Method for processing underground disease segmentation result based on three-dimensional ground penetrating radar

Abstract

The invention discloses a method for processing an underground disease segmentation result based on a three-dimensional ground penetrating radar. The method comprises the following steps: S1, carrying out overlapping clustering on road sections corresponding to data for segmentation prediction; s2, clustering the spatial sparse point clouds corresponding to all the 3D segmentation masks in the road section of each clustering cluster; and S3, performing duplicate removal on a point cloud clustering result to obtain a disease instance. The method comprises the following steps: carrying out overlapped clustering on road sections corresponding to data used for segmentation prediction, carrying out clustering on spatial sparse point clouds corresponding to all 3D segmentation masks in the road sections of each clustering cluster, and finally carrying out duplicate removal on a point cloud clustering result to obtain a disease instance. Therefore, overlapped efficient partitioning is specifically carried out according to the spatial distribution of the road acquisition data, DBSCAN clustering is carried out in the partitioning mode, the clustering performance can be guaranteed when the data size is large, the integrity of the disease instance at the partitioning boundary is guaranteed, the repeated calculation area is greatly reduced, and the processing speed of the algorithm is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of underground disease treatment technology, and in particular to a method for processing underground disease segmentation results based on three-dimensional ground-penetrating radar. Background Technology

[0002] With the continuous development of urban infrastructure and the increasing complexity of underground pipelines, 3D ground-penetrating radar (GPR) is widely used for identifying underground defects in municipal roads due to its advantages such as non-destructive operation, strong imaging capabilities, and wide adaptability. Traditional GPR systems mainly focus on the acquisition and analysis of two-dimensional data, which is insufficient to meet the needs for three-dimensional reconstruction of spatial structures in complex scenarios. Therefore, 3D GPR systems have gradually become a hot topic in research and engineering applications.

[0003] Existing methods for automatic identification of underground defects based on 3D ground-penetrating radar data and deep learning algorithms generally fall into two categories: 2D / 3D detection and 2D / 3D segmentation. The 2D segmentation method requires merging segmentation masks from multiple radar signal channels to form a 3D segmentation mask. Subsequent methods, including 3D segmentation, require converting the 3D segmentation mask into defect instances; this process is typically called post-processing of the segmentation results. Existing methods have the following problems: one is directly merging all segmentation masks, which leads to a significant performance degradation and a poor user experience when dealing with large datasets; the other is partitioning all masks, which can result in a complete defect instance at the partition boundaries being split into defect instances belonging to multiple partitions. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for processing the segmentation results of underground diseases based on three-dimensional ground penetrating radar.

[0005] The objective of this invention is achieved through the following technical solution: a method for processing the segmentation results of underground defects based on three-dimensional ground-penetrating radar, comprising the following steps:

[0006] S1: Perform overlapping clustering on the road segments corresponding to the data used for segmentation prediction;

[0007] S2: Cluster all the spatially sparse point clouds corresponding to the 3D segmentation masks in the road segments of each cluster;

[0008] S3: Remove duplicates from the point cloud clustering results to obtain disease instances.

[0009] .

[0010] Preferably, step S1 further includes the following step:

[0011] S11: Acquire 3D ground-penetrating radar CSCAN data for 3D segmentation, and calculate the CSCAN center point position based on the latitude and longitude of all measurement points of each CSCAN.

[0012] ;

[0013] in, The center point is located at the center point. Location of each A-SCAN measurement point;

[0014] S12: Perform balanced clustering on all the obtained centroids using a limited number of cluster elements.

[0015] ;

[0016] in, For the first A cluster, To limit the number of cluster elements;

[0017] S13: Merge the clusters.

[0018] ;

[0019] in, For the reason The cluster group is formed by merging the clusters on its left and right, and its three adjacent clusters.

[0020] Preferably, step S2 further includes the following step:

[0021] S21: According to Find the corresponding 3D segmentation mask and convert the non-zero voxels of the 3D mask from mask voxel coordinates to physical space coordinates.

[0022] ;

[0023] in, yes Non-zero voxels of the 3D segmentation mask in all CSCAN, It is the converted physical space point cloud;

[0024] S22: Use DBSCAN to cluster the physical spatial point cloud and obtain the instantiation of the disease spatial point cloud in the cluster group.

[0025] ;

[0026] in, This refers to all disease segmentation instances in this cluster group.

[0027] The present invention has the following advantages: The present invention performs overlapping clustering on the road segments corresponding to the data used for segmentation prediction, then clusters all spatially sparse point clouds corresponding to the 3D segmentation masks in each cluster, and finally deduplicates the point cloud clustering results to obtain disease instances. Thus, it performs overlapping and efficient partitioning based on the specific spatial distribution of road data, and performs DBSCAN clustering on the partitions. Even with a large amount of data, it can ensure clustering performance, thereby ensuring the integrity of disease instances at the partition boundaries and greatly reducing the area of ​​repeated calculation, thus ensuring the processing speed of the algorithm. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the structure of a method for processing the segmentation results of underground defects using 3D ground-penetrating radar. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0030] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0031] It should be noted that, unless otherwise specified, the embodiments and features described in this invention can be combined with each other.

[0032] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0033] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use, or the orientation or positional relationship commonly understood by those skilled in the art. They are only used for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention. In addition, the terms "first," "second," etc., are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0034] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0035] In this embodiment, as Figure 1 As shown, a method for processing the segmentation results of underground defects based on three-dimensional ground-penetrating radar includes the following steps:

[0036] S1: Perform overlapping clustering on the road segments corresponding to the data used for segmentation prediction;

[0037] S2: Cluster all the spatially sparse point clouds corresponding to the 3D segmentation masks in the road segments of each cluster;

[0038] S3: Remove duplicates from the point cloud clustering results to obtain disease instances.

[0039] .

[0040] By performing overlapping clustering on the road segments corresponding to the data used for segmentation prediction, and then clustering the spatially sparse point clouds corresponding to all 3D segmentation masks in each cluster, and finally deduplicating the point cloud clustering results, disease instances are obtained. Thus, based on the specific spatial distribution of the road data, overlapping and efficient partitioning is performed, and DBSCAN clustering is performed on the partitions. Even with a large amount of data, the clustering performance can be guaranteed. This ensures the integrity of disease instances at the partition boundaries and greatly reduces the area of ​​repeated calculation, thus guaranteeing the processing speed of the algorithm.

[0041] Furthermore, step S1 also includes the following steps:

[0042] S11: Acquire 3D ground-penetrating radar CSCAN data for 3D segmentation, and calculate the CSCAN center point position based on the latitude and longitude of all measurement points of each CSCAN.

[0043] ;

[0044] in, The center point is located at the center point. Location of each A-SCAN measurement point;

[0045] S12: Perform balanced clustering on all the obtained centroids using a limited number of cluster elements.

[0046] ;

[0047] in, For the first A cluster, To limit the number of cluster elements;

[0048] S13: Merge the clusters.

[0049] ;

[0050] in, For the reason The cluster group is formed by merging the clusters on the left and right and their adjacent clusters. Specifically, since the data collection is carried out along the road, each cluster obtained in step S12 will only be adjacent to two surrounding clusters. At this time, in step S13, these three clusters are merged into a single cluster group.

[0051] Furthermore, step S2 also includes the following steps:

[0052] S21: According to Find the corresponding 3D segmentation mask and convert the non-zero voxels of the 3D mask from mask voxel coordinates to physical space coordinates.

[0053] ;

[0054] in, yes Non-zero voxels of the 3D segmentation mask in all CSCAN, It is the converted physical space point cloud;

[0055] S22: Use DBSCAN to cluster the physical spatial point cloud and obtain the instantiation of the disease spatial point cloud in the cluster group.

[0056] ;

[0057] in, This refers to all disease segmentation instances in this cluster group.

[0058] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for processing the segmentation results of underground defects based on three-dimensional ground-penetrating radar, characterized in that: Includes the following steps: S1: Perform overlapping clustering on the road segments corresponding to the data used for segmentation prediction; S2: Cluster all the spatially sparse point clouds corresponding to the 3D segmentation masks in the road segments of each cluster; S3: Remove duplicates from the point cloud clustering results to obtain disease instances. 。 2. The method for processing the segmentation results of underground defects based on three-dimensional ground-penetrating radar according to claim 1, characterized in that: Step S1 further includes the following steps: S11: Acquire 3D ground-penetrating radar CSCAN data for 3D segmentation, and calculate the CSCAN center point position based on the latitude and longitude of all measurement points of each CSCAN. ； in, The center point is located at the center point. Location of each A-SCAN measurement point; S12: Perform balanced clustering on all the obtained centroids using a limited number of cluster elements. ； in, For the first A cluster, To limit the number of cluster elements; S13: Merge the clusters. ； in, For the reason The cluster group is formed by merging the clusters on its left and right, and its three adjacent clusters.

3. The method for processing the results of underground disease segmentation based on three-dimensional ground-penetrating radar according to claim 2, characterized in that: Step S2 further includes the following steps: S21: According to Find the corresponding 3D segmentation mask and convert the non-zero voxels of the 3D mask from mask voxel coordinates to physical space coordinates. ； in, yes Non-zero voxels of the 3D segmentation mask in all CSCAN, It is the converted physical space point cloud; S22: Use DBSCAN to cluster the physical spatial point cloud and obtain the instantiation of the disease spatial point cloud in the cluster group. ； in, This refers to all disease segmentation instances in this cluster group.

Images