Tunnel scene recognition method and apparatus, and device and storage medium

Abstract

A tunnel scene recognition method, which is executed by a computer device. The method comprises: acquiring point cloud data corresponding to sampling path points on a motion path of a vehicle (S202); segmenting spatial points in the point cloud data according to spatial positions, so as to obtain a plurality of point cloud subsets (S204); in each point cloud subset among the plurality of point cloud subsets, emitting a ray to each spatial point in the point cloud subset by using a fixed reference point as the ray origin, and determining a ray cluster to which the ray belongs (S206); on the basis of a first distance between each spatial point in each ray cluster and the ray origin, determining closed-distribution characteristics of the spatial points in each point cloud subset; on the basis of the spatial points in each ray cluster, determining the center point of the ray cluster; and on the basis of a second distance between the center point and the ray origin, determining continuous-distribution characteristics of the spatial points in each point cloud subset (S208); and on the basis of the closed-distribution characteristics and the continuous-distribution characteristics of respective spatial points in the plurality of point cloud subsets, recognizing from the motion path a motion sub-path in a tunnel scene (S210).

Description

Tunnel scene recognition method, device, equipment and storage medium

[0001] Related applications

[0002] This application claims priority to Chinese patent application number 2024100230740, filed on January 8, 2024, entitled “Method, device, equipment and storage medium for identifying tunnel scenes,” the entire text of which is hereby incorporated by reference. Technical Field

[0003] The present application relates to the field of computer technology, and in particular to a method, apparatus, device, and storage medium for identifying a tunnel scene. Background Art

[0004] As a form of high-precision mapping, point cloud maps can provide rich three-dimensional information, helping systems achieve precise positioning and perception. Currently, before performing real-time positioning and building point cloud maps based on collected point cloud data, road scene recognition is usually performed on the collection path to obtain the collection path under specific road scenarios. This allows the point cloud data of the collection path under specific road scenarios to be mapped using appropriate registration algorithms, thereby improving the accuracy of the point cloud map.

[0005] Among them, the tunnel scene is a common special road scene. For the tunnel scene, the radar equipped on the acquisition vehicle is affected by the tunnel structure and closure in the tunnel, resulting in the point cloud data collected in the tunnel environment is not accurate enough. If the registration algorithm matching the tunnel scene is not used for mapping, it will lead to large errors in the point cloud map.

[0006] However, existing tunnel scene recognition methods usually have a large amount of computation and are difficult to accurately identify the motion sub-paths in the tunnel scene in the acquisition path, resulting in poor recognition efficiency and accuracy of the tunnel scene. Summary of the Invention

[0007] Based on this, it is necessary to provide a tunnel scene recognition method, device, computer equipment, computer-readable storage medium and computer program product that can improve the tunnel scene path recognition effect in response to the above technical problems.

[0008] In a first aspect, the present application provides a method for identifying a tunnel scene, which is performed by a computer device, and the method includes:

[0009] Obtaining point cloud data corresponding to the vehicle's motion path position;

[0010] Segmenting the spatial points in the point cloud data according to spatial positions to obtain multiple point cloud subsets;

[0011] In the point cloud subset, emitting a ray toward each spatial point in the point cloud subset with a fixed reference point as a ray origin, and determining the ray cluster to which the ray belongs;

[0012] determining, based on a first distance between each spatial point in the ray cluster and the ray origin, a closed distribution feature of the spatial points in the point cloud subset;

[0013] Determining a center point of the ray cluster based on the spatial points in the ray cluster, and determining a continuity distribution feature of the spatial points in the point cloud subset according to a second distance between the center point and the ray origin; and

[0014] A motion sub-path in a tunnel scene is identified in the motion path according to the closed distribution characteristics and the continuous distribution characteristics of the spatial points of each of the plurality of point cloud subsets.

[0015] In a second aspect, the present application further provides a tunnel scene recognition device, the device comprising:

[0016] A point cloud data acquisition module, used to acquire point cloud data corresponding to the motion path position of the vehicle;

[0017] a point cloud data segmentation module, configured to segment the spatial points in the point cloud data according to spatial positions to obtain a plurality of point cloud subsets;

[0018] a ray cluster determination module, configured to emit a ray from the point cloud subset toward each spatial point in the point cloud subset using a fixed reference point as a ray origin, and determine the ray cluster to which the ray belongs;

[0019] a distribution feature determination module, configured to determine a closed distribution feature of spatial points in the point cloud subset based on a first distance between each spatial point in the ray cluster and the ray origin; determine a center point of the ray cluster based on the spatial points in the ray cluster, and determine a continuous distribution feature of spatial points in the point cloud subset based on a second distance between the center point and the ray origin; and

[0020] A scene recognition module is used to identify a motion sub-path in a tunnel scene in the motion path based on the closed distribution characteristics and continuous distribution characteristics of the spatial points of each of the multiple point cloud subsets.

[0021] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the above-mentioned tunnel scene recognition method when executing the computer program.

[0022] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned tunnel scene recognition method.

[0023] In a fifth aspect, the present application further provides a computer program product, which includes a computer program that implements the above-mentioned tunnel scene recognition method when executed by a processor.

[0024] The details of one or more embodiments of the present application are set forth in the accompanying drawings and the description below. Other features, objects, and advantages of the present application will become apparent from the description, drawings, and claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the conventional technology, the following briefly introduces the drawings required for use in the embodiments or the conventional technology descriptions. Obviously, the drawings described below are merely embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the disclosed drawings without any creative work.

[0026] FIG1 is a diagram illustrating an application environment of a method for identifying a tunnel scene in one embodiment;

[0027] FIG2 is a schematic flow chart of a method for identifying a tunnel scene in one embodiment;

[0028] FIG3 is a schematic diagram of point cloud segmentation of point cloud data in one embodiment;

[0029] FIG4 is a schematic diagram of ray clusters corresponding to a point cloud subset in one embodiment;

[0030] FIG5 is a schematic diagram of a single frame of point cloud data in one embodiment;

[0031] FIG6 is a schematic diagram of stitching point cloud data in one embodiment;

[0032] FIG7 is a schematic diagram of point cloud data in another embodiment;

[0033] FIG8 is a schematic flow chart of a method for identifying a tunnel scene in another embodiment;

[0034] FIG9 is a schematic diagram of point cloud data of a real scene in one embodiment;

[0035] FIG10 is a schematic diagram of point cloud data of a real scene in another embodiment;

[0036] FIG11 is a schematic diagram of point cloud data of a real scene in another embodiment;

[0037] FIG12 is a schematic diagram of point cloud data of a real scene in another embodiment;

[0038] FIG13 is a block diagram of a tunnel scene recognition device according to an embodiment;

[0039] FIG14 is a structural block diagram of a tunnel scene recognition device according to another embodiment;

[0040] FIG15 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0042] The tunnel scene recognition method provided in this application can be applied to the field of intelligent transportation, and specifically to the field of autonomous driving of intelligent transportation.

[0043] The tunnel scene recognition method provided in an embodiment of the present application can be applied in the application environment shown in Figure 1. Terminal 102 communicates with server 104 via a network. A data storage system can store data that server 104 needs to process. The data storage system can be integrated with server 104, or located in the cloud or on another server. The tunnel scene recognition method can be executed independently by terminal 102 or server 104, or by both terminal 102 and server 104. In some embodiments, the tunnel scene recognition method is executed by the terminal 102, which obtains point cloud data collected by the vehicle on the motion path; divides the spatial points in the point cloud data according to the spatial position to obtain multiple point cloud subsets; in the point cloud subset, uses a fixed reference point as the ray origin to emit rays toward each spatial point in the point cloud subset, and determines the ray cluster to which the ray belongs; determines the closed distribution characteristics of the spatial points in the point cloud subset based on the first distance between each spatial point in the ray cluster and the ray origin; determines the center point of the ray cluster based on the spatial points in the ray cluster, and determines the continuity distribution characteristics of the spatial points in the point cloud subset based on the second distance between the center point and the ray origin; identifies the motion sub-path under the tunnel scene in the motion path based on the closed distribution characteristics and continuity distribution characteristics of the spatial points of each of the multiple point cloud subsets.

[0044] Among them, the terminal 102 can be, but is not limited to, various desktop computers, laptops, smart phones, tablets, Internet of Things devices and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart car-mounted devices, etc. Portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The server 104 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The terminal 102 and the server 104 can be directly or indirectly connected through wired or wireless communication, and this application does not impose any restrictions on this.

[0045] In one embodiment, as shown in FIG2 , a method for identifying a tunnel scene is provided. The method is described by applying the method to the computer device in FIG1 as an example, and includes the following steps:

[0046] S202: Acquire point cloud data corresponding to the motion path position of the vehicle.

[0047] The vehicle can be a collection vehicle or another type of vehicle, such as an autonomous vehicle. A collection vehicle is a vehicle specifically designed to collect road environment data. It typically has multiple sensor devices, such as lidar, cameras, GPS, and IMUs, enabling real-time acquisition of information such as the vehicle's position, attitude, and speed, as well as point cloud data of the road environment.

[0048] A data collection vehicle typically drives along a road, scanning and collecting information about the road environment using multiple sensor devices. Devices like lidar and cameras can be used to acquire point cloud data and image data from the road environment, while devices like GPS and IMUs can be used to obtain the vehicle's position and posture information, thereby constructing its motion path.

[0049] In addition to data collection vehicles, other vehicles can also acquire motion paths and point cloud data by installing various sensor devices. For example, autonomous vehicles can use sensors such as lidar and cameras to acquire road environment data for autonomous navigation and control. It should be noted that compared to data collection vehicles, other vehicles typically require more complex algorithm design and system integration to achieve accurate and stable data collection.

[0050] The motion path can be the original motion path of the vehicle, or it can be the path obtained after sparse processing of the original motion path. The original motion path is a continuous sequence of position points passed by the vehicle during driving. Specifically, a series of position points can be used to represent the vehicle's driving route and path; the original motion path corresponds to an original point cloud dataset, which refers to a data set composed of multiple frames of point cloud data collected based on the original motion path. It usually contains point cloud data collected at multiple time points and is used to describe the road environment during the vehicle's driving process. Each frame of point cloud data contains three-dimensional point cloud information of the road environment obtained from the sensor at the time point corresponding to the frame of point cloud data.

[0051] The point cloud data corresponding to the vehicle's motion path position may be point cloud data corresponding to a path point position on the vehicle's motion path, and the point cloud data corresponding to a path point position may be all or part of the point cloud data collected at the path point.

[0052] It is understandable that the point cloud data frames in the original point cloud dataset may be unevenly distributed due to the non-constant driving speed of the vehicle during the acquisition process, such as the problem of a large number of spatial points being accumulated due to congestion or parking. In order to avoid these problems, the original motion path can be sparsely processed to obtain the motion path, and the original point cloud dataset can be sparsely processed at the same time to obtain the sparsely processed point cloud dataset. The path obtained after sparse processing corresponds to the sparsely processed point cloud dataset.

[0053] Sparse processing can specifically be to uniformly sample the original motion path to obtain a path, and based on the sparse path points on the path obtained by uniform sampling, sample the original point cloud dataset to obtain a sparsely processed point cloud dataset. Specifically, the original motion path can be equidistantly sampled at preset distance intervals to obtain sparse path points. The path composed of the sparse path points is the path obtained after sparse processing; for each sparse path point on the path obtained after sparse processing, obtain a frame of original point cloud data collected at the corresponding position of the sparse path point from the original point cloud dataset. The set of one frame of original point cloud data corresponding to each initial path point is the sparsely processed point cloud dataset. The order of the sparse path points in the path obtained after sparse processing is the frame sequence of the original point cloud data of different frames in the sparsely processed point cloud dataset.

[0054] For example, a certain original motion path is sampled at intervals of 2 meters, and 60 sparse path points are obtained in sequence, namely sparse path point 01, sparse path point 02, sparse path point 03, ..., sparse path point 60. The frame sequence of a frame of original point cloud data collected at the corresponding positions of the above sparse path points in the original point cloud dataset is frame 1, frame 5, frame 8, frame 17, ... frame 300, and the original point cloud data of each frame corresponding to these frame sequences are extracted to form a new sparsely processed point cloud dataset. In the sparsely processed point cloud dataset, the frame sequence of the above frames of original point cloud data can be recorded as frame 1, frame 2, frame 3, ..., frame 60, where the collection position of the original point cloud data of the first frame is sparse path point 01, and the collection position of the original point cloud data of the second frame is sparse path point 02. By analogy, the sparsely processed point cloud dataset contains a frame of original point cloud data collected at each sparse path point at its position.

[0055] Specifically, the terminal controls the vehicle to travel in the target environment, and collects the vehicle's position and posture data through the GPS and IMU devices configured on the vehicle during the driving process, and then determines the vehicle's original motion path based on the collected vehicle's position and posture data, and collects point cloud data of the vehicle's environment through the laser radar configured on the vehicle to obtain an original point cloud data set corresponding to the vehicle's original motion path, and determines the vehicle's motion path based on the original motion path, and determines the point cloud data set corresponding to the motion path position based on the original point cloud data set, obtains the original point cloud data corresponding to the motion path from the point cloud data set, and filters the original point cloud data corresponding to the motion path according to preset conditions to obtain the point cloud data corresponding to the motion path.

[0056] It can be understood that when the original motion path is directly determined as the motion path, the point cloud dataset corresponding to the motion path is the original point cloud dataset; when the motion path is the path obtained after sparse processing of the original motion path, the point cloud dataset corresponding to the motion path is the sparsely processed point cloud dataset.

[0057] The original point cloud data corresponding to the motion path can be all the original point cloud data in the point cloud dataset, or it can be the original point cloud data corresponding to the sampled path points on the motion path in the point cloud dataset. The sampled path points are obtained by point sampling of the motion path. Point sampling refers to selecting specific path points in the motion path.

[0058] The original point cloud data corresponding to the sampling path point in the point cloud dataset can be a frame of original point cloud data collected at the corresponding position of the sampling path point; it can be understood that since the scanning range of a single frame of original point cloud data is small and may be unevenly distributed, it is easily interfered by noise and is not accurate enough. Therefore, the original point cloud data corresponding to the sampling path point can also include the original point cloud data of other frames adjacent to the frame of original point cloud data in the point cloud dataset. That is to say, the original point cloud data corresponding to the sampling path point in the point cloud dataset can also be a frame of original point cloud data and adjacent frames of original point cloud data collected at the corresponding position of the sampling path point. The adjacent frames of original point cloud data are the original point cloud data of other frames adjacent to the frame of original point cloud data collected at the corresponding position of the sampling path point in the point cloud dataset. For example, if the original point cloud data of the 100th frame in the point cloud dataset is collected at sampling path point A, the original point cloud data of the 100th frame can be determined as the original point cloud data corresponding to sampling path point A, or the original point cloud data of the 99th, 100th and 101st frames can be determined as the original point cloud data corresponding to sampling path point A.

[0059] In one embodiment, the process of a computer device screening the original point cloud data corresponding to the motion path according to preset conditions includes: when the original point cloud data corresponding to the motion path is a frame, selecting point cloud data that meets the height condition from the original point cloud data of the frame, and the point cloud data that meets the height condition is the point cloud data corresponding to the motion path.

[0060] The height condition can specifically be to meet a preset height range. This height range can be a distance relative to the vehicle, the ground, or other reference plane. For example, the area above 2.5 meters above the ground can be selected as the height range of interest, thereby avoiding the influence of road surface elements on the tunnel scene recognition effect in subsequent processing.

[0061] It should be noted that when there are at least two frames of original point cloud data corresponding to the sampling path points, the original point cloud data of at least two frames can be spliced ​​first to obtain spliced ​​point cloud data, and point cloud data that meets the height conditions can be selected from the spliced ​​point cloud data.

[0062] S204 , dividing the spatial points in the point cloud data according to their spatial positions to obtain a plurality of point cloud subsets.

[0063] Segmenting the spatial points in point cloud data can be called point cloud segmentation, or point cloud slicing. Point cloud segmentation involves dividing point cloud data composed of a large number of spatial points into smaller point cloud subsets based on their spatial locations. Point cloud subsets are defined from the point cloud data, each containing spatial points within a specific region of the point cloud data. Point cloud subsets can also be referred to as point cloud slicing results. Point cloud segmentation can be performed along a plane perpendicular (or approximately perpendicular) to the direction of vehicle travel when collecting the point cloud data.

[0064] Specifically, after obtaining the point cloud data corresponding to the motion path position, the computer device can set appropriate segmentation parameters for the point cloud data corresponding to the motion path position, and perform point cloud segmentation on the point cloud data based on the determined segmentation parameters to obtain multiple point cloud subsets of the point cloud data.

[0065] The segmentation parameters refer to the parameters used when performing point cloud segmentation on point cloud data, which are used to determine the size and shape of the obtained point cloud subsets, as well as the spatial relationship between the point cloud subsets.

[0066] It can be understood that if the point cloud data corresponding to the motion path position is obtained by filtering all the original point cloud data in the point cloud data set corresponding to the motion path, appropriate segmentation parameters can be set so that the width of the obtained point cloud subset is larger, so that the point cloud subsets obtained by point cloud segmentation of the point cloud data correspond to a part of the sub-path in the motion path, and by analyzing the closed distribution characteristics and continuity distribution characteristics of the spatial points in the point cloud subset, the scene type of the corresponding sub-path can be determined based on the closed distribution characteristics and continuity distribution characteristics of the spatial points in the point cloud subset; if the point cloud data corresponding to the motion path position is obtained by filtering the original point cloud data corresponding to the sampling path points on the motion path, the width of each point cloud subset corresponding to each sampling path point is smaller, that is, the number of spatial points in the point cloud subset is smaller, thereby improving the efficiency of analyzing each point cloud subset for each sampling path point.

[0067] As shown in (A) in FIG3 , this is the point cloud data after stitching together multiple frames of original point cloud data in one embodiment. After the point cloud data is obtained by screening according to the height condition, the point cloud data is segmented, thereby obtaining the five point cloud subsets shown in (B) in FIG3 .

[0068] S206 , in the point cloud subset, emitting a ray toward each spatial point in the point cloud subset with the fixed reference point as the ray origin, and determining the ray cluster to which the ray belongs.

[0069] Among them, the spatial points correspond to rays, which refer to the straight-line paths from a fixed reference point to the spatial points. The fixed reference point can also be called the ray origin. The direction and length of the rays of each spatial point can reflect the shape characteristics of the point cloud subset to which it belongs, that is, the distribution characteristics of the spatial points in the point cloud subset. The length of the ray refers to the distance from the spatial point to the ray origin. A ray cluster is a collection of rays with similar directions. Rays with similar directions may refer to rays whose direction deviations are within a preset range, and the maximum value of the direction angles between any two rays in the ray cluster is less than the preset direction angle. The fixed reference point may be a point that represents the position of the vehicle when collecting spatial points in the point cloud subset. Specifically, it may be any point or center point on the vehicle, or it may be the point at the center of the point cloud subset. The center may be the center of mass.

[0070] Specifically, after obtaining each point cloud subset, the computer device determines the corresponding ray origin and reference coordinate axis for any point cloud subset, and constructs the ray corresponding to each spatial point in the point cloud subset based on the ray origin, constructs a coordinate system according to the ray origin, and selects the reference coordinate axis in the coordinate system, and determines the ray cluster to which each ray belongs based on the positional relationship between the ray and the reference coordinate axis.

[0071] In one embodiment, S206 specifically includes the following steps: in the point cloud subset, a fixed reference point is used as the ray origin, and a reference coordinate axis is determined; rays are emitted from the ray origin toward each spatial point in the point cloud subset; the angle between each ray and the reference coordinate axis is determined; and the ray cluster to which each ray belongs is determined based on the angle.

[0072] Among them, the ray origin and the reference coordinate axis can be selected according to actual needs. The position of the ray origin can specifically be the position of the laser radar itself when collecting the point cloud subset. It can be understood that the position of the laser radar itself and the position of the spatial points in the point cloud subset are discussed in the same coordinate system. If the coordinate systems of the two are inconsistent, corresponding coordinate changes need to be made to make the coordinate systems of the two consistent. For example, for convenience, the point cloud data in the world coordinate system can be converted to the laser radar coordinate system, and then the edge trimming and ray cluster construction are performed in the laser radar coordinate system. The laser radar coordinate system is a spatial reference system used to describe the point cloud data captured by the laser radar sensor. In the case where the laser radar is installed on a vehicle, the laser radar coordinate system can also be called the vehicle coordinate system. Its x-axis usually points to the direction of the vehicle's movement, the x-axis usually points to the right side of the vehicle, and the z-axis usually points vertically upward, perpendicular to the ground.

[0073] Specifically, after determining the origin of the ray, the computer device can construct a coordinate system based on the ray origin with the ray origin as the coordinate origin, the vehicle's forward direction as the positive direction of the x-axis, the vertical upward direction as the positive direction of the z-axis, and the direction perpendicular to the x-axis and z-axis and conforming to the right-hand coordinate system rule as the y-axis. The ray origin is used as the starting point and the spatial points in the point cloud subset are used as the end points to generate rays corresponding to each spatial point. The y-axis of the coordinate system is selected as the reference coordinate axis, and the angle between each ray and the reference coordinate axis, i.e., the y-axis, is determined. For any ray, the ray cluster to which it belongs is determined according to the corresponding angle.

[0074] In the above embodiment, the computer device determines the ray origin and the reference coordinate axis corresponding to the point cloud subset; takes the ray origin as the starting point and each spatial point in the point cloud subset as the end point, generates rays corresponding to each spatial point, and connects the spatial points to a common origin, so that the structure and shape in the point cloud data can be seen more clearly, and the angle between each ray and the reference coordinate axis is determined. Based on the angle, the ray cluster to which each ray belongs is determined, so that the distribution of spatial points in the point cloud subset can be analyzed based on the ray cluster, thereby improving the accuracy and efficiency of the analysis of the distribution of spatial points in the point cloud subset, and when the motion sub-path in the tunnel scene is subsequently determined based on the distribution of the point cloud data, the accuracy and efficiency of the motion sub-path identification in the tunnel scene can be improved.

[0075] In one embodiment, a process in which a computer device determines the ray cluster to which each ray belongs based on an included angle includes the following steps: obtaining a ray cluster angle value corresponding to each ray cluster; and when the angle difference between the included angle and a target ray cluster angle value in the ray cluster angle value satisfies an angle difference threshold, dividing the ray corresponding to the included angle into the ray cluster corresponding to the target ray cluster angle value.

[0076] Among them, the angle difference threshold is the maximum difference value between the ray allowed by the ray cluster and the corresponding ray cluster angle value, which can be set according to actual needs. It can be understood that when there are more spatial points in the point cloud subset, in order to reduce the amount of calculation, the angle difference threshold can be set to a smaller value to achieve sparse sampling of spatial points when constructing the ray cluster. When there are fewer spatial points in the point cloud subset, in order to improve the accuracy, the angle difference threshold can be set to a larger value. This ensures that even when there are fewer spatial points, a sufficient number of rays can be classified into appropriate ray clusters.

[0077] Among them, the ray cluster angle value is a predefined value used to characterize the angular characteristics of the ray cluster. For example, 18 ray clusters are predefined according to actual needs, and a specific angle value is assigned to each ray cluster. The position of the ray origin is the position of the lidar itself when collecting the point cloud subset. It can be determined that the angular distribution of the 18 ray clusters needs to cover a range of 180 degrees. In order to avoid errors caused by the small number of spatial points on the left and right endpoints of the point cloud subset, a ray cluster can be defined starting from 5 degrees and every 10 degrees to obtain a ray cluster set. The ray cluster angle value set corresponding to this ray cluster set is

[0078] Specifically, for any ray, after determining the angle between the ray and the reference coordinate axis, the angle difference between the angle and the angle value of each ray cluster can be determined respectively, and each angle difference can be compared with the angle difference threshold. When the angle difference between the angle and the angle value of the target ray cluster is less than the angle difference threshold, the ray corresponding to the angle is divided into the ray cluster corresponding to the angle value of the target ray cluster.

[0079] In one embodiment, the angle θ and the ray cluster angle θ i When the relationship between the angle difference threshold ε satisfies the following relationship, the ray corresponding to the angle θ is divided into the target ray cluster angle value θ i In the corresponding ray cluster, in the embodiment of the present application, ε can be taken as 0.2 degrees: |θ-θ i |<ε

[0080] FIG4 is a schematic diagram of ray clusters corresponding to multiple point cloud subsets of a certain sampling path point in one embodiment. The rays shown in the figure are rays of the corresponding ray cluster, and the angle between the ray and the reference coordinate axis is the ray cluster angle value corresponding to the corresponding ray cluster.

[0081] In the above embodiment, the computer device obtains the ray cluster angle value corresponding to each ray cluster, and when the angle difference between the included angle and the target ray cluster angle value in the ray cluster angle value meets the angle difference threshold, the computer device divides the ray corresponding to the included angle into the ray cluster corresponding to the target ray cluster angle value, thereby accurately classifying the rays into the corresponding ray clusters and avoiding misclassification. When the distribution of spatial points in the point cloud subset is subsequently analyzed based on the ray cluster, the accuracy and efficiency of the distribution analysis of spatial points in the point cloud subset can be improved. When the motion sub-path in the tunnel scene is subsequently determined based on the distribution of point cloud data, the accuracy of the motion sub-path identification in the tunnel scene can be improved. By adjusting the angle difference threshold, point cloud data of various complexities can be flexibly processed to adapt to different scenarios and needs.

[0082] S208, determining the closed distribution characteristics of the spatial points in the point cloud subset based on the first distance between each spatial point in the ray cluster and the ray origin; determining the center point of the ray cluster based on the spatial points in the ray cluster, and determining the continuity distribution characteristics of the spatial points in the point cloud subset based on the second distance between the center point and the ray origin.

[0083] Among them, the closed distribution characteristics and continuous distribution characteristics of the spatial points in the point cloud subset are two distribution characteristics of the spatial points in the point cloud subset. The distribution characteristics refer to the distribution patterns and geometric properties of the spatial points in the point cloud subset. The closed distribution characteristics are used to describe whether the distribution of each spatial point in the point cloud subset forms a closed structure, such as a boundary around a certain area or shape. Specifically, it can be closed or not. Closure means that the distribution of spatial points in the point cloud subset forms a closed outline or boundary, such as a circle, polygon, etc., which indicates a high degree of closure. Non-closure means that the distribution of spatial points in the point cloud subset does not form a closed structure, such as a linear distribution or a scattered distribution. The continuous distribution characteristics are used to describe whether the distribution of each spatial point in the point cloud subset is continuous. Specifically, it can be continuous or discontinuous. Continuity means that the distribution of the point cloud subset in space is smooth and consistent, without obvious gaps or breaks. Discontinuity means that the distribution of the point cloud subset in space has gaps, interruptions or irregular distributions.

[0084] The center point refers to a feature point determined based on a spatial point in the ray cluster. The feature point is used to geometrically represent the ray cluster as a whole, and specifically may be the centroid or symmetry center of the ray cluster.

[0085] Specifically, for any ray cluster, the computer device can determine the first distance from each spatial point to the ray origin based on the position coordinates of each spatial point in the ray cluster and the position coordinates of the ray origin, and determine the position coordinates of the center point of the ray cluster according to the position coordinates of each spatial point, determine the second distance from the center point to the ray origin according to the position coordinates of the center point and the position coordinates of the ray origin, and determine the closed distribution characteristics and continuity distribution characteristics of the spatial points in the point cloud subset according to the first distance and second distance corresponding to each ray cluster in the point cloud subset.

[0086] S210 , identifying a motion sub-path in a tunnel scene in the motion path according to the closed distribution characteristics and continuous distribution characteristics of the spatial points of each of the plurality of point cloud subsets.

[0087] In one embodiment, if the point cloud data corresponding to the motion path position is obtained by screening all the original point cloud data in the point cloud data set corresponding to the motion path, the computer device determines the motion sub-path corresponding to each point cloud subset in the motion path after obtaining the closed distribution characteristics and continuity distribution characteristics of the spatial points in each point cloud subset. When the closed distribution characteristics of the spatial points in the target point cloud subset indicate that the point cloud subset is closed, and the continuity distribution characteristics indicate that the point cloud subset is continuous, the target point cloud subset is determined to be a tunnel point cloud subset, and the motion sub-path corresponding to the target point cloud subset in the motion path is identified as the motion sub-path in the tunnel scene. If the motion sub-path in the tunnel scene can be identified in the motion path, it means that the tunnel scene has been identified; if the motion sub-path in the tunnel scene cannot be identified in the motion path, it means that the tunnel scene has not been identified.

[0088] In one embodiment, if the point cloud data corresponding to the motion path position is obtained by filtering the original point cloud data corresponding to the sampling path points on the motion path, for any sampling path point in the motion path, after obtaining the closed distribution characteristics and continuity distribution characteristics of the spatial points in each point cloud subset of the sampling path point, the computer device can determine whether the distribution of the point cloud data corresponding to the sampling path point matches the typical characteristics of the tunnel scene based on the closed distribution characteristics and continuity distribution characteristics of the spatial points in each point cloud subset. The tunnel usually appears as a closed and continuous structure, and its shape and size remain consistent within a certain range. When the distribution of the point cloud data corresponding to the sampling path point matches the typical characteristics of the tunnel scene, it can be determined that the sampling path point is in the tunnel scene, thereby determining the sampling path point in the motion path that is in the tunnel scene, and identifying the motion sub-path in the tunnel scene based on the sampling path point in the motion path that is in the tunnel scene.

[0089] The above-mentioned tunnel scene recognition method provides a new solution for identifying tunnel scenes. The computer device performs point cloud segmentation on the point cloud data corresponding to the vehicle's position on the motion path and analyzes the distribution of each point cloud subset. Without analyzing the entire point cloud data, the closed distribution characteristics and continuity distribution characteristics of the spatial points in multiple point cloud subsets that can reflect the distribution characteristics of the point cloud data can be obtained. Based on the closed distribution characteristics and continuity distribution characteristics of the spatial points in the multiple point cloud subsets, the motion sub-path belonging to the tunnel scene in the motion path can be identified, thereby improving the recognition efficiency of the motion sub-path in the tunnel scene. By performing point cloud segmentation on the point cloud data and determining the ray cluster to which each spatial point in the obtained point cloud subset belongs, the closed distribution characteristics and continuity distribution characteristics of the spatial points in the point cloud subset can be accurately and quickly determined based on a first distance from each spatial point in the ray cluster to the ray origin and a second distance from the center point of the ray cluster to the ray origin. Subsequently, based on the closed distribution characteristics and continuity distribution characteristics of the spatial points in each point cloud subset, the motion sub-path in the tunnel scene in the motion path can be determined, further improving the accuracy and efficiency of motion sub-path recognition in the tunnel scene.

[0090] In one embodiment, the process of a computer device acquiring point cloud data corresponding to the motion path position of a vehicle includes the following steps: extracting at least two frames of original point cloud data corresponding to the sampling path points in the motion path from a point cloud data set corresponding to the motion path position of the vehicle; aligning and splicing the at least two frames of original point cloud data to obtain spliced ​​point cloud data; and selecting point cloud data that meets the height condition from the spliced ​​point cloud data.

[0091] Among them, registration and stitching refers to spatially aligning two or more frames of point cloud data to form a unified and continuous three-dimensional representation.

[0092] Specifically, after obtaining the motion path of the vehicle, the computer device can also perform sparse sampling on the motion path to obtain each sampling path point, and for each sampling path point, extract a frame of original point cloud data collected at the corresponding position of the sampling path point from the point cloud data set, and obtain the original point cloud data of the adjacent frames of the original point cloud data of the frame in the point cloud data set, thereby obtaining at least two frames of original point cloud data, and use a preset alignment algorithm to align and splice the at least two frames of original point cloud data to obtain spliced ​​point cloud data, and obtain the height threshold corresponding to the height condition, and select each spatial point whose height is greater than or equal to the height threshold from the spliced ​​point cloud data to obtain point cloud data that meets the height condition.

[0093] It should be noted that the position coordinates of each space in the original point cloud data can specifically be the coordinates in the lidar coordinate system. When splicing the original point cloud data, the coordinate system can be converted on the left side of the position of the original point cloud data to obtain the position coordinates of the original point cloud data in the world coordinate system, and at least two frames of original point cloud data can be aligned and spliced ​​in the world coordinate system.

[0094] For example, the initial posture T of the vehicle coordinate system in the world coordinate system can be determined based on the GPS and IMU devices configured on the vehicle. wv The initial laser radar external parameters of the laser radar are used as the transformation matrix T from the vehicle coordinate system to the laser radar coordinate system vl , so according to the transformation matrix T vl and the initial posture T wv The position T of the laser radar coordinate origin in the world coordinate system can be determined wl , where the transformation relationship involved in the above process is as follows: T wl =T wv ·T vl

[0095] The position coordinates of the spatial point P in any frame of the original point cloud data extracted can be recorded as P(X, Y, Z) in the laser radar coordinate system, and the position T of the laser radar coordinate origin in the world coordinate system is wl , and the following transformation relationship can be used to convert the position coordinates of the spatial point P in any frame of original point cloud data into the coordinates in the world coordinate system, where the transformation relationship is as follows: P w =T wl ·P

[0096] By performing coordinate transformation on at least two frames of original point cloud data, at least two frames of original point cloud data in a world coordinate system are obtained, and the at least two frames of original point cloud data are aligned and spliced ​​in the world coordinate system, so that more evenly distributed spliced ​​point cloud data can be obtained. In the process of lidar scanning, due to the occlusion of interference objects, a certain frame of scanned point cloud data may be inaccurate due to the existence of holes. By aligning and splicing at least two frames of original point cloud data, the hole problem caused by occlusion can be alleviated, thereby improving the recognition accuracy when performing scene recognition based on the spliced ​​point cloud data.

[0097] As shown in Figure 5, a frame of original point cloud data is shown, and as shown in Figure 6, the spliced ​​point cloud data obtained after registration and splicing of multiple frames of original point cloud data. It can be seen from the figure that the overall distribution of the spliced ​​point cloud data is relatively uniform.

[0098] In the above embodiment, the computer device can obtain more continuous and complete point cloud data corresponding to the sampled path points by aligning and splicing at least two frames of original point cloud data, thereby reducing the omission problem that may be encountered when processing single-frame data, thereby improving the recognition accuracy of motion sub-paths in tunnel scenarios. In addition, by selecting point cloud data that meets specific height conditions, it helps to eliminate interfering elements on the ground (such as pedestrians, vehicles, road markings, etc.), focus on key structures such as the tunnel top, and avoid interference with tunnel scene recognition caused by irrelevant elements, thereby improving the recognition accuracy and efficiency of motion sub-paths in tunnel scenarios.

[0099] In one embodiment, a computer device divides spatial points in point cloud data according to spatial positions to obtain multiple point cloud subsets, including the following steps: determining the segmentation plane, point cloud subset size and point cloud subset interval of the point cloud data; determining the point cloud subset coordinate conditions based on the segmentation plane, point cloud subset size and point cloud subset interval; dividing the spatial points in the point cloud data that meet the same point cloud subset coordinate conditions into the same point cloud subset to obtain a preset number of point cloud subsets.

[0100] The segmentation parameters may specifically include a segmentation plane, a point cloud subset size, and a point cloud subset interval. The segmentation plane is a two-dimensional plane used to divide the point cloud data into smaller parts. The segmentation plane can be determined according to actual needs. For example, the segmentation plane can be a horizontal plane, a vertical plane, or a plane at any angle. The point cloud subset size presets the size of the spatial range covered by the point cloud subset. For example, a point cloud subset can be set to cover a space with a thickness of 2cm, 5cm, or 10cm; the point cloud subset interval refers to the distance between adjacent point cloud subsets when performing point cloud segmentation of point cloud data. The point cloud subset interval determines whether there is overlap or gap between point cloud subsets. A smaller interval may cause overlap between point cloud subsets, while a larger interval may cause gaps between data. In an embodiment of the present application, the point cloud subset is a sparse sampling of the point cloud data. Therefore, the setting size of the point cloud subset interval satisfies the requirement that there is a certain gap between adjacent point cloud subsets.

[0101] The splitting plane, point cloud subset size, and point cloud subset interval can be selected based on the purpose of point cloud data analysis. Specifically, an appropriate splitting plane for point cloud data can be determined based on the data analysis purpose of identifying motion subpaths in tunnel scenes. For example, in the embodiment of the present application, for identifying tunnel scenes, the splitting plane for point cloud data in tunnel scenes can be selected to be a plane parallel to the cross-section of the point cloud data, where the cross-section of the point cloud data refers to a plane perpendicular to the direction of vehicle travel. Alternatively, an inclined plane obtained by rotating the cross-section by a certain angle about the common edge between the horizontal plane (also called the ground) and the cross-section can be selected. This splitting plane can effectively retain the tunnel scene features in the point cloud subset for subsequent analysis. The point cloud subset size can be determined based on the data accuracy and coverage required for identifying motion subpaths in tunnel scenes. For example, a thinner point cloud subset provides higher accuracy, while a thicker point cloud subset covers a wider area. The point cloud subset interval can be determined based on the processing efficiency required for identifying motion subpaths in tunnel scenes. For example, selecting a smaller interval can increase the number and coverage of point cloud subsets, while a larger interval can reduce the number of point cloud subsets and improve processing efficiency.

[0102] The point cloud subset coordinate condition refers to the condition used to determine which spatial points are included in each point cloud subset when performing point cloud segmentation of point cloud data.

[0103] Specifically, after determining the segmentation plane, point cloud subset size and point cloud subset interval of the point cloud data, the determined segmentation plane, point cloud subset size and point cloud subset interval and the point cloud data can be input into the point cloud data processing application. The segmentation plane, point cloud subset size and point cloud subset interval are analyzed by the point cloud data processing application to determine the coordinate conditions of each point cloud subset, and the spatial points in the point cloud data that meet the same point cloud subset coordinate condition are divided into the same point cloud subset to obtain a preset number of point cloud subsets.

[0104] In the above embodiment, the computer device determines the segmentation plane, point cloud subset size and point cloud subset interval of the point cloud data; based on the segmentation plane, point cloud subset size and point cloud subset interval, determines the point cloud subset coordinate condition; thereby, the spatial points in the point cloud data that meet the same point cloud subset coordinate condition can be divided into the same point cloud subset, and a preset number of point cloud subsets are obtained, and large point cloud data are divided into smaller fragments, so that the analysis and processing of specific areas or structures becomes easier to manage, so that the distribution characteristics of the point cloud data can be accurately and quickly determined, so that when the motion sub-path in the tunnel scene is subsequently determined based on the distribution characteristics of the point cloud data, the recognition accuracy and efficiency of the motion sub-path in the tunnel scene can be improved.

[0105] In one embodiment, the process of determining the segmentation plane, point cloud subset size and point cloud subset interval by a computer device includes: obtaining scanning parameters of a multi-line laser radar used to collect point cloud data; determining the segmentation plane based on the central scanning line direction and the scanning direction; determining the point cloud subset size based on the number of laser lines and the field of view angle; and determining the point cloud subset interval based on the vehicle driving speed and the scanning frequency.

[0106] It should be noted that the laser radar used in the embodiment of the present application can specifically be a multi-line laser radar, such as a 64-line laser radar, and the scanning parameters include the vehicle speed and the number of laser lines, field of view angle, central scanning line direction, scanning direction, and scanning frequency of the multi-line laser radar.

[0107] It is understandable that the central scanning light of the laser radar installed on the vehicle usually has an inclination angle with the horizontal plane. During the scanning process, when there is an obstacle in the light path of the scanning light, a hole will be formed behind the obstacle along the light path (as shown in Figure 7). If the segmentation plane is specifically a plane parallel to the cross section of the point cloud data, that is, the section is segmented in the vertical direction, the obtained point cloud subset cannot fully reflect the environmental characteristics, thereby affecting the accuracy of the subsequent tunnel scene recognition results; if the segmentation plane is a plane that separates the cross section from the horizontal plane (also called the ground) and the horizontal plane (also called the ground), the point cloud subset will not be completely reflected. When the common edge between the cross sections is rotated by a certain angle to obtain an inclined plane, and the angle is equal to the inclination angle between the central scanning light and the horizontal plane, that is, the segmentation plane is parallel to the scanning plane determined by the central scanning light during the lidar scanning process, then the obtained point cloud subset can fully reflect the specific environment, thereby improving the accuracy of subsequent tunnel scene recognition results; therefore, according to the data analysis purpose of identifying the motion sub-path of the tunnel scene, the scanning plane can be determined in combination with the central scanning line direction and scanning direction of the multi-line lidar, and the section plane is determined to be a plane parallel to the scanning plane.

[0108] The number of laser lines refers to the number of laser beams emitted simultaneously by a multi-line lidar. The higher the number of laser lines, the better the vertical resolution of the radar, and the more detailed vertical details it can capture. The field of view refers to the vertical range covered by the radar scan. A wide field of view means that the radar can capture data within a wider vertical range. The size of the point cloud subset can be determined based on the data accuracy and coverage required for identifying the motion sub-paths of the tunnel scene, combined with the number of laser lines and the field of view.

[0109] The vehicle speed and scanning frequency determine the scanning range of each frame of point cloud data. The number of preset point cloud subsets is determined based on the processing efficiency required for identifying the motion sub-paths of the tunnel scene. The point cloud subset interval is determined based on the preset number of point cloud subsets and the scanning range.

[0110] In the above embodiment, the computer device obtains the scanning parameters of the multi-line laser radar used to collect point cloud data, so that the appropriate segmentation plane, point cloud subset size and point cloud subset interval can be determined based on the scanning parameters of the multi-line laser radar, so that when the point cloud data is segmented based on the segmentation plane, point cloud subset size and point cloud subset interval, the important spatial features of the point cloud data can be captured, so that the distribution characteristics of the point cloud data can be accurately and quickly determined, so that when the motion sub-path in the tunnel scene is subsequently determined based on the distribution characteristics of the point cloud data, the recognition accuracy and efficiency of the motion sub-path in the tunnel scene can be improved.

[0111] In one embodiment, a computer device determines the closed distribution characteristics of spatial points in a point cloud subset based on a first distance between each spatial point in the ray cluster and the ray origin; determines the center point of the ray cluster based on the spatial points in the ray cluster, and determines the continuity distribution characteristics of spatial points in the point cloud subset based on a second distance between the center point and the ray origin. The process includes the following steps: determining the closed distribution characteristics of spatial points in the point cloud subset based on the first distance between each spatial point in the ray cluster and the ray origin; determining the center point of the ray cluster based on the spatial points in the ray cluster, and determining the continuity distribution characteristics of spatial points in the point cloud subset based on the second distance between the center point and the ray origin.

[0112] Specifically, for each spatial point in any ray cluster, its first distance to the ray origin is determined respectively, and the consistency distribution of the spatial points in the ray cluster is determined based on the first distance of each spatial point to the ray origin, and the closed distribution characteristics of the spatial points in the point cloud subset are determined based on the consistency distribution of the spatial points in each ray cluster in the point cloud subset; for any ray cluster, the second distance from its target junction point to the ray origin is determined, and the continuity distribution characteristics of the spatial points in the point cloud subset are determined based on the relationship between the second distances of adjacent ray clusters in the point cloud subset.

[0113] In the above embodiment, the computer device determines the closed distribution characteristics and continuity distribution characteristics of the spatial points in the point cloud subset by calculating the distance from the spatial points in the ray cluster to the ray origin, so as to perform a deeper and more accurate analysis of the point cloud data, thereby improving the recognition accuracy of the motion sub-path in the tunnel scene when subsequently determining the motion sub-path in the tunnel scene based on the closed distribution characteristics and continuity distribution characteristics of the spatial points in the point cloud subset.

[0114] In one embodiment, a process in which a computer device determines the closed distribution characteristics of spatial points in a point cloud subset based on a first distance between each spatial point in the ray cluster and the ray origin includes the following steps: determining a first distance from each spatial point to the ray origin; determining the ray cluster as a candidate ray cluster when the difference between the maximum value and the minimum value in the first distance meets a first distance difference condition; and determining the closed distribution characteristics of spatial points in the point cloud subset based on the proportion of candidate ray clusters in each ray cluster in the point cloud subset.

[0115] The first distance difference condition determines whether the space points corresponding to the rays in the ray cluster can form a closed structure. Specifically, the first distance difference threshold may be greater than or equal to a preset first distance difference threshold.

[0116] Specifically, for any ray cluster, the first distance from the spatial point corresponding to each ray in the ray cluster to the ray origin is determined, and the maximum and minimum values ​​of the first distances are selected from the obtained first distances, the difference between the maximum and minimum values ​​is determined, and the difference is compared with a preset first distance difference threshold. When the difference is less than the preset first distance difference threshold, the ray cluster is determined to be a candidate ray cluster, so as to obtain whether each ray cluster in the point cloud subset is a candidate ray cluster, and the number of candidate ray clusters in each ray cluster of the point cloud subset is counted, and the ratio of the number of candidate ray clusters to all ray clusters in the point cloud subset is determined. When the ratio is greater than or equal to the first preset ratio, the closure distribution feature of the spatial points in the point cloud subset is determined to be closed. When the ratio is less than the first preset ratio, the closure distribution feature of the spatial points in the point cloud subset is determined to be open.

[0117] In one embodiment, the maximum value of the first distance of the ray cluster is max(d ij ), the minimum value of the first distance min(d ij ) and the first distance difference threshold ∈ satisfy the following relationship, the ray cluster is determined to be a candidate ray cluster. In the embodiment of the present application, ∈ can be taken as 1m: |max(d ij )-min(d ij )|<ε

[0118] In one embodiment, the number of candidate ray clusters N valid , the number of ray clusters N and the first preset ratio δ n When the relationship between satisfies the following relationship, it is determined that the closed distribution feature of the spatial points in the point cloud subset is closed. In the embodiment of the present application, δ n The value can be 0.833:

[0119] It should be noted that, when there is no ray corresponding to a spatial point in a certain ray cluster, the ray cluster can be directly determined to be a non-candidate ray cluster.

[0120] In the above embodiment, the computer device determines a first distance from each spatial point to the ray origin; when the difference between the maximum value and the minimum value in the first distance meets the first distance difference condition, the ray cluster is determined to be a candidate ray cluster; based on the proportion of candidate ray clusters in each ray cluster of the point cloud subset, the closed distribution characteristics of the spatial points in the point cloud subset are determined, so that it can more accurately judge whether the point cloud subset forms a closed structure, and thus when the motion sub-path in the tunnel scene is subsequently determined based on the closed distribution characteristics and continuity distribution characteristics of the spatial points in the point cloud subset, the recognition accuracy of the motion sub-path in the tunnel scene can be improved.

[0121] In one embodiment, a computer device determines the center point of a ray cluster based on the spatial points in the ray cluster, and determines the continuity distribution characteristics of the spatial points in the point cloud subset according to the second distance between the center point and the ray origin, including the following steps: when the closed distribution characteristic characterizes that the point cloud subset is in a closed state, determining the second distance from the center point of each candidate ray cluster in the point cloud subset to the ray origin; when the difference between the second distance of the candidate ray cluster and the second distance of the adjacent candidate ray cluster meets the second distance difference condition, determining that the candidate ray cluster is a normal ray cluster; based on the proportion of normal ray clusters in each candidate ray cluster in the point cloud subset, determining the continuity distribution characteristics of the spatial points in the point cloud subset.

[0122] The second distance difference condition determines whether the spatial points corresponding to the rays in the ray cluster are possibly distributed continuously. Specifically, the second distance difference threshold may be greater than or equal to a preset second distance difference threshold.

[0123] Specifically, for any candidate ray cluster, the center point of the candidate ray cluster is determined, and the second distance from the center point to the ray origin is determined. The second distance of the candidate ray cluster is compared with the second distances of the left and right adjacent candidate ray clusters respectively. When the difference between its second distance and the second distance of the left adjacent candidate ray cluster is less than a preset second distance difference threshold, and the difference between its second distance and the second distance of the right adjacent candidate ray cluster is less than the preset second distance difference threshold, the candidate ray cluster is determined to be a normal ray cluster, thereby obtaining whether each candidate ray cluster in the point cloud subset is a normal ray cluster, and counting the number of normal ray clusters in each candidate ray cluster in the point cloud subset, and determining the ratio of the number of normal ray clusters to the number of candidate ray clusters in the point cloud subset. When the ratio is greater than or equal to the second preset ratio, it is determined that the continuity distribution feature of the spatial points in the point cloud subset is continuous. When the ratio is less than the second preset ratio, it is determined that the continuity distribution feature of the spatial points in the point cloud subset is discontinuous.

[0124] In one embodiment, the second distance of any candidate ray cluster The second distance of the left adjacent candidate ray cluster The second distance of the right adjacent candidate ray cluster and the second distance difference threshold ε n When the relationship between ε and ε satisfies the following relationship, the candidate ray cluster is determined to be a normal ray cluster. In the embodiment of the present application, n The value can be 5m:

[0125] and

[0126] In the above embodiment, when the closure distribution feature indicates that the point cloud subset is in a closed state, the computer device determines the second distance from the center point of each candidate ray cluster in the point cloud subset to the ray origin; when the difference between the second distance of the candidate ray cluster and the second distance of the adjacent candidate ray cluster satisfies the second distance difference condition, the candidate ray cluster is determined to be a normal ray cluster; based on the proportion of normal ray clusters in each candidate ray cluster in the point cloud subset, the continuity distribution feature of the spatial points in the point cloud subset can be more accurately judged, so that when the motion sub-path in the tunnel scene is subsequently determined based on the closure distribution feature and the continuity distribution feature of the spatial points in the point cloud subset, the recognition accuracy of the motion sub-path in the tunnel scene can be improved.

[0127] In one embodiment, a computer device determines the closed distribution characteristics of spatial points in a point cloud subset based on a first distance between each spatial point in the ray cluster and the ray origin; determines the center point of the ray cluster based on the spatial points in the ray cluster, and determines the continuity distribution characteristics of the spatial points in the point cloud subset based on a second distance between the center point and the ray origin. The process includes the following steps: determining a first distance from each spatial point to the ray origin; when the difference between the maximum value and the minimum value in the first distance meets a first distance difference condition, determining the ray cluster to be a first candidate type; determining a second distance from the center point of each ray cluster in the point cloud subset to the ray origin; when the difference between the second distance of the ray cluster and the second distance of the adjacent ray cluster meets a second distance difference condition, determining the ray cluster to be a second candidate type; and determining the closed distribution characteristics and continuity distribution characteristics of the spatial points in the point cloud subset based on the proportion of ray clusters of both the first candidate type and the second candidate type in the point cloud subset.

[0128] Specifically, for any point cloud subset, the computer device determines that the closure distribution feature of the spatial points in the point cloud subset is closed, and determines that the continuity distribution feature of the spatial points in the point cloud subset is continuous when the ratio of the number of ray clusters in the point cloud subset that are both the first candidate type and the second candidate type to the total number of ray clusters in the point cloud subset is greater than a preset threshold.

[0129] In one embodiment, a process in which a computer device identifies a motion sub-path in a tunnel scenario in a motion path based on the closed distribution characteristics and continuity distribution characteristics of spatial points of each of a plurality of point cloud subsets includes the following steps: determining the scene type of the sampling path points based on the closed distribution characteristics and continuity distribution characteristics of the spatial points in the point cloud subsets; and determining the motion sub-path in the motion path to which at least two consecutive sampling path points belong, whose scene type is a target type, as a motion sub-path in a tunnel scenario.

[0130] The scene type may specifically be a target type or a non-target type. The target type may be a tunnel scene type, and the non-target type may be other types other than the tunnel scene.

[0131] Specifically, for any sampling path point in the motion path, after obtaining the closed distribution characteristics and continuity distribution characteristics of the spatial points in each point cloud subset of the sampling path point, the corresponding scene type of the sampling path point can be determined according to the closed distribution characteristics and continuity distribution characteristics of the spatial points in each point cloud subset, and the sampling path points of the target type can be marked in the motion path to obtain the marked motion path, and the path segment to which the continuous sampling path points in the marked motion path belong and whose scene type is the target type belongs can be determined as the motion sub-path under the tunnel scene.

[0132] In the above embodiment, the computer device determines the scene type by analyzing the closed distribution characteristics and continuous distribution characteristics of the spatial points in the point cloud subset, and can more accurately identify specific scenes, such as tunnels. The motion sub-path in the motion path whose scene type is the target type and to which at least two consecutive sampling path points belong is determined as the motion sub-path in the tunnel scene. By combining the continuous sampling path points, the entire tunnel section can be more accurately identified, rather than just a part of the tunnel, thereby improving the recognition accuracy of the motion sub-path in the tunnel scene.

[0133] In one embodiment, a process for a computer device to identify a motion sub-path in a tunnel scenario in a motion path based on the closed distribution characteristics and continuous distribution characteristics of the spatial points of each of a plurality of point cloud subsets includes the following steps: determining the number of sampling path points of non-target scene types between the motion sub-paths in two adjacent tunnel scenarios; when the number of sampling path points of non-target scene types is less than a preset number threshold, or the distance between the motion sub-paths in two adjacent tunnel scenarios is less than a preset distance threshold, determining the sub-path between the motion sub-paths in the two adjacent tunnel scenarios as the corrected motion sub-path in the tunnel scenario.

[0134] For example, in a mountain tunnel scenario, the non-open-air tunnel section of the tunnel can be identified first, and the distance between two adjacent non-open-air tunnel sections can be determined. When the distance is less than 200 meters, the section between the two adjacent non-open-air tunnel sections is an open-air section, and it is merged with the two adjacent non-open-air tunnel sections to obtain a complete tunnel section.

[0135] In one embodiment, a process in which a computer device determines the scene type of a sampling path point based on the closed distribution characteristics and continuity distribution characteristics of the spatial points in the point cloud subset includes the following steps: when the closed distribution characteristics of the spatial points in the point cloud subset characterize that the point cloud subset is in a closed state, and the continuity distribution characteristics characterize that the point cloud subset is continuous, the point cloud subset is determined to be a suspected tunnel point cloud subset; when the proportion of the suspected tunnel point cloud subset in the point cloud subset corresponding to the sampling path point reaches a proportion threshold, the scene type corresponding to the sampling path point is determined to be a target type.

[0136] Specifically, for any point cloud subset, after the computer device obtains the closure distribution characteristics and continuity distribution characteristics of the spatial points in the point cloud subset, if the closure distribution characteristic is closed and the continuity distribution characteristic is continuous, that is, the closure distribution characteristic characterizes that the point cloud subset is in a closed state, and the continuity distribution characteristic characterizes that the point cloud subset is continuous, the point cloud subset may be a tunnel or other similar closed structure, then the point cloud subset is determined to be a suspected tunnel point cloud subset, and the proportion of the suspected tunnel point cloud subset in all point cloud subsets corresponding to each sampling path point is counted. When the proportion reaches the proportion threshold, the scene type corresponding to the corresponding sampling path point is determined to be the target type, that is, the scene type corresponding to the corresponding sampling path point is determined to be the tunnel scene type.

[0137] For example, if the ratio threshold is 1 / 2, there are 5 point cloud subsets corresponding to a certain sampling path point, of which the number of suspected tunnel point cloud subsets is 3, then the scene type of the sampling path point is determined to be a tunnel scene type.

[0138] In the above embodiment, the computer device can more accurately determine whether the point cloud subset represents a tunnel scene by evaluating the closure and continuity of the point cloud subset, which helps to distinguish tunnels from other similar structures, such as bridges or culverts. When the proportion of suspected tunnel point cloud subsets in the point cloud subset corresponding to the sampling path point reaches a proportion threshold, the scene type corresponding to the sampling path point is determined to be the target type, thereby avoiding misjudgment and ensuring that the scene type is determined as a tunnel type only when there is sufficient evidence. Therefore, when the motion sub-path in the tunnel scene is subsequently determined based on the sampling path point of the tunnel type, the recognition accuracy of the motion sub-path in the tunnel scene can be improved.

[0139] In one embodiment, the above-mentioned tunnel scene recognition method also includes the following steps: during the point cloud map update process, when the motion sub-path corresponding to the point cloud data to be processed is the motion sub-path under the tunnel scene, the point cloud data to be processed is aligned with the point cloud map to obtain an updated point cloud map.

[0140] Among them, the point cloud map refers to an existing point cloud map, which is a point cloud map obtained by updating frame by frame based on the processed point cloud data in the point cloud dataset. For example, the point cloud dataset corresponding to the motion path contains 20 frames of point cloud data, and the processing of the first 9 frames of point cloud data has been completed. Then, the point cloud map is obtained based on the first 9 frames of point cloud data.

[0141] The point cloud data to be processed refers to the point cloud data that needs to be processed and analyzed, which can be the point cloud data of the current frame. For example, the point cloud data set corresponding to the motion path contains 20 frames of point cloud data. The processing of the first 9 frames of point cloud data has been completed, and the point cloud data of the 10th frame is the point cloud data to be processed.

[0142] Specifically, after determining the motion sub-path under the tunnel scene in the motion path, the computer device can determine that in the original point cloud data set, each frame of point cloud data corresponding to the motion sub-path under the tunnel scene is the point cloud data under the tunnel scene. During the point cloud map update process, when the motion sub-path corresponding to the point cloud data to be processed is the motion sub-path under the tunnel scene, that is, when the point cloud data to be processed is the point cloud data under the tunnel scene, the preset alignment algorithm applicable to the tunnel scene is used to align the point cloud data to be processed with the point cloud map to obtain an updated point cloud map.

[0143] Among them, the preset registration algorithm suitable for tunnel scenarios is an algorithm specially designed for processing point cloud data in tunnel environments. Specifically, it can be an algorithm obtained after adjusting the parameters of the iterative closest point (ICP) algorithm, feature reference registration, standard template matching, etc. to adapt to tunnel scenarios.

[0144] In the above embodiment, during the point cloud map update process, the computer device performs special point cloud data registration and map update for the motion sub-path in the tunnel scene, which can significantly improve the accuracy and practicality of the point cloud map.

[0145] In one embodiment, as shown in FIG8 , a method for identifying a tunnel scene is provided. The method is described by applying the method to the computer device in FIG1 as an example, and includes the following steps:

[0146] S802 , extracting at least two frames of original point cloud data corresponding to sampling path points in the motion path from a point cloud data set corresponding to the motion path position of the vehicle; the sampling path points are obtained by sampling the point positions of the motion path.

[0147] S804: Register and stitch at least two frames of original point cloud data to obtain stitched point cloud data.

[0148] S806: Select point cloud data that meets the height condition from the stitched point cloud data.

[0149] S808: performing point cloud segmentation on the cloud data to obtain a point cloud subset of the point cloud data.

[0150] S810 , in the point cloud subset, using a fixed reference point as a ray origin and determining a reference coordinate axis; and emitting a ray from the ray origin toward each spatial point in the point cloud subset.

[0151] S812, determining the angle between each ray and the reference coordinate axis; obtaining the ray cluster angle value corresponding to each ray cluster.

[0152] S814 , when the angle difference between the included angle and the target ray cluster angle value in the ray cluster angle value satisfies the angle difference threshold, dividing the ray corresponding to the included angle into the ray cluster corresponding to the target ray cluster angle value.

[0153] S816 , determining a first distance from each spatial point to the ray origin; when the difference between the maximum value and the minimum value in the first distance meets a first distance difference condition, determining the ray cluster as a candidate ray cluster.

[0154] S818 , determining the closed distribution characteristics of the spatial points in the point cloud subset based on the proportion of the candidate ray clusters in each ray cluster of the point cloud subset.

[0155] S820, when the closed distribution feature indicates that the point cloud subset is in a closed state, determining the second distance from the center point of each candidate ray cluster in the point cloud subset to the ray origin; when the difference between the second distance of the candidate ray cluster and the second distance of the adjacent candidate ray cluster meets the second distance difference condition, determining that the candidate ray cluster is a normal ray cluster.

[0156] S822: Determine the continuity distribution characteristics of the spatial points in the point cloud subset based on the proportion of normal ray clusters in each candidate ray cluster in the point cloud subset.

[0157] S824 , when the closed distribution feature of the spatial points in the point cloud subset indicates that the point cloud subset is closed, and the continuity distribution feature indicates that the point cloud subset is continuous, determine that the point cloud subset is a suspected tunnel point cloud subset.

[0158] S826: When the proportion of the suspected tunnel point cloud subset in the point cloud subset corresponding to the sampling path point reaches a proportion threshold, determine that the scene type corresponding to the sampling path point is a target type.

[0159] S828: Determine a motion sub-path in which the scene type of the motion path is the target type and to which at least two consecutive sampling path points belong as a motion sub-path in the tunnel scene.

[0160] The present application also provides an application scenario, which applies the above-mentioned tunnel scene identification method. The tunnel scene identification method specifically includes the following steps:

[0161] 1. Point sampling and local point cloud stitching

[0162] The collection vehicle collects a large amount of GPS / IMU information and laser point cloud data at various households. The complete motion path during collection is determined based on the GPS / IMU information, and the collected laser point cloud data is organized according to the complete motion path to obtain the original point cloud dataset. For the complete motion path, data preprocessing is first performed by equidistant sampling (such as 2m intervals) to obtain the motion path, and a frame of original point cloud data collected at each path point on the motion path is extracted from the original point cloud dataset. A point cloud dataset is formed based on the frame of original point cloud data collected at each path point. Then, the motion path is sampled at a spacing of 30m to obtain each sampling path point, and a frame of original point cloud data collected at the sampling path point and the two frames of original point cloud data adjacent to the frame of original point cloud data are extracted from the point cloud dataset to obtain three frames of point cloud data corresponding to the sampling path point, and these three frames of point cloud data are spliced ​​to obtain spliced ​​point cloud data.

[0163] 2. Point cloud ray cluster construction

[0164] The spliced ​​point cloud data corresponding to each sampling path point is segmented to obtain each point cloud subset.

[0165] Among them, this scheme starts from the radar scan itself, and the direction of segmenting the point cloud subset and the direction of subsequent ray cluster construction are all along the direction parallel to the central scanning line of the multi-line lidar. To achieve this goal, it is necessary to stitch the point cloud in the world system Back-projected to the radar center corresponding to the sampling path point i Department:

[0166] Since the z-axis of the laser radar coordinate system is perpendicular to the scan line and the origin is located at the starting point of the scan line, it is only necessary to keep the z value near 0 (i.e. ), which are obtained by scanning the central scan line. Considering factors such as the scanning range and resolution of the lidar point cloud, taking a 64-line lidar as an example, its field of view angle range is -25° to 15°, with an average resolution of 0.625°. When the laser scans an object at a distance of 10m, the scanning line spacing is approximately 10cm. In practice, due to the uneven distribution of laser scan lines, the resolution increases toward the middle area. This scheme also mainly samples the middle scanning area. Therefore, the point cloud subset size is set to 10cm (i.e., δ = 0.05m). To ensure coverage in the front-to-back direction, each spliced ​​point cloud is segmented five times, and the point cloud subset interval is set to 0.5m. To avoid interference with subsequent tunnel recognition from surrounding traffic, guardrails, pedestrians, etc., only point cloud subsets above a certain height (for example, 1m above the vehicle body and approximately 2.5m above the ground) are retained.

[0167] The bottom midpoint of each point cloud subset is used as the ray origin O, and rays are drawn from the origin to the spatial points on the point cloud subset. After the rays are drawn from the closed point cloud, a semicircle should be formed. In order to reduce the amount of calculation, specific angles are sampled at each interval as candidate angles, and only rays of the corresponding angles are constructed.

[0168] Preset the ray cluster angle value corresponding to each ray cluster, determine the angle between each ray and the reference coordinate axis, and when the angle difference between the angle and the target ray cluster angle value in the ray cluster angle value meets the angle difference threshold, divide the ray corresponding to the angle into the ray cluster corresponding to the target ray cluster angle value. For example, construct 18 ray clusters, where the ray cluster angle values ​​corresponding to each ray cluster are

[0169] 3. Tunnel identification

[0170] For any spatial point corresponding to a ray cluster, determine the first distance from each spatial point to the origin. When the difference between the maximum and minimum values ​​in the first distance meets the first distance difference condition, the ray cluster is determined to be a valid ray cluster. For any valid ray cluster S i , take the effective ray cluster S i Corresponding space point p ij The center of mass And calculate the second distance between the centroid and the origin.

[0171] (1) Closure: For any point cloud subset, if the point cloud cross section is closed, then the number of valid ray clusters N in the ray clusters of the point cloud subset is valid Should be equal to or close to (taking into account the influence of noise) the total number of ray clusters N. If the number is large, as shown in the following relationship: |NN valid |>δ n

[0172] In this embodiment, δ n =3, then it can be determined that the point cloud subset is a non-tunnel type point cloud subset.

[0173] (2) Continuity: For any point cloud subset, since the inner surface of the tunnel is relatively regular and smooth, the distances from the surface points of adjacent areas in the same cross section to the geometric center of the cross section are relatively consistent, and the second distance from the adjacent ray clusters in the ray cluster to the origin is relatively consistent. and as well as Should be relatively close. If the deviation is too large, as shown in the following relationship:

[0174] and

[0175] In this embodiment, ε n =5m, it is considered abnormal. When the number of abnormally long ray clusters in the valid ray clusters of a point cloud subset exceeds a certain threshold, for example, greater than 3, the point cloud subset is determined to be a non-tunnel type point cloud subset.

[0176] (3) Consistency: To further improve the accuracy of tunnel scene recognition and reduce the interference caused by road elements such as gantries and dense roadside trees, this solution utilizes the consistency of tunnels and combines the tunnel recognition results of all point cloud subsets at each sampling path point to vote on whether the scene type of the sampling path point is a tunnel scene. The voting strategy adopted is that when the number of point cloud subsets judged as suspected tunnels at each sampling path point is less than the preset number, the scene type of the sampling path point is determined to be a non-tunnel scene type.

[0177] (4) Noise filtering and leak filling: Considering that a single sampling path point only contains local information, some scenes such as culverts and overpasses are basically the same as tunnels in local features, but from a global perspective, they should be distinguished from tunnels due to their shorter length. Based on this, this application scenario judges the tunnel recognition results of continuous sampling path points. Only when the sampling path point is continuously identified as a tunnel scene type, it is considered to be a path point in a valid tunnel scene. Specifically, if a sampling path point is a tunnel scene type, and the two adjacent sampling path points before and after are both non-tunnel scene types, then the sampling path point can be filtered out as noise. After filtering out the noise, for scenes such as mountain tunnels where there are some open-air non-tunnels between tunnels, the open-air section and the front and back tunnel sections should be merged into a complete tunnel, and the merging interval threshold is set to 200m. Under the action of the merging strategy, if some tunnel sampling path points are misjudged as non-tunnels and missed calls, the leaks can still be filled by the fact that the front and back sampling path points are both tunnels.

[0178] The following illustrates the recognition effect of the tunnel scene recognition method provided in this application through several real scenarios. Referring to scene 1 shown in Figure 9, in the process of identifying tunnel scenes using the tunnel scene recognition method provided in this application, the number of valid ray clusters in the ray clusters corresponding to sampling path point ① and sampling path point ③ is much smaller than the total number of corresponding ray clusters, which does not meet the closure requirement. Therefore, the scene types corresponding to sampling path point ① and sampling path point ③ are both non-tunnel scenes; sampling path point ② meets the closure and continuity requirements, and the scene type corresponding to sampling path point ② is a tunnel scene, but the adjacent sampling path point ① and sampling path point ③ of sampling path point ② are both non-tunnel scenes. Therefore, scene 1 is determined to be a non-tunnel scene, and the corresponding motion sub-path is a motion sub-path under a non-tunnel scene.

[0179] Referring to scene 2 shown in Figure 10, in the process of identifying tunnel scenes using the tunnel scene identification method provided in this application, the number of valid ray clusters in the ray clusters corresponding to each sampling path point is close to the corresponding total number of ray clusters, which satisfies the closure property; however, the distances from the center of mass of adjacent ray clusters to the ray origin in the same point cloud subset differ greatly, and there are many ray clusters in this situation, which does not satisfy the continuity property. Therefore, scene 2 is determined to be a non-tunnel scene, and the corresponding motion sub-path is the motion sub-path under the non-tunnel scene.

[0180] Referring to Scene 3 shown in FIG11 , during tunnel scene recognition using the tunnel scene recognition method provided in this application, the ray clusters in the point cloud subset at each of the three sampling path points all meet the requirements of closure, continuity, and consistency, confirming that the scene type of each sampling path point is a tunnel scene. Furthermore, the three sampling path points are continuous, thus determining that Scene 3 is a tunnel scene, and the corresponding motion subpath is a motion subpath in a tunnel scene. Furthermore, sampling path point ② presents a complex scene of intersecting tunnels, which can still be accurately identified using the tunnel scene recognition method provided in this application.

[0181] In another real scenario, the tunnel scene recognition method provided in this application is used to recognize the tunnel scene, thereby obtaining the scene schematic diagram obtained after recognizing a certain motion path as shown in Figure 12. The part on the left of the figure covered by a large-scale point cloud is a non-tunnel area, and the white rectangle indicates that the scene type of the sampling path point is a non-tunnel scene. The part on the right covered by a small-scale point cloud is a non-tunnel area, and the black umbrella indicates that the scene type of the sampling path point is a tunnel scene. It can be seen from the figure that the tunnel scene recognition method provided in this application can effectively identify the tunnel area in the road and output the complete tunnel range.

[0182] It should be understood that, although the steps in the flowcharts of the above embodiments are shown in sequence as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be performed in other orders. Moreover, at least a portion of the steps in the flowcharts of the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily performed at the same time, but can be performed at different times. The execution order of these steps or stages is not necessarily to be performed in sequence, but can be performed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0183] Based on the same inventive concept, embodiments of the present application also provide a tunnel scene recognition device for implementing the tunnel scene recognition method described above. The solution provided by this device is similar to the solution described in the method described above. Therefore, the specific limitations of the one or more tunnel scene recognition device embodiments provided below can be found in the limitations of the tunnel scene recognition method described above and will not be repeated here.

[0184] In one embodiment, as shown in FIG13 , a tunnel scene recognition device is provided, comprising: a point cloud data acquisition module 1302 , a point cloud data segmentation module 1304 , a ray cluster determination module 1306 , a distribution feature determination module 1308 , and a scene recognition module 1310 , wherein:

[0185] The point cloud data acquisition module 1302 is used to acquire point cloud data corresponding to the vehicle's motion path position.

[0186] The point cloud data segmentation module 1304 is used to segment the spatial points in the point cloud data according to spatial positions to obtain multiple point cloud subsets.

[0187] The ray cluster determination module 1306 is configured to emit a ray from the point cloud subset toward each spatial point in the point cloud subset using a fixed reference point as a ray origin, and determine the ray cluster to which the ray belongs.

[0188] The distribution feature determination module 1308 is used to determine the closed distribution features of the spatial points in the point cloud subset based on the first distance between each spatial point in the ray cluster and the ray origin; determine the center point of the ray cluster based on the spatial points in the ray cluster, and determine the continuity distribution features of the spatial points in the point cloud subset based on the second distance between the center point and the ray origin.

[0189] The scene recognition module 1310 is configured to recognize a motion sub-path in a tunnel scene in the motion path based on the closed distribution characteristics and continuous distribution characteristics of the spatial points of each of the plurality of point cloud subsets.

[0190] In the above embodiment, point cloud segmentation is performed on the point cloud data corresponding to the vehicle position on the motion path, and the distribution of each point cloud subset is analyzed. Without analyzing the entire point cloud data, the closed distribution characteristics and continuity distribution characteristics of the spatial points in multiple point cloud subsets that can reflect the distribution characteristics of the point cloud data can be obtained. Then, based on the closed distribution characteristics and continuity distribution characteristics of the spatial points in the multiple point cloud subsets, the motion sub-path belonging to the tunnel scene in the motion path can be identified, thereby improving the efficiency of identifying motion sub-paths in the tunnel scene. By performing point cloud segmentation on the point cloud data and determining the ray cluster to which each spatial point in the obtained point cloud subset belongs, the closed distribution characteristics and continuity distribution characteristics of the spatial points in the point cloud subset can be accurately and quickly determined based on the first distance from each spatial point in the ray cluster to the ray origin and the second distance from the center point of the ray cluster to the ray origin. Subsequently, based on the closed distribution characteristics and continuity distribution characteristics of the spatial points in each point cloud subset, the motion sub-path in the tunnel scene in the motion path can be determined, further improving the accuracy and efficiency of identifying motion sub-paths in the tunnel scene.

[0191] In one embodiment, the point cloud data acquisition module 1302 is further used to: extract at least two frames of original point cloud data corresponding to the sampling path points in the motion path from the point cloud data set corresponding to the motion path position of the vehicle; align and splice the at least two frames of original point cloud data to obtain spliced ​​point cloud data; and select point cloud data that meets the height condition from the spliced ​​point cloud data.

[0192] In one embodiment, the point cloud data segmentation module 1304 is also used to: determine the segmentation plane, point cloud subset size and point cloud subset interval of the point cloud data; determine the point cloud subset coordinate conditions based on the segmentation plane, point cloud subset size and point cloud subset interval; divide the spatial points in the point cloud data that meet the same point cloud subset coordinate conditions into the same point cloud subset to obtain a preset number of point cloud subsets.

[0193] In one embodiment, the point cloud data segmentation module 1304 is further used to: obtain scanning parameters of a multi-line laser radar used to collect point cloud data; the scanning parameters include the vehicle speed and the number of laser lines, field of view angle, central scanning line direction, scanning direction, and scanning frequency of the multi-line laser radar; determine the segmentation plane based on the central scanning line direction and the scanning direction; determine the point cloud subset size based on the number of laser lines and the field of view angle; and determine the point cloud subset interval based on the vehicle speed and the scanning frequency.

[0194] In one embodiment, the ray cluster determination module 1306 is further used to: determine the ray origin and reference coordinate axis corresponding to the point cloud subset; generate rays corresponding to each spatial point with the ray origin as the starting point and each spatial point in the point cloud subset as the end point; determine the angle between each ray and the reference coordinate axis; and determine the ray cluster to which each ray belongs based on the angle.

[0195] In one embodiment, the ray cluster determination module 1306 is further configured to: obtain a ray cluster angle value corresponding to each ray cluster; and when the angle difference between the included angle and a target ray cluster angle value in the ray cluster angle value satisfies an angle difference threshold, divide the ray corresponding to the included angle into the ray cluster corresponding to the target ray cluster angle value.

[0196] In one embodiment, the distribution feature determination module 1308 is further used to: determine a first distance from each spatial point to the ray origin; when the difference between the maximum value and the minimum value in the first distance meets the first distance difference condition, determine the ray cluster as a candidate ray cluster; based on the proportion of candidate ray clusters in each ray cluster in the point cloud subset, determine the closed distribution characteristics of the spatial points in the point cloud subset.

[0197] In one embodiment, the distribution feature determination module 1308 is further used to: determine the second distance from the center point of each candidate ray cluster in the point cloud subset to the ray origin when the closed distribution feature indicates that the point cloud subset is in a closed state; determine that the candidate ray cluster is a normal ray cluster when the difference between the second distance of the candidate ray cluster and the second distance of the adjacent candidate ray cluster meets the second distance difference condition; and determine the continuity distribution feature of the spatial points in the point cloud subset based on the proportion of normal ray clusters in each candidate ray cluster in the point cloud subset.

[0198] In one embodiment, the scene recognition module 1310 is used to: determine the scene type of the sampling path points based on the closed distribution characteristics and continuous distribution characteristics of the spatial points in the point cloud subset; and determine the motion sub-path in the motion path to which the scene type is a tunnel scene type and to which at least two consecutive sampling path points belong as a motion sub-path under the tunnel scene.

[0199] In one embodiment, the scene recognition module 1310 is used to: determine that the point cloud subset is a suspected tunnel point cloud subset when the closed distribution characteristics of the spatial points in the point cloud subset represent that the point cloud subset is in a closed state, and the continuity distribution characteristics represent that the point cloud subset is continuous; and when the proportion of the suspected tunnel point cloud subset in the point cloud subset corresponding to the sampling path point reaches a proportion threshold, determine that the scene type corresponding to the sampling path point is a tunnel scene type.

[0200] In one embodiment, as shown in Figure 14, the device also includes a point cloud map updating module 1312, which is used to: during the point cloud map updating process, when the motion sub-path corresponding to the point cloud data to be processed is a motion sub-path in a tunnel scene, align the point cloud data to be processed with the point cloud map to obtain an updated point cloud map.

[0201] Each module in the tunnel scene recognition device described above can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in hardware form, or stored in a computer device memory in software form, so that the processor can call and execute the corresponding operations of each module.

[0202] In one embodiment, a computer device is provided, which may be a terminal. Its internal structure diagram may be as shown in Figure 15. The computer device includes a processor, memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are connected to the system bus via the input / output interface. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and computer program in the non-volatile storage medium to run. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals via wired or wireless communication. The wireless communication may be achieved via Wi-Fi, a mobile cellular network, NFC (near-field communication), or other technologies. When executed by the processor, the computer program implements a method for identifying tunnel scenarios. The display unit of the computer device is used to form a visually visible image, and can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a path ball or a touchpad set on the computer device casing, or an external keyboard, touchpad or mouse, etc.

[0203] Those skilled in the art will understand that the structure shown in FIG15 is merely a block diagram of a portion of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0204] In one embodiment, a computer device is further provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0205] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0206] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0207] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions.

[0208] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processor involved in the various embodiments provided herein may be, but are not limited to, a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, and the like.

[0209] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0210] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A method for identifying a tunnel scenario, which is executed by a computer device, and the method includes: Obtaining point cloud data corresponding to the movement path position of a vehicle; Dividing the spatial points in the point cloud data according to spatial positions to obtain a plurality of point cloud subsets; In the point cloud subsets, using a fixed reference point as the origin of a ray to emit rays towards each spatial point in the point cloud subsets, and determining the ray clusters to which the rays belong; Determining the closedness distribution characteristics of the spatial points in the point cloud subsets according to the first distances between the spatial points in the ray clusters and the ray origin; Determining the center point of the ray cluster based on the spatial points in the ray cluster, and determining the continuity distribution characteristics of the spatial points in the point cloud subsets according to the second distance between the center point and the ray origin; And Identifying the movement sub-paths in the tunnel scenario in the movement path according to the closedness distribution characteristics and the continuity distribution characteristics of the spatial points of each of the plurality of point cloud subsets.

2. The method according to claim 1, wherein the obtaining of the point cloud data corresponding to the movement path position of the vehicle includes: Extracting at least two frames of original point cloud data corresponding to the sampled path points in the movement path from a point cloud data set corresponding to the movement path position of the vehicle; Performing registration and stitching on the at least two frames of original point cloud data to obtain stitched point cloud data; Selecting the point cloud data that meets the height condition from the stitched point cloud data.

3. The method according to claim 1 or 2, wherein the dividing of the spatial points in the point cloud data according to spatial positions to obtain a plurality of point cloud subsets includes: Determining the segmentation plane, the point cloud subset size, and the point cloud subset interval of the point cloud data; Determining the point cloud subset coordinate conditions based on the segmentation plane, the point cloud subset size, and the point cloud subset interval; Dividing the spatial points in the point cloud data that meet the same point cloud subset coordinate conditions into the same point cloud subset to obtain point cloud subsets with a preset number of point cloud subsets.

4. The method according to claim 3, wherein the determining of the segmentation plane, the point cloud subset size, and the point cloud subset interval of the point cloud data includes: Obtaining the scanning parameters of the multi-line lidar for collecting the point cloud data; the scanning parameters include the vehicle driving speed and the number of laser lines, the field of view angle, the central scanning line direction, the scanning direction, and the scanning frequency of the multi-line lidar; Determining the segmentation plane based on the central scanning line direction and the scanning direction; Determining the point cloud subset size based on the number of laser lines and the field of view angle; Determining the point cloud subset interval based on the vehicle driving speed and the scanning frequency.

5. The method according to any one of claims 1 to 4, wherein in the point cloud subsets, using a fixed reference point as the origin of a ray to emit rays towards each spatial point in the point cloud subsets, and determining the ray clusters to which the rays belong includes: In the point cloud subsets, using a fixed reference point as the origin of a ray, and determining the reference coordinate axes; Emitting rays from the ray origin towards each spatial point in the point cloud subsets; Determining the angles between each of the rays and the reference coordinate axes; Determining the ray clusters to which each of the rays belong based on the angles.

6. The method according to claim 5, wherein the determining the ray clusters to which the rays belong based on the included angle comprises: Obtaining the ray cluster angle values corresponding to each ray cluster; When the angle difference between the included angle and the target ray cluster angle value among the ray cluster angle values satisfies the angle difference threshold, classifying the ray corresponding to the included angle into the ray cluster corresponding to the target ray cluster angle value.

7. The method according to any one of claims 1 to 6, wherein the determining the closed distribution characteristics of the spatial points in the point cloud subset according to the first distances between the spatial points in the ray cluster and the ray origin comprises: Determining the first distances from each of the spatial points to the ray origin; When the difference between the maximum value and the minimum value among the first distances satisfies the first distance difference condition, determining the ray cluster as a candidate ray cluster; Based on the proportion of the candidate ray clusters in each ray cluster of the point cloud subset, determining the closed distribution characteristics of the spatial points in the point cloud subset.

8. The method according to any one of claims 1 to 7, wherein the determining the center point of the ray cluster based on the spatial points in the ray cluster and determining the continuous distribution characteristics of the spatial points in the point cloud subset according to the second distance between the center point and the ray origin comprises: When the closed distribution characteristics indicate that the point cloud subset is in a closed state, determining the second distances from the center points of the candidate ray clusters in the point cloud subset to the ray origin; When the difference between the second distance of the candidate ray cluster and the second distance of the adjacent candidate ray cluster satisfies the second distance difference condition, determining the candidate ray cluster as a normal ray cluster; Based on the proportion of the normal ray clusters in each candidate ray cluster of the point cloud subset, determining the continuous distribution characteristics of the spatial points in the point cloud subset.

9. The method according to any one of claims 1 to 8, wherein the identifying the motion sub-path in the tunnel scenario in the motion path according to the closed distribution characteristics and the continuous distribution characteristics of the spatial points of each of the multiple point cloud subsets comprises: Determining the scene type of the sampling path points according to the closed distribution characteristics and the continuous distribution characteristics of the spatial points in the point cloud subset; Determining the motion sub-path to which at least two continuous sampling path points whose scene type is the tunnel scene type in the motion path belong as the motion sub-path in the tunnel scenario.

10. The method according to claim 9, wherein the determining the scene type of the sampling path points according to the closed distribution characteristics and the continuous distribution characteristics of the spatial points in the point cloud subset comprises: When the closed distribution characteristics of the spatial points in the point cloud subset indicate that the point cloud subset is in a closed state and the continuous distribution characteristics indicate that the point cloud subset is continuous, determining the point cloud subset as a suspected tunnel point cloud subset; When the proportion of the suspected tunnel point cloud subset in the point cloud subset corresponding to the sampling path point reaches the proportion threshold, determining the scene type corresponding to the sampling path point as the tunnel scene type.

11. The method according to any one of claims 1 to 10, wherein the method comprises: During the update process of the point cloud map, when the motion sub-path corresponding to the point cloud data to be processed is a motion sub-path in a tunnel scenario, register the point cloud data to be processed with the point cloud map to obtain an updated point cloud map.

12. An identification device for a tunnel scenario, the device comprising: A point cloud data acquisition module for acquiring point cloud data corresponding to the position of the motion path of a vehicle; A point cloud data segmentation module for segmenting the spatial points in the point cloud data according to spatial positions to obtain a plurality of point cloud subsets; A ray cluster determination module for, in the point cloud subsets, emitting rays from a fixed reference point as the ray origin towards each spatial point in the point cloud subsets and determining the ray clusters to which the rays belong; A distribution feature determination module for determining the closedness distribution feature of the spatial points in the point cloud subsets according to the first distances between the spatial points in the ray clusters and the ray origin; Determine the center point of the ray cluster based on the spatial points in the ray cluster, and determine the continuity distribution feature of the spatial points in the point cloud subsets according to the second distance between the center point and the ray origin; and A scenario identification module for identifying the motion sub-path in the tunnel scenario in the motion path according to the closedness distribution feature and the continuity distribution feature of the spatial points of each of the plurality of point cloud subsets.

13. A computer device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the steps of the method according to any one of claims 1 to 11 when executing the computer program.

14. A computer-readable storage medium having a computer program stored thereon, the computer program implementing the steps of the method according to any one of claims 1 to 11 when executed by a processor.

15. A computer program product comprising a computer program which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 11.

Images