loss function may be, but is not limited to, expressed as: , in, represents the loss value, Indicates The loss value of each voxel, is the number of categories, is the scale factor, is the total variation regularization term.

[0086] It should be noted that the above explanations of the embodiment of the method for three-dimensional reconstruction of underground objects are also applicable to the three-dimensional reconstruction device for underground objects of this embodiment, and will not be repeated here.

[0087] According to the underground object three-dimensional reconstruction device proposed in the embodiment of the present application, a target neural network model can be constructed based on the target data set, the neural network architecture and the target loss function to input the three-dimensional ground penetrating radar signal data into the target neural network model, thereby outputting the three-dimensional reconstruction result of the underground target body corresponding to the three-dimensional ground penetrating radar signal data. Thus, it is realized that through the engineering scene survey, the initial neural network model is trained by using the simulation tool to generate the ground penetrating radar simulation signal data with high accuracy, so as to improve the practical application ability of the target neural network model; and combined with the three-dimensional deep convolutional neural network architecture, the three-dimensional reconstruction performance of the target neural network is effectively improved through the residual connection and the jump connection. Further, the present application can also further enhance the geometric consistency of the three-dimensional reconstruction result by combining the composite loss function of the Dice loss and the total variation regularization term; finally, the present application can realize the end-to-end accurate recognition of underground pipelines and cavities, and at the same time of realizing the accurate classification of the underground target body category, it can also reconstruct the geometric shape of the underground target body, significantly improve the efficiency and accuracy of underground space detection, and can be widely used in the data post-processing stage of two-dimensional or three-dimensional ground penetrating radar and various engineering scenarios such as urban infrastructure management, road monitoring and underground disease assessment, which helps to improve the safety and management efficiency of urban infrastructure. This solves the problem that ground penetrating radar signal processing technology in related technologies still faces many challenges in three-dimensional reconstruction, such as insufficient target shape modeling, classification bias caused by data imbalance, and poor geometric consistency. In addition, most of them require manual operation, and have certain deficiencies in detection efficiency and accuracy.

[0088] Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include: A memory 501 , a processor 502 , and a computer program stored in the memory 501 and executable on the processor 502 .

[0089] When the processor 502 executes the program, the method for three-dimensional reconstruction of underground objects provided in the above embodiment is implemented.

[0090] Furthermore, the electronic device further comprises: The communication interface 503 is used for communication between the memory 501 and the processor 502 .

[0091] The memory 501 is used to store computer programs that can be executed on the processor 502 .

[0092] The memory 501 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0093] If the memory 501, the processor 502 and the communication interface 503 are implemented independently, the communication interface 503, the memory 501 and the processor 502 can be connected to each other through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0094] Optionally, in a specific implementation, if the memory 501, the processor 502 and the communication interface 503 are integrated on a chip, the memory 501, the processor 502 and the communication interface 503 can communicate with each other through an internal interface.

[0095] The processor 502 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0096] An embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned method for three-dimensional reconstruction of underground objects.

[0097] An embodiment of the present application also provides a computer program product, including a computer program, which can run computer instructions. When the computer instructions are executed by a processor, the three-dimensional reconstruction method of underground objects provided in the embodiment of the present application is implemented.

[0098] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0099] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0100] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0101] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection with one or N wirings (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways as necessary and then storing it in a computer memory.

[0102] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above embodiment, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented by hardware, as in another embodiment, it can be implemented by any one or a combination of multiple of the following technologies known in the art: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0103] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0104] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0105] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A method for three-dimensional reconstruction of underground objects, characterized in that: The following steps are involved: Construct the target dataset, neural network architecture, and target loss function; Using the target data set, the neural network architecture and the target loss function, a target neural network model is constructed for the purpose of three-dimensional reconstruction of an underground target body; The three-dimensional ground penetrating radar signal data is input into the target neural network model, so as to output the three-dimensional reconstruction result of the underground target body corresponding to the three-dimensional ground penetrating radar signal data through the target neural network model.

2. The method for three-dimensional reconstruction of underground objects according to claim 1, characterized in that: The construction of the target data set, neural network architecture and target loss function includes: Acquire simulated ground penetrating radar signal data of a preset typical underground target body and perform data preprocessing to obtain preprocessed data; Constructing a three-dimensional matrix corresponding to the three-dimensional model of the typical underground target body, and extracting the material spatial distribution label of the typical underground target body; The target data set is constructed based on the preprocessed data, the three-dimensional matrix and the material spatial distribution labels.

3. The method for 3D reconstruction of underground objects according to claim 2, characterized in that: The method of using the target data set, the neural network architecture and the target loss function to construct a target neural network model for the purpose of three-dimensional reconstruction of an underground target body includes: The preprocessed data is input into an initial neural network model to output the material category of each voxel in the underground space corresponding to the simulated ground penetrating radar signal data to construct the target neural network model.

4. The method for 3D reconstruction of underground objects according to claim 1, characterized in that: The outputting of the three-dimensional reconstruction result of the underground target body corresponding to the three-dimensional ground penetrating radar signal data through the target neural network model includes: Identify the type information and geometric information of the underground target body corresponding to the three-dimensional ground penetrating radar signal data through the mapping model in the neural network architecture; The three-dimensional reconstruction result is determined according to the type information and the geometric information.

5. The method for 3D reconstruction of underground objects according to claim 1, characterized in that: The method of using the target data set, the neural network architecture and the target loss function to construct a target neural network model for the purpose of three-dimensional reconstruction of an underground target body includes: Based on the target encoding-decoding architecture, build the target 3D convolutional neural network architecture; Based on the target three-dimensional convolutional neural network architecture, the neural network architecture is determined to construct the target neural network model.

6. The method for 3D reconstruction of underground objects according to claim 1, characterized in that: The objective loss function is: ， in, represents the loss value, Indicates The loss value of each voxel, is the number of categories, is the scale factor, is the total variation regularization term.

7. A device for three-dimensional reconstruction of underground objects, characterized in that: include: The first building module is used to build the target dataset, neural network architecture and target loss function; A second construction module is used to construct a target neural network model for the purpose of three-dimensional reconstruction of an underground target body by using the target data set, the neural network architecture and the target loss function; The reconstruction module is used to input the three-dimensional ground penetrating radar signal data into the target neural network model to output the three-dimensional reconstruction result of the underground target body corresponding to the three-dimensional ground penetrating radar signal data through the target neural network model.

8. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method for three-dimensional reconstruction of underground objects as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the three-dimensional reconstruction method of underground objects as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed, it is used to implement the three-dimensional reconstruction method of underground objects as described in any one of claims 1-6.

Images