Target identification method and system based on point cloud data

Abstract

The invention provides a target identification method and system based on point cloud data, and the method can collect the point cloud data of a target region, and extract a target point cloud and a non-target point cloud from the point cloud data. And segmenting the target point cloud into sub-target point clouds, and constructing a projection plane based on the sub-target point clouds. And respectively projecting the sub-target point cloud and the non-target point cloud to a projection plane to obtain a sub-target projection point cloud and a non-target projection point cloud. And extracting target data points from the non-target projection point clouds based on the sub-target projection point clouds. And finally identifying the target type of the sub-target point cloud according to the target data point. According to the method, target recognition of the point cloud data is carried out by utilizing the feature attributes of the background point cloud formed by the target, and the accuracy of target recognition is improved.

Description

Technical Field

[0001] This application relates to the technical field of target recognition, and particularly to a target recognition method and system based on point cloud data. Background Art

[0002] A lidar (Laser Radar) is a radar system that detects the position, speed, and other characteristic quantities of a target by emitting laser beams. The lidar can emit laser pulses into the environment and calculate the distance of an object based on the time required for the pulses to return from the object. During this process, the lidar will obtain detailed three-dimensional point cloud data of the surrounding environment in terms of depth and height. The target recognition method is a process of identifying and positioning objects or targets in the environment based on the point cloud data. The target recognition method can be widely applied to fields such as autonomous driving, robot navigation, line obstacle detection, environmental perception, and security monitoring. For example, in a line obstacle detection system, there may be situations where rocks, landslides, and collapses occur on a railway line, forming obstacles that block the line and endangering the safety of train operation. The target recognition method based on lidar three-dimensional point cloud data can identify the obstacles affecting the line and issue warnings in a timely manner, improving the safety of railway operation.

[0003] When identifying obstacles affecting the line, for branch-shaped targets such as grass branches and tree branches that roll onto the railway line, they do not affect the safety of train operation and do not require warnings. If the branch-shaped targets such as grass branches and tree branches are not accurately identified, a large number of false alarms will be generated, affecting the normal operation of the train and reducing the practicality of the system. To accurately identify different types of targets, deep learning methods can be used, but deep learning methods require a large amount of sample data, resulting in a large workload for sample collection of various targets. They also have high requirements for computing power, requiring high hardware support and cost investment, and the design of deep learning models is complex and the recognition results cannot be explained. This will reduce the accuracy of target recognition. Summary of the Invention

[0004] This application provides a target recognition method and system based on point cloud data to solve the problem of low accuracy of target recognition.

[0005] In a first aspect, this application provides a target recognition method based on point cloud data, including:

[0006] Collecting point cloud data of a target area;

[0007] Extracting target point clouds and non-target point clouds from the point cloud data, where the target point clouds include data points in the point cloud data whose distance from the data points in the background point cloud is greater than a first distance, and the non-target point clouds include data points in the point cloud data whose distance from the data points in the background point cloud is less than or equal to the first distance, and the background point cloud is the point cloud data after performing rasterized downsampling processing;

[0008] Divide the target point cloud into sub-target point clouds, where the sub-target point clouds are sets of data points in the target point cloud with distances between data points less than a second distance;

[0009] Construct a projection plane based on the sub-target point cloud, where the projection plane passes through the average coordinate point of the sub-target point cloud, and the normal of the projection plane passes through the reference coordinate point of the point cloud data and the average coordinate point of the sub-target point cloud;

[0010] Project the sub-target point cloud and the non-target point cloud onto the projection plane respectively to obtain a sub-target projected point cloud and a non-target projected point cloud;

[0011] Extract target data points in the non-target projected point cloud based on the sub-target projected point cloud, where the target data points are data points located within a boundary polygon region, and the boundary polygon region is a region constructed based on the boundary of the sub-target projected point cloud;

[0012] Identify the target type of the sub-target point cloud according to the target data points.

[0013] In an alternative embodiment, the step of constructing a projection plane based on the sub-target point cloud includes:

[0014] Calculate the average coordinate point of the sub-target point cloud, where the average coordinate point is the average value of the coordinate values of the data points in the sub-target point cloud;

[0015] Obtain the reference coordinate point of the point cloud data, where the reference coordinate point is the position point where the laser signal is emitted when collecting the point cloud data;

[0016] Construct a normal passing through the average coordinate point and the reference coordinate point according to the average coordinate point and the reference coordinate point;

[0017] Construct a projection plane perpendicular to the normal and passing through the average coordinate point.

[0018] In an alternative embodiment, the step of calculating the average coordinate point of the sub-target point cloud includes:

[0019] Calculate the average value of the horizontal axis coordinate values of the data points in the sub-target point cloud to obtain the horizontal axis coordinate value of the average coordinate point;

[0020] Calculate the average value of the vertical axis coordinate values of the data points in the sub-target point cloud to obtain the vertical axis coordinate value of the average coordinate point;

[0021] Calculate the average value of the vertical axis coordinate values of the data points in the sub-target point cloud to obtain the vertical axis coordinate value of the average coordinate point.

[0022] In an alternative embodiment, the step of extracting target data points from the non-target projected point cloud based on the sub-target projected point cloud includes:

[0023] Identifying the boundary of the sub-target projected point cloud;

[0024] Extracting the data points on the boundary;

[0025] Constructing a boundary polygon region with the data points on the boundary as vertices;

[0026] Extracting target data points located within the boundary polygon region from the non-target projected point cloud.

[0027] In an alternative embodiment, the step of extracting target data points located within the boundary polygon region from the non-target projected point cloud includes:

[0028] Traversing the data points of the non-target projected point cloud;

[0029] If the data point is located within the boundary polygon region, marking the data point as a target data point;

[0030] If the data point is not located within the boundary polygon region, marking the data point as a non-target data point.

[0031] In an alternative embodiment, the step of identifying the target type of the sub-target point cloud based on the target data points includes:

[0032] Obtaining the total number of points of the sub-target projected point cloud;

[0033] Obtaining the number of data points of the target data points;

[0034] Calculating a point ratio, where the point ratio is the ratio of the number of data points to the total number of points;

[0035] If the point ratio is greater than a ratio threshold, marking the target type of the sub-target point cloud as a branched type;

[0036] If the point ratio is less than or equal to the ratio threshold, marking the target type of the sub-target point cloud as a non-branched type.

[0037] In an alternative embodiment, the step of extracting the target point cloud and the non-target point cloud from the point cloud data includes:

[0038] Performing rasterized downsampling processing on the point cloud data to obtain a background point cloud;

[0039] Calculating the distance between the data points in the point cloud data and the data points in the background point cloud;

[0040] If the distance between the data points in the point cloud data and the data points in the background point cloud is greater than a first distance, classify the data points in the point cloud data into the target point cloud;

[0041] If the distance between the data points in the point cloud data and the data points in the background point cloud is less than or equal to the first distance, classify the data points in the point cloud data into the non-target point cloud.

[0042] In an alternative embodiment, the step of dividing the target point cloud into sub-target point clouds includes:

[0043] Calculate the distances between the data points in the target point cloud;

[0044] If the distance between two of the data points is less than a second distance, classify the two data points into the sub-target point cloud.

[0045] In a second aspect, the present application provides a target recognition system based on point cloud data, including:

[0046] A point cloud data acquisition module for acquiring point cloud data of a target area;

[0047] A target recognition module for extracting a target point cloud and a non-target point cloud from the point cloud data, where the target point cloud includes the data points in the point cloud data whose distance from the data points in the background point cloud is greater than a first distance, and the non-target point cloud includes the data points in the point cloud data whose distance from the data points in the background point cloud is less than or equal to the first distance, and the background point cloud is the point cloud data after performing rasterized downsampling processing;

[0048] Divide the target point cloud into sub-target point clouds, where the sub-target point cloud is a set of data points in the target point cloud whose distances between the data points are less than a second distance;

[0049] Construct a projection plane based on the sub-target point cloud, where the projection plane passes through the average coordinate point of the sub-target point cloud, and the normal line of the projection plane passes through the reference coordinate point of the point cloud data and the average coordinate point of the sub-target point cloud;

[0050] Project the sub-target point cloud and the non-target point cloud onto the projection plane respectively to obtain a sub-target projection point cloud and a non-target projection point cloud;

[0051] Extract target data points from the non-target projection point cloud based on the sub-target projection point cloud, where the target data points are the data points located within the boundary polygon area, and the boundary polygon area is an area constructed based on the boundary of the sub-target projection point cloud;

[0052] Identify the target type of the sub-target point cloud according to the target data points.

[0053] In an optional embodiment, the step of the target recognition module extracting target data points from the non-target projected point cloud based on the sub-target projected point cloud is further configured to:

[0054] Identify the boundary of the sub-target projected point cloud;

[0055] Extract the data points on the boundary;

[0056] Construct a boundary polygon region with the data points on the boundary as vertices;

[0057] Extract target data points located within the boundary polygon region from the non-target projected point cloud.

[0058] As can be seen from the above technical solutions, the present application provides a target recognition method and system based on point cloud data. The method can collect point cloud data of a target area, extract target point clouds and non-target point clouds from the point cloud data. Then, the target point cloud is segmented into sub-target point clouds, and a projection plane is constructed based on the sub-target point clouds. Then, the sub-target point clouds and non-target point clouds are respectively projected onto the projection plane to obtain sub-target projected point clouds and non-target projected point clouds. Then, based on the sub-target projected point clouds, target data points are extracted from the non-target projected point clouds. Finally, the target type of the sub-target point cloud is identified according to the target data points. The method utilizes the characteristic attributes of the target to form the background point cloud to perform target recognition of the point cloud data, improving the accuracy of target recognition. Description of the Drawings

[0059] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0060] Figure 1 It is a schematic flowchart of the target recognition method provided by the embodiment of the present application;

[0061] Figure 2 It is a schematic diagram of a picture and point cloud data containing a branched target in a railway line provided by the embodiment of the present application;

[0062] Figure 3 It is a partially enlarged schematic diagram of the branched target provided by the embodiment of the present application;

[0063] Figure 4 It is a schematic flowchart of extracting target point clouds and non-target point clouds provided by the embodiment of the present application;

[0064] Figure 5 A flowchart of the process for dividing a target point cloud into sub-target point clouds provided by an embodiment of the present application;

[0065] Figure 6 A flowchart of the process for constructing a projection plane provided by an embodiment of the present application;

[0066] Figure 7 A flowchart of the process for identifying the target type of sub-target point clouds provided by an embodiment of the present application. Detailed implementation manners

[0067] The embodiments will be described in detail below, and the examples are shown in the accompanying drawings. When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementation manners described in the following embodiments do not represent all implementation manners consistent with the present application. They are only examples of systems and methods consistent with some aspects of the present application detailed in the claims.

[0068] In an on-line obstacle system, on a railway line, there will be situations where falling rocks, landslides, and collapses form obstacles blocking the line, endangering the safety of train operation. The target recognition method based on lidar three-dimensional point cloud data can identify obstacles affecting the line and issue warnings in a timely manner, improving the safety of railway operation.

[0069] When identifying obstacles affecting the line, for branched targets such as grass branches and tree branches that roll onto the railway line, they do not affect the safety of train operation and do not require warnings. If the branched targets such as grass branches and tree branches are not accurately identified, a large number of false alarms will be generated, affecting the normal operation of the train and reducing the practicality of the system.

[0070] In some embodiments, in order to accurately identify different types of targets, deep learning methods can be used. However, deep learning methods require a large amount of sample data. In the on-line obstacle system, in addition to identifying branched targets, it is also necessary to identify various targets such as animals, people, trains, rain, snow, and fog. The workload of sample collection for various targets is large. The computing power requirement is also very high. For example, for a long railway line, the three-dimensional radar monitoring unit is installed near the track, and each radar monitors a distance of 30 - 100 meters, and each monitoring unit operates independently on-site. And the design of the deep learning model is complex, and the recognition results cannot be explained. Thus, the accuracy of target recognition will be reduced.

[0071] In some embodiments, it is also possible to identify based on the shape of the branched target. The branched target is a tree-like structure, that is, there are many branches on the main trunk. However, affected by factors such as the resolution ability and detection distance of the lidar, the accuracy of target recognition only from the shape is relatively low.

[0072] To solve the problem of low accuracy in target recognition, an embodiment of the present application provides a target recognition system based on point cloud data. As Figure 1 shown, the system includes a point cloud data acquisition module and a target recognition module. The point cloud data acquisition module is used to acquire the point cloud data of the target area. For example, in an online obstacle system, the point cloud data acquisition module is a lidar, and the target area is the railway line that needs to be monitored for obstacles. The point cloud data refers to a set of data points obtained by lidar scanning and measurement, which can reflect the true situation of the target area with high precision, such as the ground state, the reflection characteristics of ground objects, and the characteristics of obstacles.

[0073] As Figure 2 , 3 shown, Figure 2 is a schematic diagram of a picture and point cloud data of a dendritic target included in a railway line provided by an embodiment of the application. Figure 3 is a partial enlarged schematic diagram of the dendritic target provided by an embodiment of the present application. When the lidar scans dendritic targets such as grass branches and tree branches, the dendritic targets will form point cloud data. At the same time, the lidar scanning points will penetrate the gaps of the dendritic targets and form background point clouds at the back of the dendritic targets.

[0074] Therefore, based on the dendritic target point cloud and the background point cloud, the dendritic targets in the point cloud data can be recognized. An embodiment of the present application can use a lidar to scan the railway track in real time and output three-dimensional point cloud data. The target recognition module recognizes obstacle targets based on the three-dimensional point cloud data. When dendritic target types such as grass branches and tree branches are recognized, no warning information is output to reduce false alarms caused by obstacles such as grass branches and tree branches and reduce the occurrence of events of false blocking of trains. When target types such as falling rocks and debris flows are recognized, warning information is output to improve the safety of railway operation and improve the practicality and accuracy of the three-dimensional obstacle system.

[0075] It should be noted that the technical solutions provided in the embodiments of the present application are described by taking the above-mentioned obstacle system as an example. It should be understood that the provided technical solutions are not limited to being applied in the obstacle system, but can also be applied to other systems related to target recognition.

[0076] As Figure 1 shown, the lidar collects the point cloud data of the railway line in real time and sends the point cloud data to the target recognition module for target recognition. Among them, the point cloud data obtained by the point cloud data acquisition module scanning the railway line completely once is called single-field point cloud data (PCD_FIELD_SINGLE).

[0077] The target recognition module executes a target recognition method based on point cloud data, which specifically includes the following content:

[0078] S100: Obtain point cloud data. The point cloud data is the three-dimensional point cloud data obtained by the lidar scanning the target area in real time. According to the radar point cloud resolution, in order to describe the regional scene more accurately and finely, multiple fields of point cloud data can be obtained for target recognition. To improve the recognition efficiency, rasterization downsampling processing can also be performed on the multiple fields of point cloud data to reduce the amount of point cloud data for subsequent analysis and processing. Among them, the point cloud data after rasterization downsampling processing can be called background point cloud (PCD_BK_GRID).

[0079] S200: Extract target point cloud and non-target point cloud from the point cloud data.

[0080] Among them, the target point cloud includes the data points in the point cloud data whose distance from the data points in the background point cloud is greater than the first distance, and the non-target point cloud includes the data points in the point cloud data whose distance from the data points in the background point cloud is less than or equal to the first distance. The background point cloud is the point cloud data after rasterization downsampling processing.

[0081] As Figure 4 shown, it is a schematic flow chart of extracting target point cloud and non-target point cloud provided by an embodiment of the present application. After the target recognition module obtains the point cloud data, rasterization downsampling processing can be performed on the point cloud data to obtain the background point cloud. Then traverse the data points in the point cloud data and calculate the distance between the data points in the point cloud data and the data points in the background point cloud. If the distance between the data points in the point cloud data and the data points in the background point cloud is greater than the first distance, then classify the data point in the point cloud data into the target point cloud. If the distance between the data points in the point cloud data and the data points in the background point cloud is less than or equal to the first distance, then classify the data point in the point cloud data into the target point cloud.

[0082] For example, after obtaining the single-field point cloud data (PCD_FIELD_SINGLE) collected by the lidar each time, distinguish the target point cloud and the non-target point cloud. The data points in the point cloud data (PCD_FIELD_SINGLE) whose distance from the data points in the background point cloud (PCD_BK_GRID) is greater than the first distance (DIST_BK) are classified into the target point cloud (PCD_OBJ), and the data points whose distance is less than or equal to the first distance (DIST_BK) are classified into the non-target point cloud (PCD_OBJ_NOT).

[0083] S300: Segment the target point cloud into sub-target point clouds.

[0084] Among them, the sub-target point cloud is a set of data points in the target point cloud whose distance between data points is less than the second distance.

[0085] After the target recognition module extracts the target point cloud and the non-target point cloud, it performs target segmentation and clustering on the data points in the target point cloud, and divides the target point cloud into multiple sub-target point clouds. As Figure 5 shown, it is a schematic flowchart of dividing the target point cloud into sub-target point clouds provided by an embodiment of the present application. The target recognition module traverses the data points in the target point cloud and calculates the distances between the data points in the target point cloud. If the distance between two data points is less than the second distance (DIST_SEG), the two data points are classified into the sub-target point cloud. That is, the data points in the target point cloud with a distance less than the second distance (DIST_SEG) are classified into the same sub-target point cloud (PCD_OBJ_SEG). After segmentation, the target point cloud will form one or more sub-target point clouds (PCD_OBJ_SEG).

[0086] S400: Construct a projection plane based on the sub-target point cloud.

[0087] After the target recognition module obtains the sub-target point cloud, it can perform target recognition on the sub-target point cloud to identify the target type of the sub-target point cloud. If the target type is a branch type, no alarm is output. If the target type is a non-branch type, an alarm is output.

[0088] The target recognition module can construct a projection plane based on the sub-target point cloud and perform subsequent recognition processing according to the projection plane. As Figure 6 shown, it is a schematic flowchart of constructing a projection plane provided by an embodiment of the present application. The target recognition module can calculate the average coordinate point of the sub-target point cloud. Obtain the reference coordinate point of the point cloud data. Construct a normal line passing through the average coordinate point and the reference coordinate point according to the average coordinate point and the reference coordinate point, and then construct a projection plane perpendicular to the normal line and passing through the average coordinate point. Among them, the average coordinate point is the average value of the coordinate values of the data points in the sub-target point cloud. The reference coordinate point is the position point where the laser signal is emitted when collecting the point cloud data. During the process of collecting point cloud data, the lidar emits laser signals around and receives the reflected laser signals to obtain the three-dimensional coordinate information of the target. This process takes the radar center as the reference point, and this reference coordinate point is fixed and represents the position of the lidar. The projection plane passes through the average coordinate point of the sub-target point cloud, and the normal line of the projection plane passes through the reference coordinate point of the point cloud data and the average coordinate point of the sub-target point cloud.

[0089] In some embodiments, when the target recognition module calculates the average coordinate point of the sub-target point cloud, it can calculate the average value of the horizontal axis coordinate values of the data points in the sub-target point cloud to obtain the horizontal axis coordinate value of the average coordinate point. Calculate the average value of the vertical axis coordinate values of the data points in the sub-target point cloud to obtain the vertical axis coordinate value of the average coordinate point. Calculate the average value of the vertical axis coordinate values of the data points in the sub-target point cloud to obtain the vertical axis coordinate value of the average coordinate point.

[0090] For example, continuing with the above example, calculate the average coordinate point P_CENTER(x, y, z) of the sub-target point cloud (PCD_OBJ_SEG) according to the following formula:

[0091] P_CENTER.x = the average value of the x-coordinate values of all data points in the sub-target point cloud;

[0092] P_CENTER.y = the average value of the y-coordinate values of all data points in the sub-target point cloud;

[0093] P_CENTER.z = the average value of the z-coordinate values of all data points in the sub-target point cloud;

[0094] Among them, P_CENTER.x, P_CENTER.y, and P_CENTER.z are the x-axis coordinate value, y-axis coordinate value, and z-axis coordinate value of the average coordinate point P_CENTER(x, y, z) respectively.

[0095] The reference coordinate point is R_CENTER(x, y, z), which is the radar center point. According to the average coordinate point P_CENTER(x, y, z) and the reference coordinate point R_CENTER(x, y, z), construct a projection plane. Among them, the projection plane passes through the average coordinate point P_CENTER(x, y, z) of the sub-target point cloud, and its normal passes through the reference coordinate point R_CENTER(x, y, z) and the average coordinate point P_CENTER(x, y, z).

[0096] The projection plane is as follows:

[0097] a(x - P_CENTER.x) + b(y - P_CENTER.y) + c(z - P_CENTER.z) = 0;

[0098] a = (R_CENTER.x - P_CENTER.x) / dist_pr;

[0099] b = (R_CENTER.y - P_CENTER.y) / dist_pr;

[0100] c = (R_CENTER.z - P_CENTER.z) / dist_pr;

[0101] Among them, dist_pr is the distance between the reference coordinate point R_CENTER(x, y, z) and the average coordinate point P_CENTER(x, y, z) of the sub-target point cloud.

[0102] S500: Project the sub-target point cloud and the non-target point cloud onto the projection plane respectively to obtain the sub-target projection point cloud and the non-target projection point cloud.

[0103] Such asFigure 7 As shown in the figure, it is a schematic flowchart of identifying the target type of the sub-target point cloud provided by the embodiment of the present application. After the target recognition module constructs a projection plane based on the sub-target point cloud, the sub-target point cloud and the non-target point cloud can be projected onto the projection plane respectively to convert the three-dimensional point cloud data into a two-dimensional projected point cloud. That is, the sub-target point cloud (PCD_OBJ_SEG) is projected onto the projection plane (PLANE_PROJECT) to form a two-dimensional sub-target projected point cloud (PCD_OBJ_SEG_PROJECT). The non-target point cloud (PCD_OBJ_NOT) is projected onto the projection plane (PLANE_PROJECT) to form a two-dimensional non-target projected point cloud (PCD_OBJ_NOT_PROJECT).

[0104] S600: Extract target data points from the non-target projected point cloud based on the sub-target projected point cloud.

[0105] Among them, the target data points are the data points located within the boundary polygon region, and the boundary polygon region is a region constructed based on the boundary of the sub-target projected point cloud. The target recognition module can identify the boundary of the sub-target projected point cloud, extract the data points on the boundary, use the data points on the boundary as vertices to construct the boundary polygon region. Finally, the target data points located within the boundary polygon region are extracted from the non-target projected point cloud.

[0106] The target recognition module can traverse the data points of the non-target projected point cloud. If the data point is located within the boundary polygon region, the data point is marked as a target data point. If the data point is not located within the boundary polygon region, the data point is marked as a non-target data point. Thus, a point cloud data set of the target data points located within the boundary polygon region is obtained for subsequent target recognition.

[0107] For example, following the above example, identify the concave boundary of the sub-target projected point cloud (PCD_OBJ_SEG_PROJECT), and store the data points on the boundary into the vertex vector (VERTEX_POLYGONS). Use the data points in the vertex vector (VERTEX_POLYGONS) to enclose a planar polygon, and extract the data points located inside the boundary polygon from the non-target projected point cloud (PCD_OBJ_NOT_PROJECT) and store them into the point cloud data set (PCD_OBJ_NOT_PROJECT_IN).

[0108] S700: Identify the target type of the sub-target point cloud according to the target data points.

[0109] After the target recognition module extracts the target data points located within the boundary polygon area, it can obtain the number of data points of the target data points and the total number of points of the sub-target projected point cloud. Then calculate the point ratio, where the point ratio is the ratio of the number of data points to the total number of points. If the point ratio is greater than the ratio threshold, mark the target type of the sub-target point cloud as the branch type. If the point ratio is less than or equal to the ratio threshold, mark the target type of the sub-target point cloud as the non-branch type.

[0110] For example, continuing with the above example, calculate the ratio of the number of data points in the point cloud data set (PCD_OBJ_NOT_PROJECT_IN) to the number of data points in the sub-target projected point cloud (PCD_OBJ_SEG_PROJECT). If this value is greater than the preset ratio threshold (GRASS_PERCENT), then determine that the sub-target point cloud is a branch target.

[0111] Thus, the determination of whether a single sub-target point cloud (PCD_OBJ_SEG) is of the branch type is completed. Repeat the above process to determine whether all other sub-target point clouds in the single-field point cloud data are branch targets. After processing the single-field point cloud data, receive new single-field point cloud data from the lidar and repeat the above processing process to monitor the railway line in real time.

[0112] For the single-field point cloud data, when a sub-target point cloud of the branch type is recognized, no alarm output is made. When target types such as falling rocks and debris flows are recognized, alarm information is output. Thus, effectively reducing false alarms caused by branch targets, improving the practicability and accuracy of the three-dimensional line obstacle system, and reducing the occurrence of false alarms and false train blocking events.

[0113] It should be noted that in the embodiments of the present application, no specific limitations are imposed on the above-mentioned first distance, second distance, ratio threshold, etc. Those skilled in the art can set them according to actual needs. For example, the parameters of the three-dimensional lidar are a horizontal angular resolution of 0.0625° and a vertical angular resolution of 0.4°. The first distance DIST_BK for distinguishing point cloud data and background point cloud is 8 cm. The second distance DIST_SEG for target segmentation is 12 cm. The ratio threshold GRASS_PERCENT for branch type determination is 15%.

[0114] Based on the above target recognition system based on point cloud data, the embodiments of the present application further provide a target recognition method based on point cloud data, and the method includes:

[0115] Collect the point cloud data of the target area.

[0116] Extract the target point cloud and non-target point cloud from the point cloud data. Among them, the target point cloud includes the data points in the point cloud data whose distance from the data points in the background point cloud is greater than the first distance. The non-target point cloud includes the data points in the point cloud data whose distance from the data points in the background point cloud is less than or equal to the first distance. The background point cloud is the point cloud data after performing rasterized downsampling processing.

[0117] Segment the target point cloud into sub-target point clouds. Among them, the sub-target point cloud is a set of data points in the target point cloud whose distance between data points is less than the second distance.

[0118] Construct a projection plane based on the sub-target point cloud. Among them, the projection plane passes through the average coordinate point of the sub-target point cloud, and the normal line of the projection plane passes through the reference coordinate point of the point cloud data and the average coordinate point of the sub-target point cloud.

[0119] Project the sub-target point cloud and the non-target point cloud onto the projection plane respectively to obtain the sub-target projected point cloud and the non-target projected point cloud.

[0120] Extract target data points from the non-target projected point cloud based on the sub-target projected point cloud. Among them, the target data points are the data points located within the boundary polygon region, and the boundary polygon region is a region constructed based on the boundary of the sub-target projected point cloud.

[0121] Identify the target type of the sub-target point cloud according to the target data points.

[0122] As can be seen from the above technical solutions, the target recognition method and system based on point cloud data provided by the above embodiments can collect the point cloud data of the target area, extract the target point cloud and the non-target point cloud from the point cloud data. Then segment the target point cloud into sub-target point clouds, and construct a projection plane based on the sub-target point cloud. Then project the sub-target point cloud and the non-target point cloud onto the projection plane respectively to obtain the sub-target projected point cloud and the non-target projected point cloud. Then extract target data points from the non-target projected point cloud based on the sub-target projected point cloud. Finally, identify the target type of the sub-target point cloud according to the target data points. The method uses the characteristic attributes of the target forming the background point cloud to perform target recognition of the point cloud data, improving the accuracy of target recognition.

[0123] For the similar parts between the embodiments provided in this application, reference can be made to each other. The specific implementation manners provided above are only several examples under the general concept of this application, and do not constitute a limitation on the protection scope of this application. For those skilled in the art, any other implementation manner extended based on the solution of this application without creative efforts belongs to the protection scope of this application.

Claims

1. A target recognition method based on point cloud data, characterized in that Including: Collecting point cloud data of a target area; Extracting target point cloud and non-target point cloud from the point cloud data, where the target point cloud includes data points in the point cloud data whose distance from the data points in the background point cloud is greater than a first distance, and the non-target point cloud includes data points in the point cloud data whose distance from the data points in the background point cloud is less than or equal to the first distance, and the background point cloud is the point cloud data after performing rasterization downsampling processing; Dividing the target point cloud into sub-target point clouds, where the sub-target point cloud is a set of data points in the target point cloud whose distance between data points is less than a second distance; Constructing a projection plane based on the sub-target point cloud, where the projection plane passes through the average coordinate point of the sub-target point cloud, and the normal line of the projection plane passes through the reference coordinate point of the point cloud data and the average coordinate point of the sub-target point cloud; Projecting the sub-target point cloud and the non-target point cloud onto the projection plane respectively to obtain sub-target projected point cloud and non-target projected point cloud; Based on the sub-target projected point cloud, extracting target data points in the non-target projected point cloud, where the target data points are data points located within a boundary polygon area, and the boundary polygon area is an area constructed based on the boundary of the sub-target projected point cloud; Identifying the target type of the sub-target point cloud according to the target data points.

2. The object recognition method based on point cloud data according to claim 1, wherein The step of constructing a projection plane based on the sub-target point cloud includes: Calculating the average coordinate point of the sub-target point cloud, where the average coordinate point is the average value of the coordinate values of the data points in the sub-target point cloud; Obtaining the reference coordinate point of the point cloud data, where the reference coordinate point is the position point where the laser signal is emitted when collecting the point cloud data; Constructing a normal line passing through the average coordinate point and the reference coordinate point according to the average coordinate point and the reference coordinate point; Constructing a projection plane perpendicular to the normal line and passing through the average coordinate point.

3. The object recognition method based on point cloud data according to claim 2, wherein The step of calculating the average coordinate point of the sub-target point cloud includes: Calculating the average value of the horizontal axis coordinate values of the data points in the sub-target point cloud to obtain the horizontal axis coordinate value of the average coordinate point; Calculating the average value of the vertical axis coordinate values of the data points in the sub-target point cloud to obtain the vertical axis coordinate value of the average coordinate point; Calculating the average value of the vertical axis coordinate values of the data points in the sub-target point cloud to obtain the vertical axis coordinate value of the average coordinate point.

4. The object recognition method based on point cloud data according to claim 1, characterized in that, The step of extracting target data points in the non-target projected point cloud based on the sub-target projected point cloud includes: Identifying the boundary of the sub-target projected point cloud; Extracting the data points on the boundary; Constructing a boundary polygon area with the data points on the boundary as vertices; Extracting target data points located within the boundary polygon area in the non-target projected point cloud.

5. The object recognition method based on point cloud data according to claim 4, wherein The step of extracting target data points located within the boundary polygon area in the non-target projected point cloud includes: Traversing the data points of the non-target projected point cloud; If the data point is located within the boundary polygon area, marking the data point as a target data point; If the data point is not located within the boundary polygon area, marking the data point as a non-target data point.

6. The object recognition method based on point cloud data according to claim 1, characterized in that The step of identifying the target type of the sub-target point cloud according to the target data points includes: Obtain the total number of points of the sub-target projected point cloud; Obtain the number of data points of the target data points; Calculate the point ratio, where the point ratio is the ratio of the number of data points to the total number of points; If the point ratio is greater than the ratio threshold, mark the target type of the sub-target point cloud as the branch type; If the point ratio is less than or equal to the ratio threshold, mark the target type of the sub-target point cloud as the non-branch type.

7. The target recognition method based on point cloud data according to claim 1, wherein The step of extracting the target point cloud and the non-target point cloud from the point cloud data includes: Perform rasterized downsampling processing on the point cloud data to obtain the background point cloud; Calculate the distance between the data points in the point cloud data and the data points in the background point cloud; If the distance between the data points in the point cloud data and the data points in the background point cloud is greater than the first distance, classify the data points in the point cloud data into the target point cloud; If the distance between the data points in the point cloud data and the data points in the background point cloud is less than or equal to the first distance, classify the data points in the point cloud data into the non-target point cloud.

8. The object recognition method based on point cloud data according to claim 1, characterized in that, The step of dividing the target point cloud into sub-target point clouds includes: Calculate the distance between the data points in the target point cloud; If the distance between two data points is less than the second distance, classify the two data points into the sub-target point cloud.

9. An object recognition system based on point cloud data, characterized in that, Includes: A point cloud data acquisition module for acquiring point cloud data of a target area; A target recognition module for extracting a target point cloud and a non-target point cloud from the point cloud data. The target point cloud includes data points in the point cloud data whose distance from the data points in the background point cloud is greater than the first distance, and the non-target point cloud includes data points in the point cloud data whose distance from the data points in the background point cloud is less than or equal to the first distance. The background point cloud is the point cloud data after rasterized downsampling processing; Divide the target point cloud into sub-target point clouds, where the sub-target point cloud is a set of data points in the target point cloud whose distance between data points is less than the second distance; Construct a projection plane based on the sub-target point cloud. The projection plane passes through the average coordinate point of the sub-target point cloud, and the normal of the projection plane passes through the reference coordinate point of the point cloud data and the average coordinate point of the sub-target point cloud; Project the sub-target point cloud and the non-target point cloud onto the projection plane respectively to obtain a sub-target projected point cloud and a non-target projected point cloud; Based on the sub-target projected point cloud, extract target data points from the non-target projected point cloud. The target data points are data points located within the boundary polygon area, and the boundary polygon area is an area constructed based on the boundary of the sub-target projected point cloud; Identify the target type of the sub-target point cloud according to the target data points.

10. The object recognition system based on point cloud data according to claim 9, characterized in that, The step of the target recognition module extracting target data points from the non-target projected point cloud based on the sub-target projected point cloud is also used for: Identify the boundary of the sub-target projected point cloud; Extract the data points on the boundary; Construct a boundary polygon area with the data points on the boundary as vertices; Extract target data points located within the boundary polygon region from the non-target projection point cloud.

Images