Underground mine 4D millimeter wave radar ghosting noise automatic marking method and system

Abstract

The invention discloses an underground mine 4D millimeter wave radar ghosting noise automatic marking method and system. The method comprises the steps that first laser radar point cloud data and first 4D millimeter wave radar point cloud data are aligned at the same time reference through a timestamp nearest neighbor matching method, and a time synchronization data pair is obtained; according to the method, the function of accurately identifying and marking the ghosting noise of the 4D millimeter wave radar by combining the high-precision characteristic of the laser radar is realized, the marking efficiency is greatly improved, the marking cost is reduced, the function of detecting the ghosting noise in a mode of combining a distance threshold method and a clustering analysis method is realized, and the accuracy of the ghosting noise of the 4D millimeter wave radar is improved. The detection speed is guaranteed, the marking precision is improved, various complex conditions in the underground mine environment can be handled without depending on a specific signal model and an environment hypothesis, the safety and stability of a mine automation system are improved, remarkable economic and social benefits are achieved, and the method is suitable for being widely popularized and used.

Description

Technical Field

[0001] The present invention relates to the technical field of radar ghost noise marking, and in particular to a method and system for automatically marking ghost noise of a 4D millimeter-wave radar in an underground mine. Background Art

[0002] Underground mining environments feature cramped spaces, irregular terrain, extremely low light conditions, high humidity, and high dust levels, placing extremely high demands on environmental perception technology. In the field of mine automation, 4D millimeter-wave radar has become a key perception tool due to its strong penetration, anti-interference capabilities, and all-weather operation. Compared to traditional 3D radar, 4D millimeter-wave radar not only provides three-dimensional spatial information but also Doppler velocity data, enabling more robust target detection and environmental mapping.

[0003] Currently, the application of 4D millimeter-wave radar in underground mines faces the problem of ghost noise. Ghost noise is the generation of false point cloud data caused by multipath reflections, antenna interference, or environmental complexity, which is particularly serious in underground mining environments. These false point clouds not only reduce data quality but also easily interfere with target detection and autonomous navigation, posing a significant safety risk. Existing ghost noise treatment methods mostly focus on the radar signal processing stage and rely on specific signal models and environmental assumptions, making them difficult to adapt to the complexity of mining environments and the sparsity of 4D millimeter-wave radar point clouds. Manual labeling is also time-consuming, labor-intensive, and costly. Therefore, it is necessary to design a method and system for automatically labeling ghost noise for 4D millimeter-wave radar in underground mines. Summary of the Invention

[0004] The purpose of the present invention is to overcome the shortcomings of the existing technology and to better and effectively solve the problems that the existing ghost noise processing methods are mostly concentrated in the radar signal processing stage and rely on specific signal models and environmental assumptions, which makes it difficult to adapt to the complexity of the mining environment and the sparsity of 4D millimeter-wave radar point clouds. At the same time, the manual marking method is time-consuming, labor-intensive and costly. A method and system for automatic marking of 4D millimeter-wave radar ghost noise in underground mines is provided, which realizes the function of accurately identifying and marking 4D millimeter-wave radar ghost noise by combining the high-precision characteristics of laser radar, which not only greatly improves the marking efficiency, but also reduces the marking cost. It also realizes the function of detecting ghost noise by combining the distance threshold method with the cluster analysis method, which not only ensures the detection speed, but also improves the marking accuracy. It can cope with various complex situations in the underground mining environment without relying on specific signal models and environmental assumptions, which not only improves the safety and stability of the mine automation system, but also has significant economic and social benefits.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] A method for automatically marking ghost noise of 4D millimeter-wave radar in underground mines, comprising the following steps:

[0007] Step A, aligning the first lidar point cloud data and the first 4D millimeter-wave radar point cloud data on the same time basis using a timestamp nearest neighbor matching method to obtain a time-synchronized data pair;

[0008] Step B: spatially aligning the first laser radar point cloud data and the first 4D millimeter wave radar point cloud data to obtain second laser radar point cloud data and second 4D millimeter wave radar point cloud data respectively;

[0009] Step C, based on the time-synchronized data pair, using a distance threshold method to filter ghost noise from the second lidar point cloud data and the second 4D millimeter-wave radar point cloud data to obtain a first ghost noise point set;

[0010] In step D, the DBSCAN clustering method is used to verify and mark the first set of ghost noise points, thereby obtaining a second set of ghost noise points, thereby completing the automatic labeling of underground mine 4D millimeter wave radar ghost noise.

[0011] In the aforementioned method for automatically labeling ghost noise in underground mines using 4D millimeter-wave radar, step A uses the timestamp nearest neighbor matching method to align the lidar point cloud data and the 4D millimeter-wave radar point cloud data to the same time base and obtain a time-synchronized data pair. The specific steps are as follows:

[0012] Step A1: Assume that the timestamp sequence of the lidar point cloud and the timestamp sequence of the 4D millimeter wave radar are T L and T R , then find the 4D millimeter-wave radar timestamp closest to the lidar timestamp, as shown in formula (1);

[0013] j=arg min|T L [i]-T R [j]| (1)

[0014] Among them, T L [i] is the timestamp of the i-th frame of the lidar point cloud, T R [j] is the timestamp of the j-th frame 4D millimeter-wave radar point cloud;

[0015] Step A2: Pair the first laser radar point cloud data and the first 4D millimeter wave radar point cloud data with the closest timestamps to form a time-synchronized data pair (T L [i], T R [j]).

[0016] The aforementioned method for automatic marking of ghost noise of 4D millimeter-wave radar in underground mines, step B, spatially aligning the first laser radar point cloud data and the first 4D millimeter-wave radar point cloud data, respectively obtaining the second laser radar point cloud data and the second 4D millimeter-wave radar point cloud data, wherein the spatial alignment is specifically to uniformly convert the first laser radar point cloud coordinates in the first laser radar point cloud data and the first 4D millimeter-wave radar point cloud coordinates in the first 4D millimeter-wave radar point cloud data into the reference coordinate system of the acquisition device and respectively obtain the second laser radar point cloud data and the second 4D millimeter-wave radar point cloud data, and the unified conversion of the point cloud coordinates is specifically to use the rotation matrix and the translation vector for linear transformation, assuming that the coordinate system of the laser radar is L, the coordinate system of the 4D millimeter-wave radar is R and the reference coordinate system is B, then the coordinates of the first laser radar point cloud in the laser radar coordinate system are (x L ,,y L ,,z L ), the first 4D millimeter wave radar point cloud coordinate in the 4D millimeter wave radar coordinate system is (x R ,y R , z R ) and the point cloud coordinates in the reference coordinate system are (x B ,y B ,,z B ), the first laser radar point cloud coordinate conversion and the first 4D millimeter wave radar point cloud coordinate conversion processes are shown in formula (2) and formula (3) respectively;

[0017]

[0018] Among them, R L→B and R R→B are the laser radar rotation matrix and the 4D millimeter wave radar rotation matrix respectively, [t Lx t Ly t Lz ] T and [t Rx t Ry t Rz ] T They are the lidar translation vector and 4D millimeter-wave radar translation vector respectively.

[0019] The aforementioned method for automatically marking ghost noise of 4D millimeter-wave radar in underground mines, step C, based on the time-synchronized data pair, uses the distance threshold method to screen the ghost noise of the second laser radar point cloud data and the second 4D millimeter-wave radar point cloud data to obtain a first ghost noise point set. Specifically, the distance threshold method is used to calculate the nearest neighbor distance between the second 4D millimeter-wave radar point in each second 4D millimeter-wave radar point cloud data and the second laser radar point cloud in the second laser radar point cloud data. The specific steps are as follows:

[0020] Step C1, set the second 4D millimeter wave radar point cloud as the set And the second lidar point cloud is a collection Calculate every second 4D millimeter wave radar point With the second lidar point cloud N L The nearest neighbor distance in As shown in formula (4),

[0021]

[0022] Among them, ‖·‖ is the Euclidean distance, is the second lidar point;

[0023] Step C2: If the second laser radar point cloud N L There is a second lidar point in satisfy Where δ is the point distance threshold, then the second laser radar point Mark it as the first ghost noise point, then count all the first ghost noise points and form a first ghost noise point set.

[0024] In the aforementioned method for automatically labeling ghost noise of 4D millimeter-wave radar in underground mines, step D uses the DBSCAN clustering method to verify and label the first set of ghost noise points, thereby obtaining a second set of ghost noise points, and then completing the automatic labeling operation of the 4D millimeter-wave radar ghost noise in underground mines. Specifically, the DBSCAN algorithm is used to cluster the first ghost noise points and optimize the clustering effect by adjusting the neighborhood radius and the minimum number of cluster points to obtain the second ghost noise points, and then obtain the second set of ghost noise points. The specific steps are as follows:

[0025] Step D1, using the DBSCAN algorithm to automatically identify sparsely distributed noise points and distinguish point cloud clusters of real objects based on the point cloud density characteristics of the first ghost noise point set, assuming that the noise lidar point cloud and the noise 4D millimeter wave radar point cloud in the first ghost noise point set after processing by the DBSCAN algorithm are respectively divided into noise lidar clusters Clustering of 4D millimeter-wave radar with noise in and Noisy lidar clustering Clustering of 4D millimeter-wave radar with noise Then calculate each noise 4D millimeter wave radar cluster subset Clustering subsets with the nearest noisy lidar distance As shown in formula (5),

[0026]

[0027] Step D2: If the noise 4D millimeter wave radar cluster subset satisfy Where λ is the set cluster distance threshold, then the noise 4D millimeter wave radar cluster subset All points in are second ghost noise points, and then all second ghost noise points are counted to form a second ghost noise point set.

[0028] A system for automatically marking ghost noise from 4D millimeter-wave radars in underground mines includes a time alignment module, a spatial alignment module, a noise screening module, and a noise marking module. The time alignment module is used to align first laser radar point cloud data and first 4D millimeter-wave radar point cloud data to the same time base using a timestamp nearest neighbor matching method and obtain a time-synchronized data pair; the spatial alignment module is used to spatially align the first laser radar point cloud data and the first 4D millimeter-wave radar point cloud data and obtain second laser radar point cloud data and second 4D millimeter-wave radar point cloud data, respectively; the noise screening module is used to screen ghost noise from the second laser radar point cloud data and the second 4D millimeter-wave radar point cloud data based on the time-synchronized data pair and using a distance threshold method to obtain a first ghost noise point set; the noise marking module is used to verify and mark the first ghost noise point set using a DBSCAN clustering method to obtain a second ghost noise point set, thereby completing the automatic marking operation of 4D millimeter-wave radar ghost noise in underground mines.

[0029] The beneficial effects of the present invention are as follows: the present invention provides an underground mine 4D millimeter-wave radar ghost noise automatic marking method and system, firstly adopts the timestamp nearest neighbor matching method to align the first lidar point cloud data and the first 4D millimeter-wave radar point cloud data to the same time base and obtains a time synchronization data pair, then spatially aligns the first lidar point cloud data and the first 4D millimeter-wave radar point cloud data to obtain the second lidar point cloud data and the second 4D millimeter-wave radar point cloud data respectively, then based on the time synchronization data pair and using the distance threshold method, the ghost noise of the second lidar point cloud data and the second 4D millimeter-wave radar point cloud data is screened to obtain a first ghost noise point set, then adopts the DBSCAN clustering method to verify and mark the first ghost noise point set to obtain the second ghost noise point set, thereby completing the automatic marking operation of the underground mine 4D millimeter-wave radar ghost noise; effectively realizing that the underground mine 4D millimeter-wave radar ghost noise automatic marking method and system have the characteristics of combining the high-precision characteristics of the lidar to mark the 4D millimeter-wave The function of accurately identifying and marking radar ghost noise not only greatly improves the marking efficiency, but also reduces the marking cost. Compared with manual marking, which may take hours or even days, the present invention can complete the marking task of data of the same scale in just a few minutes. It also realizes the function of detecting ghost noise by combining the distance threshold method with the cluster analysis method, which not only ensures the detection speed but also improves the marking accuracy. This dual mechanism of preliminary screening and secondary verification makes the marking accuracy far exceed the performance of traditional single algorithms. The present invention can cope with various complex situations in underground mining environments without relying on specific signal models and environmental assumptions, and indirectly enhances its application effect in underground mining environments by improving the quality of 4D millimeter-wave radar data. This provides a more reliable perception basis for tasks such as autonomous navigation, collision avoidance and environmental perception, which not only improves the safety and stability of mine automation systems, but also has significant economic and social benefits. It can also be promoted and applied to 4D millimeter-wave radar data processing in other complex environments, and has broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is an overall flow chart of an automatic marking method for underground mine 4D millimeter wave radar ghost noise according to the present invention;

[0031] Figure 2 This is a schematic diagram of the working principle of the underground mine 4D millimeter wave radar ghost noise automatic marking system of the present invention;

[0032] Figure 3This is a comparison of the labeling results of the present invention in a straight tunnel scenario, where the green points are LiDAR references, the blue points are real radar points, and the red points are ghost radar points; (a) is the result of manual labeling, (b) is the result of labeling using the proposed method, (c) is the result of labeling using the distance threshold method, and (d) is the result of labeling using the DBSCAN method;

[0033] Figure 4 This is a comparison of the labeling results of the present invention in the side tunnel scenario, where the green points are LiDAR references, the blue points are real radar points, and the red points are ghost radar points; (a) is the result of manual labeling, (b) is the result of labeling using the proposed method, (c) is the result of labeling using the distance threshold method, and (d) is the result of labeling using the DBSCAN method;

[0034] Figure 5 This is a comparison chart of the marking results of the present invention in the cross tunnel scene, where the green points are Lidar references, the blue points are real radar points, and the red points are ghost radar points; (a) the result of manual marking, (b) the marking result using the method proposed in this invention, (c) the marking result using the distance threshold method, and (d) the marking result using the DBSCAN method. DETAILED DESCRIPTION

[0035] The present invention will be further described below with reference to the accompanying drawings.

[0036] like Figure 1 As shown, the present invention provides an automatic marking method for underground mine 4D millimeter wave radar ghost noise, comprising the following steps:

[0037] Step A: Use the timestamp nearest neighbor matching method to align the lidar point cloud data and the 4D millimeter wave radar point cloud data to the same time base and obtain a time-synchronized data pair. The specific steps are as follows:

[0038] Step A1: Assume that the timestamp sequence of the lidar point cloud and the timestamp sequence of the 4D millimeter wave radar are T L and T R , then find the 4D millimeter-wave radar timestamp closest to the lidar timestamp, as shown in formula (1);

[0039] j=arg min|T L [i]-T R [j]| (1)

[0040] Among them, T L [i] is the timestamp of the i-th frame of the lidar point cloud, T R [j] is the timestamp of the j-th frame 4D millimeter-wave radar point cloud;

[0041] Step A2: Pair the first laser radar point cloud data and the first 4D millimeter wave radar point cloud data with the closest timestamps to form a time-synchronized data pair (T L [i], T R [j]).

[0042] In step A, for each frame of the lidar point cloud, the corresponding synchronized 4D millimeter-wave radar point cloud must meet the minimum absolute difference between the timestamps. The present invention can effectively solve the time alignment problem caused by the different sampling frequencies of the two sensors through time synchronization, providing consistency guarantee for subsequent automatic labeling work.

[0043] Step B, spatially aligning the first laser radar point cloud data and the first 4D millimeter wave radar point cloud data to obtain second laser radar point cloud data and second 4D millimeter wave radar point cloud data, respectively. The spatial alignment is specifically to uniformly convert the first laser radar point cloud coordinates in the first laser radar point cloud data and the first 4D millimeter wave radar point cloud coordinates in the first 4D millimeter wave radar point cloud data into the reference coordinate system of the acquisition device and obtain the second laser radar point cloud data and the second 4D millimeter wave radar point cloud data, respectively. The unified conversion of the point cloud coordinates is specifically to use the rotation matrix and the translation vector for linear transformation. Assuming that the coordinate system of the laser radar is L, the coordinate system of the 4D millimeter wave radar is R, and the reference coordinate system is B, then the coordinates of the first laser radar point cloud in the laser radar coordinate system are (x L ,,y L ,,z L ), the first 4D millimeter wave radar point cloud coordinate in the 4D millimeter wave radar coordinate system is (x R ,y R , z R ) and the point cloud coordinates in the reference coordinate system are (x B ,y B ,,z B ), the first laser radar point cloud coordinate conversion and the first 4D millimeter wave radar point cloud coordinate conversion processes are shown in formula (2) and formula (3) respectively;

[0044]

[0045] Among them, R L→B and R R→B are the laser radar rotation matrix and the 4D millimeter wave radar rotation matrix respectively, [t Lx t Ly t Lz ] T and [t Rx t Ry t Rz ] T They are the lidar translation vector and 4D millimeter-wave radar translation vector respectively.

[0046] In step B, the present invention uses spatial alignment to unify the lidar point cloud data and 4D millimeter-wave radar point cloud data into the same coordinate system, ensuring the consistency of the two sensors in geometric space, thereby avoiding error accumulation caused by coordinate system deviation, and providing high-quality point cloud input for subsequent labeling tasks.

[0047] Step C, based on the time-synchronized data pair, uses the distance threshold method to filter the ghost noise of the second lidar point cloud data and the second 4D millimeter-wave radar point cloud data to obtain a first ghost noise point set. Specifically, the distance threshold method is used to calculate the nearest neighbor distance between each second 4D millimeter-wave radar point in the second 4D millimeter-wave radar point cloud data and the second lidar point cloud in the second lidar point cloud data. The specific steps are as follows:

[0048] Step C1, set the second 4D millimeter wave radar point cloud as the set And the second lidar point cloud is a collection Calculate every second 4D millimeter wave radar point With the second lidar point cloud N L The nearest neighbor distance in As shown in formula (4),

[0049]

[0050] Among them, ‖·‖ is the Euclidean distance, is the second lidar point;

[0051] Step C2: If the second laser radar point cloud N L There is a second lidar point in satisfy Where δ is the point distance threshold, then the second laser radar point Mark it as the first ghost noise point, then count all the first ghost noise points and form a first ghost noise point set.

[0052] In step C, the present invention uses KD-Tree to accelerate the calculation in the nearest neighbor search and constructs the laser radar point cloud N L The spatial index structure can quickly find each 4D millimeter wave radar point cloud point The nearest neighbor point of , significantly reduces the computational complexity.

[0053] Step D: Use the DBSCAN clustering method to verify and mark the first ghost noise point set, thereby obtaining a second ghost noise point set, and then completing the automatic labeling of underground mine 4D millimeter wave radar ghost noise. Specifically, the DBSCAN algorithm is used to cluster the first ghost noise points and optimize the clustering effect by adjusting the neighborhood radius and the minimum number of cluster points to obtain the second ghost noise points, and then obtain the second ghost noise point set. The specific steps are as follows:

[0054] Step D1, using the DBSCAN algorithm to automatically identify sparsely distributed noise points and distinguish point cloud clusters of real objects based on the point cloud density characteristics of the first ghost noise point set, assuming that the noise lidar point cloud and the noise 4D millimeter wave radar point cloud in the first ghost noise point set after processing by the DBSCAN algorithm are respectively divided into noise lidar clusters Clustering of 4D millimeter-wave radar with noise in and Noisy lidar clustering Clustering of 4D millimeter-wave radar with noise Then calculate each noise 4D millimeter wave radar cluster subset Clustering subsets with the nearest noisy lidar distance As shown in formula (5),

[0055]

[0056] Step D2: If the noise 4D millimeter wave radar cluster subset satisfy Where λ is the set cluster distance threshold, then the noise 4D millimeter wave radar cluster subset All points in are second ghost noise points, and then all second ghost noise points are counted to form a second ghost noise point set.

[0057] like Figure 2-5As shown, a 4D millimeter-wave radar ghost noise automatic labeling system for underground mines includes a time alignment module, a spatial alignment module, a noise screening module and a noise labeling module. The time alignment module is used to align the first laser radar point cloud data and the first 4D millimeter-wave radar point cloud data to the same time base by using the timestamp nearest neighbor matching method and obtain a time-synchronized data pair; the spatial alignment module is used to spatially align the first laser radar point cloud data and the first 4D millimeter-wave radar point cloud data and obtain the second laser radar point cloud data and the second 4D millimeter-wave radar point cloud data respectively; the noise screening module is used to screen the ghost noise of the second laser radar point cloud data and the second 4D millimeter-wave radar point cloud data based on the time-synchronized data pair and using the distance threshold method to obtain a first ghost noise point set; the noise labeling module is used to verify and label the first ghost noise point set by using the DBSCAN clustering method to obtain a second ghost noise point set, thereby completing the automatic labeling operation of the 4D millimeter-wave radar ghost noise in the underground mine.

[0058] In summary, the present invention provides an underground mine 4D millimeter-wave radar ghost noise automatic labeling method and system, firstly uses the timestamp nearest neighbor matching method to align the first lidar point cloud data and the first 4D millimeter-wave radar point cloud data to the same time base and obtains a time synchronization data pair, then spatially aligns the first lidar point cloud data and the first 4D millimeter-wave radar point cloud data to obtain the second lidar point cloud data and the second 4D millimeter-wave radar point cloud data respectively, then based on the time synchronization data pair and using the distance threshold method, the ghost noise of the second lidar point cloud data and the second 4D millimeter-wave radar point cloud data is screened to obtain a first ghost noise point set, then the DBSCAN clustering method is used to verify and label the first ghost noise point set to obtain a second ghost noise point set, thereby completing the automatic labeling operation of the underground mine 4D millimeter-wave radar ghost noise; effectively realizing that the underground mine 4D millimeter-wave radar ghost noise automatic labeling method and system have the characteristics of combining the high-precision characteristics of the lidar to label the 4D millimeter-wave radar The function of accurately identifying and marking ghost noise not only greatly improves the marking efficiency, but also reduces the marking cost. Compared with manual marking, which may take hours or even days, the present invention can complete the marking task of data of the same scale in just a few minutes. It also realizes the function of detecting ghost noise by combining the distance threshold method with the cluster analysis method, which not only ensures the detection speed but also improves the marking accuracy. This dual mechanism of preliminary screening and secondary verification makes the marking accuracy far exceed the performance of traditional single algorithms. The present invention can cope with various complex situations in underground mining environments without relying on specific signal models and environmental assumptions, and indirectly enhances its application effect in underground mining environments by improving the quality of 4D millimeter-wave radar data. This provides a more reliable perception basis for tasks such as autonomous navigation, collision avoidance and environmental perception, which not only improves the safety and stability of mine automation systems, but also has significant economic and social benefits. It can also be promoted and applied to 4D millimeter-wave radar data processing in other complex environments, and has broad application prospects.

[0059] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the foregoing embodiments. The foregoing embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for automatically marking ghost noise from 4D millimeter-wave radar in underground mines, characterized by: The following steps are included: Step A, aligning the first lidar point cloud data and the first 4D millimeter-wave radar point cloud data on the same time basis using a timestamp nearest neighbor matching method to obtain a time-synchronized data pair; Step B: spatially aligning the first laser radar point cloud data and the first 4D millimeter wave radar point cloud data to obtain second laser radar point cloud data and second 4D millimeter wave radar point cloud data respectively; Step C, based on the time-synchronized data pair, using a distance threshold method to filter ghost noise from the second lidar point cloud data and the second 4D millimeter-wave radar point cloud data to obtain a first ghost noise point set; In step D, the DBSCAN clustering method is used to verify and mark the first set of ghost noise points, thereby obtaining a second set of ghost noise points, thereby completing the automatic labeling of underground mine 4D millimeter wave radar ghost noise.

2. The method for automatically marking ghost noise using 4D millimeter-wave radar in underground mines according to claim 1, characterized in that: Step A: Use the timestamp nearest neighbor matching method to align the lidar point cloud data and the 4D millimeter wave radar point cloud data to the same time base and obtain a time-synchronized data pair. The specific steps are as follows: Step A1: Assume that the timestamp sequence of the lidar point cloud and the timestamp sequence of the 4D millimeter wave radar are T L and T R , then find the 4D millimeter-wave radar timestamp closest to the lidar timestamp, as shown in formula (1); j=arg min|T L [i]-T R [j]| (1) Among them, T L [i] is the timestamp of the i-th frame of the lidar point cloud, T R [j] is the timestamp of the j-th frame 4D millimeter-wave radar point cloud; Step A2: Pair the first laser radar point cloud data and the first 4D millimeter wave radar point cloud data with the closest timestamps to form a time-synchronized data pair (T L [i], T R [j]).

3. The method for automatically marking underground mine 4D millimeter-wave radar ghost noise according to claim 2, characterized in that: Step B, spatially aligning the first laser radar point cloud data and the first 4D millimeter wave radar point cloud data to obtain second laser radar point cloud data and second 4D millimeter wave radar point cloud data, respectively. The spatial alignment is specifically to uniformly convert the first laser radar point cloud coordinates in the first laser radar point cloud data and the first 4D millimeter wave radar point cloud coordinates in the first 4D millimeter wave radar point cloud data into the reference coordinate system of the acquisition device and obtain the second laser radar point cloud data and the second 4D millimeter wave radar point cloud data, respectively. The unified conversion of the point cloud coordinates is specifically to use the rotation matrix and the translation vector for linear transformation. Assuming that the coordinate system of the laser radar is L, the coordinate system of the 4D millimeter wave radar is R, and the reference coordinate system is B, then the coordinates of the first laser radar point cloud in the laser radar coordinate system are (x L ,,y L ,,z L ), the first 4D millimeter wave radar point cloud coordinate in the 4D millimeter wave radar coordinate system is (x R ,y R , z R ) and the point cloud coordinates in the reference coordinate system are (x B ,y B ,,z B ), the first laser radar point cloud coordinate conversion and the first 4D millimeter wave radar point cloud coordinate conversion processes are shown in formula (2) and formula (3) respectively; Among them, R L→B and R R→B are the laser radar rotation matrix and the 4D millimeter wave radar rotation matrix respectively, [t Lx t Ly t Lz ] T and [t Rx t Ry t Rz ] T They are the lidar translation vector and 4D millimeter-wave radar translation vector respectively.

4. The method for automatically marking ghost noise using 4D millimeter-wave radar in underground mines according to claim 3, characterized in that: Step C, based on the time-synchronized data pair, uses the distance threshold method to filter the ghost noise of the second lidar point cloud data and the second 4D millimeter-wave radar point cloud data to obtain a first ghost noise point set. Specifically, the distance threshold method is used to calculate the nearest neighbor distance between each second 4D millimeter-wave radar point in the second 4D millimeter-wave radar point cloud data and the second lidar point cloud in the second lidar point cloud data. The specific steps are as follows: Step C1, set the second 4D millimeter wave radar point cloud as the set And the second lidar point cloud is a collection Calculate every second 4D millimeter wave radar point With the second lidar point cloud N L The nearest neighbor distance in As shown in formula (4), Among them, ‖·‖ is the Euclidean distance, is the second lidar point; Step C2: If the second laser radar point cloud N L There is a second lidar point in satisfy Where δ is the point distance threshold, then the second laser radar point Mark it as the first ghost noise point, then count all the first ghost noise points and form a first ghost noise point set.

5. The method for automatically marking ghost noise using 4D millimeter-wave radar in underground mines according to claim 4, characterized in that: Step D: Use the DBSCAN clustering method to verify and mark the first ghost noise point set, thereby obtaining a second ghost noise point set, and then completing the automatic labeling of underground mine 4D millimeter wave radar ghost noise. Specifically, the DBSCAN algorithm is used to cluster the first ghost noise points and optimize the clustering effect by adjusting the neighborhood radius and the minimum number of cluster points to obtain the second ghost noise points, and then obtain the second ghost noise point set. The specific steps are as follows: Step D1, using the DBSCAN algorithm to automatically identify sparsely distributed noise points and distinguish point cloud clusters of real objects based on the point cloud density characteristics of the first ghost noise point set, assuming that the noise lidar point cloud and the noise 4D millimeter wave radar point cloud in the first ghost noise point set after processing by the DBSCAN algorithm are respectively divided into noise lidar clusters Clustering of 4D millimeter-wave radar with noise in and Noisy lidar clustering Clustering of 4D millimeter-wave radar with noise Then calculate each noise 4D millimeter wave radar cluster subset Clustering subsets with the nearest noisy lidar distance As shown in formula (5), Step D2: If the noise 4D millimeter wave radar cluster subset satisfy Where λ is the set cluster distance threshold, then the noise 4D millimeter wave radar cluster subset All points in are second ghost noise points, and then all second ghost noise points are counted to form a second ghost noise point set.

6. A 4D millimeter-wave radar ghost noise automatic marking system for underground mines, wherein the specific automatic marking process of the ghost noise automatic marking system is based on the ghost noise automatic marking method according to any one of claims 1 to 5, characterized in that: It includes a time alignment module, a space alignment module, a noise screening module and a noise marking module. The time alignment module is used to align the first laser radar point cloud data and the first 4D millimeter wave radar point cloud data to the same time base using a timestamp nearest neighbor matching method and obtain a time synchronized data pair; The spatial alignment module is used to spatially align the first laser radar point cloud data and the first 4D millimeter wave radar point cloud data and obtain the second laser radar point cloud data and the second 4D millimeter wave radar point cloud data respectively; The noise screening module is used to screen the ghost noise of the second lidar point cloud data and the second 4D millimeter wave radar point cloud data based on the time synchronization data and using the distance threshold method to obtain a first ghost noise point set; The noise labeling module is used to verify and label the first ghost noise point set using the DBSCAN clustering method, thereby obtaining a second ghost noise point set, and further completing the automatic labeling operation of underground mine 4D millimeter wave radar ghost noise.

Images