Point cloud data compliance processing method, device, storage medium and program product

Abstract

The application discloses a point cloud data compliance processing method and device, a storage medium and a program product, relates to the technical field of data compliance, and comprises the following steps: generating a target image based on a first point cloud data set collected by a target device for a target road; the target image comprises at least one linear element of the target road; generating a road surface area of the target road in a target space based on a target trajectory point, the target image and a road width of the target road; the target trajectory point comprises a trajectory point collected in a collection process of the first point cloud data set; and determining point cloud data in the first point cloud data set matched with the road surface area as a second point cloud data set of the target road based on the target space. In this way, the accuracy of compliance cropping of the point cloud data set can be improved.

Description

Technical Field

[0001] This application belongs to the field of data compliance technology, and in particular relates to a point cloud data compliance processing method, device, storage medium and program product. Background Technology

[0002] Point cloud data in intelligent assisted driving is three-dimensional spatial data collected by radar (such as lidar). It can accurately reconstruct the physical environment around the vehicle and provide key spatial information for high-precision map construction, target detection and intelligent driving model training.

[0003] To ensure the security and privacy compliance of point cloud data, such data must undergo anonymization before use (e.g., before training intelligent driving models). A key step in this process is to remove irrelevant or sensitive point cloud data (such as buildings and pedestrians) outside the vehicle's traffic area. However, many point cloud cropping technologies rely on simple bounding boxes or single sensor data, lacking precise perception of road structures, resulting in insufficient accuracy in compliant point cloud data cropping. Summary of the Invention

[0004] This application provides a point cloud data compliance processing method, device, storage medium, and program product, which can improve the accuracy of point cloud data compliance trimming.

[0005] In a first aspect, embodiments of this application provide a point cloud data compliance processing method, including:

[0006] A target image is generated based on a first point cloud dataset collected by the target device for the target road; the target image includes at least one line feature of the target road;

[0007] Based on the target trajectory points, the target image, and the road width of the target road, a road surface area of ​​the target road is generated in the target space; the target trajectory points include the trajectory points collected during the first point cloud dataset collection process;

[0008] Based on the target space, the point cloud data in the first point cloud dataset that matches the road surface area is determined as the second point cloud dataset of the target road.

[0009] In some embodiments, generating the road surface area of ​​the target road in the target space based on the target trajectory points, the target image, and the road width of the target road includes:

[0010] In the target space, at least two adjacent trajectory points are connected to form at least two trajectory point segments;

[0011] For each of the at least two trajectory point segments, the following processing is performed to obtain a first boundary point set and a second boundary point set: The perpendicular line of the trajectory point segment is determined, and the intersection of the perpendicular line and the target line element in the target image is added to the first boundary point set as a first boundary point; based on the first boundary point and the road width of the target road, a second boundary point is determined on the perpendicular line and added to the second boundary point set; the target line element is a line element in the target image that meets preset conditions.

[0012] The road surface area of ​​the target road is determined based on the first set of boundary points and the second set of boundary points.

[0013] In some embodiments, determining the road surface area based on the first set of boundary points and the second set of boundary points includes:

[0014] Connect the first boundary points in the first boundary point set sequentially to form a first boundary; and connect the second boundary points in the second boundary point set sequentially to form a second boundary;

[0015] The area between the first boundary and the second boundary is defined as the road surface area.

[0016] In some embodiments, the target line feature is a line feature in the target image that matches the traffic rules followed by the target device.

[0017] In some embodiments, when the target device follows a right-hand driving rule, the target line feature is a line feature in the target image located to the right of the trajectory point line segment; or,

[0018] When the target device follows the left-hand driving rule, the target line element is the line element located to the left of the trajectory point line segment in the target image.

[0019] In some embodiments, generating a target image based on a first point cloud dataset collected by the target device for the target road includes:

[0020] The point cloud data of the first point cloud dataset is divided into regions to form at least two point cloud data region blocks;

[0021] Based on each of the point cloud data region blocks, generate a point cloud pseudo-image corresponding to each of the point cloud data region blocks;

[0022] Based on the spatial coordinate information of the line elements in each of the point cloud pseudo-images, the line elements in each of the point cloud pseudo-images are fused to form the target image.

[0023] In some embodiments, generating a pseudo-point cloud map corresponding to each point cloud data region block based on each point cloud data region block includes:

[0024] The following processing is performed on each of the aforementioned point cloud data regions:

[0025] The point cloud data of the point cloud data region block is allocated to a preset pseudo-map grid;

[0026] The preset pseudo-image mesh after the point cloud data is allocated is filled with background.

[0027] The preset pseudo-image mesh after background filling is image-synthesized to form the point cloud pseudo-image corresponding to the point cloud data region block.

[0028] In some embodiments, before generating the road surface area of ​​the target road in the target space based on the target trajectory points, the target image, and the road width of the target road, the method further includes:

[0029] At least two target trajectory points are projected onto the road network map corresponding to the target map, wherein the road network map includes at least two road segments;

[0030] Based on the distance between each target trajectory point and each road segment in the road network map, the candidate road segment corresponding to each target trajectory point is determined;

[0031] The target road is determined based on the candidate road segments corresponding to each target trajectory point;

[0032] Extract the road width of the target road from the target map.

[0033] In some embodiments, determining the target road based on the candidate road segments corresponding to each target trajectory point includes:

[0034] Based on the candidate road segments corresponding to each target trajectory point, path planning is performed to generate at least two candidate paths, and each candidate path includes at least one of the candidate road segments.

[0035] Among the at least two candidate paths, the candidate path whose path length is closest to the trajectory point migration distance is the target road, wherein the trajectory point migration distance is the sum of the distances between any two adjacent target trajectory points.

[0036] In some embodiments, determining the point cloud data in the first point cloud dataset that matches the road surface area as the second point cloud dataset of the target road, based on the target space, includes:

[0037] Project each point cloud data in the first point cloud dataset onto the target space to obtain the point cloud projection point corresponding to each point cloud data;

[0038] The point cloud projection points that match the road surface area among the point cloud projection points are determined as the target projection points;

[0039] A second point cloud dataset is generated based on the point cloud data corresponding to each of the target projection points in the first point cloud dataset.

[0040] Secondly, embodiments of this application also provide an electronic device, the electronic device comprising: a processor and a memory storing computer program instructions;

[0041] When the processor executes the computer program instructions, it implements the point cloud data compliance processing method as described in any of the first aspects.

[0042] Thirdly, embodiments of this application also provide a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the point cloud data compliance processing method as described in any one of the first aspects.

[0043] Fourthly, embodiments of this application also provide a computer program product, wherein the instructions in the computer program product, when executed by the processor of an electronic device, cause the electronic device to implement the point cloud data compliance processing method as described in any one of the first aspects.

[0044] In this embodiment, the electronic device can generate a target image including at least one line element of the target road based on a first point cloud dataset collected by the target device for the target road. Then, based on the target image, the road width of the target road, and the target trajectory points collected by the target device at the same time as collecting the first point cloud dataset, the road surface area of ​​the target road is generated. That is, by integrating multi-source data including trajectory points, road width, and point cloud data, the road surface area of ​​the target road traversed by the target device is accurately generated. Furthermore, based on the accurate road surface area, a second point cloud dataset that accurately matches the target road is determined from the original first point cloud dataset, thereby improving the accuracy of compliant cropping of the point cloud dataset. Attached Figure Description

[0045] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1This is a flowchart illustrating an embodiment of the point cloud data compliance processing method provided in this application;

[0047] Figure 2 This is a schematic diagram of the target image in an embodiment of the point cloud data compliance processing method provided in this application;

[0048] Figure 3 This is a schematic diagram of trajectory points projected onto a road network map in an embodiment of the point cloud data compliance processing method provided in this application;

[0049] Figure 4 This is a schematic diagram of the regional division of point cloud data in an embodiment of the point cloud data compliance processing method provided in this application;

[0050] Figure 5 This is a schematic diagram illustrating the determination of boundary points in an embodiment of the point cloud data compliance processing method provided in this application;

[0051] Figure 6 This is a schematic diagram of a road surface area in an embodiment of the point cloud data compliance processing method provided in this application;

[0052] Figure 7 This is a schematic diagram of the structure of an embodiment of the point cloud data compliance processing device provided in this application;

[0053] Figure 8 This is a schematic diagram of the structure of an embodiment of the electronic device provided in this application. Detailed Implementation

[0054] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples of this application.

[0055] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0056] Point cloud data is three-dimensional spatial data collected by radar (such as lidar). It can accurately reconstruct the surrounding physical environment and provide key spatial information for high-precision map construction, target detection and model training.

[0057] However, to ensure the security and privacy compliance of point cloud data, point cloud data must be anonymized before use. A key step in this process is to remove irrelevant or sensitive point cloud data (such as buildings, pedestrians, etc.) from areas outside of vehicle traffic routes.

[0058] For example, in the training of intelligent driving models, after obtaining the point cloud dataset collected by the vehicle while driving on a road, in order to ensure that the point cloud dataset is compliant when used for training the intelligent driving model, it is necessary to remove the point cloud data in the range of more than 15 meters away from the road boundary in the point cloud dataset to achieve compliant processing of the point cloud data, and then use the compliant point cloud dataset as sample data for training the intelligent driving model.

[0059] However, point cloud cropping technology in related technologies often relies on simple bounding boxes or single sensor data, lacking accurate perception of road structures, resulting in insufficient cropping accuracy.

[0060] To address the aforementioned issue of insufficient accuracy in point cloud data cropping, this application provides a point cloud data compliance processing method, electronic device, storage medium, and program product. The method generates a target image based on a first point cloud dataset collected by a target device for a target road; the target image includes at least one line feature of the target road; a road surface area of ​​the target road is generated in a target space based on target trajectory points, the target image, and the road width of the target road; the target trajectory points include trajectory points collected during the acquisition of the first point cloud dataset; and based on the target space, point cloud data in the first point cloud dataset that matches the road surface area is determined as a second point cloud dataset for the target road.

[0061] The aforementioned target device can be a point cloud data acquisition device; the target device can be, but is not limited to, a first device that can move autonomously on the road and is equipped with a point cloud data acquisition unit, such as, but is not limited to, any one or any combination of vehicles, robot dogs, robotic robots, scooters, etc., equipped with point cloud acquisition units; the target device can also be a second device that cannot move autonomously on the road but is equipped with a point cloud data acquisition unit. When implementing this scheme, surveyors can carry the second device and move it on the target road to collect the first point cloud data, or the second device can be placed on a third device that can move autonomously (such as a vehicle, robot dog, or robot, etc.), and then the second device collects the first point cloud dataset of the target road while moving with the third device.

[0062] The aforementioned point cloud acquisition unit may include, but is not limited to, at least one or any combination of lidar, RGB-D camera, synthetic aperture radar, etc.

[0063] As one embodiment, the aforementioned target device may also be configured with a trajectory point acquisition unit. The trajectory point acquisition unit can acquire the target trajectory points of the target device during the process of the target device acquiring the first point cloud dataset. The trajectory point acquisition unit may include, but is not limited to, any one or any combination of satellite navigation and positioning devices (such as at least one of GPS positioning devices, Beidou navigation and positioning devices, GLONASS positioning devices, etc.), cellular network positioning devices, inertial navigation devices (INS), visual positioning devices, ultrasonic positioning devices, and infrared positioning devices.

[0064] In addition, the above-mentioned point cloud data compliance processing method can be executed by an electronic device, which can be the target device mentioned above. For example, if the electronic device is a vehicle equipped with a point cloud acquisition unit, the above-mentioned point cloud data compliance processing method can be executed by the vehicle's on-board terminal. Alternatively, the above-mentioned electronic device can also be a processing device other than the target device, such as a server, server cluster, other terminals, IoT devices, vehicle networking devices, etc. That is, the target device sends the collected point cloud data and target trajectory points to the electronic device, and the electronic device executes the above-mentioned point cloud data compliance processing method based on the point cloud data and target trajectory points.

[0065] For ease of understanding, this application embodiment uses a vehicle equipped with a point cloud acquisition unit and a trajectory point acquisition unit as the target device, and the above-mentioned electronic device is a specific example of the vehicle for illustration.

[0066] During the process of the vehicle traveling on road A (i.e., the target road), the vehicle collects point cloud data through a point cloud acquisition unit and simultaneously collects trajectory points through a trajectory point acquisition unit. The vehicle can use all the point cloud data collected on road A as the first point cloud dataset and at least a portion of the collected trajectory points (e.g., trajectory points collected at preset intervals (e.g., 10 seconds) as the target trajectory points. The vehicle's onboard terminal generates a target image including at least one line element of road A based on the first point cloud dataset. Then, based on the target trajectory points, the target image, and the road width of road A (which can be extracted from map data), the onboard terminal generates a road surface area of ​​road A in the map coordinate system (i.e., the target space). Next, the onboard terminal determines a point cloud data removal area in the map coordinate system that is more than a preset range value (e.g., 15 meters) from the boundary of the road surface area required by regulations. Point cloud data whose projection points in the map coordinate system are located within this point cloud data removal area are removed. The first point cloud dataset after point cloud data removal is used as the second point cloud dataset.

[0067] Based on this, the electronic device can generate a target image including at least one line feature of the target road based on the first point cloud dataset collected by the target device for the target road; then, based on the target image, the road width of the target road, and the target trajectory points collected by the target device at the same time as the first point cloud dataset, the road surface area of ​​the target road is generated. That is, by integrating multi-source data including trajectory points, road width, and point cloud data, the road surface area of ​​the target road traversed by the target device is accurately generated, and further, based on the accurate road surface area, a second point cloud dataset that accurately matches the target road is determined from the original first point cloud dataset, thereby improving the accuracy of compliant cropping of the point cloud dataset.

[0068] Figure 1 This is a flowchart illustrating an embodiment of the point cloud data compliance processing method provided in this application. The point cloud data compliance processing method of this application is applied to the aforementioned electronic device. Figure 1 As shown, the point cloud data compliance processing method includes, but is not limited to, the following steps S101 to S104.

[0069] Step S101: Generate a target image based on the first point cloud dataset collected by the target device for the target road.

[0070] In this step, the electronic device can acquire the first point cloud dataset collected by the target device for the target road, and generate the target image based on the first point cloud dataset.

[0071] The electronic device acquires the first point cloud dataset collected by the target device for the target road. This can be done in two ways: if the electronic device is the target device, the target device directly collects the first point cloud dataset through the point cloud data unit; or if the electronic device is a processing device other than the target device, the target device collects the first point cloud dataset through the point cloud dataset unit and then transmits the first point cloud dataset to the electronic device.

[0072] It should be noted that the above-mentioned point cloud data compliance processing method of this application can be executed online, that is, after the target device collects the first point cloud dataset, it transmits it to the electronic device in real time, and the electronic device executes the above-mentioned steps S101 to S103 in real time; or, the above-mentioned point cloud data compliance processing method can be executed offline, that is, after the target device collects the first point cloud dataset, it does not need to transmit it to the electronic device in real time, and the electronic device does not execute the above-mentioned steps S101 to S103 in real time. For example, the electronic device can receive the point cloud dataset collected by at least one vehicle through the point cloud data unit, and then execute the above-mentioned steps S101 to S103 in turn according to each point cloud dataset (that is, the point cloud dataset used at this time is the first point cloud dataset).

[0073] The aforementioned target road refers to the road that the target device travels on within a certain period of time. The target road can be a pre-set known road, such as a fixed road in the test site; or, the target road can be a non-pre-set known road. The electronic device can determine the target road associated with the aforementioned first point cloud dataset, for example, by determining it on the map based on the navigation path of the target device during the collection period of the first point cloud dataset, that is, by finding a road on the map that matches the navigation path traveled during that period as the target road.

[0074] The target image mentioned above includes at least one line feature of the target road. The line feature refers to the feature of the road surface linear markers representing the target road extracted from the first point cloud dataset, such as lane lines, center lines or boundary lines. Based on the at least one line feature, the road surface area features can be reflected to a certain extent.

[0075] For example, based on the first point cloud dataset mentioned above, it is possible to generate, as follows: Figure 2 The target image shown includes line features such as boundary line 21 and center line 22.

[0076] The above-mentioned generation of the target image based on the first point cloud dataset collected by the target device for the target road can be achieved by the electronic device being equipped with a point cloud image generation model based on deep learning. The electronic device inputs the point cloud data of the first point cloud dataset into the point cloud image generation model, and the point cloud image generation model outputs the above-mentioned target image.

[0077] The aforementioned point cloud image generation model can be trained using a large amount of first training sample data. Each first training data can include a historical point cloud dataset and image labels corresponding to the historical point cloud dataset. The historical point cloud dataset is a point cloud dataset collected on a historical road. The image labels can be images with linear features that are manually labeled based on the historical point cloud dataset.

[0078] Step S102: Based on the target trajectory points, target image, and road width of the target road, generate the road surface area of ​​the target road in the target space.

[0079] In this step, while the electronic device generates the target image, it can also acquire the target trajectory points and the road width of the target road collected by the target device during the first point aggregation acquisition process, and generate the road surface area of ​​the target road in the target space based on the target trajectory points, the target image and the road width.

[0080] The aforementioned target trajectory points are the trajectory points collected by the target device during the first point cloud dataset acquisition process. That is, the target device collects trajectory points simultaneously with the acquisition of the first point cloud dataset. For example, if a vehicle is traveling on road B for a certain period of time, the vehicle can collect point cloud data during its travel on road B through its point cloud acquisition unit during that period of time, and the vehicle can also collect at least two trajectory points during its travel on road B through its trajectory point acquisition unit during that period of time.

[0081] It should be noted that the electronic device may use each track point collected as the target track point; or it may use a portion of the collected track points as the target track point. For example, it may use the track points collected at a preset time interval (such as 5 seconds) as the target track point.

[0082] The aforementioned target trajectory points are typically spatial coordinate data (including latitude, longitude, and altitude data, or three-dimensional spatial coordinate data) representing the trajectory position of the target device in a global coordinate system (such as the world coordinate system or map coordinate system (such as the GCJ02 coordinate system)). The aforementioned target road width usually refers to the distance between the two road boundary lines in the global coordinate system. The aforementioned point cloud data is usually three-dimensional spatial data in a local coordinate system (such as the ENU coordinate system). Therefore, the line features in the target image generated based on the aforementioned first point cloud dataset also reflect the characteristics of linear markings on the road surface in the local coordinate system.

[0083] Therefore, to ensure that the road surface area of ​​the target road is generated based on data (including target trajectory points, target image, and road width of the target road) in different coordinate systems, the data can be processed in the same coordinate system to generate the road surface area of ​​the target road. Specifically, the road surface area of ​​the target road can be generated in a global coordinate system based on the target trajectory points, target image, and road width of the target road; that is, the target space is the global coordinate system.

[0084] The above-mentioned method of generating the road surface area of ​​the target road in the target space based on the target trajectory points, target image, and road width of the target road can be achieved by first determining two road boundary lines based on the target trajectory points, target image, and road width of the target road, and then determining the area in the target space located between the two road surface boundary lines as the road surface area of ​​the target road.

[0085] The above-mentioned determination of two road boundary lines based on the target trajectory point, the target image, and the road width of the target road can be achieved by the electronic device also being equipped with a road boundary line model based on deep learning. The electronic device inputs the target trajectory point, the aforementioned target image, and the road width of the target road into the road boundary line model, and the road boundary line model can output two road boundary lines in the target space.

[0086] It should be noted that the aforementioned road boundary line model can be obtained by training the neural network model with a large amount of second training sample data. Each second sample data can include historical trajectory points, historical images, the road width of historical roads, and two historical road boundary lines in the corresponding target space. These two historical road boundary lines can be manually generated based on the annotations of historical trajectory points, historical images, and the road width of historical roads.

[0087] Step S103: Based on the target space, the point cloud data that matches the road surface area in the first point cloud dataset is determined as the second point cloud dataset of the target road.

[0088] In this step, the point cloud data in the first point cloud dataset that matches the road surface area is determined as the second point cloud dataset of the target road based on the target space. Alternatively, in the target space, a point cloud removal area can be determined based on the road surface area and compliance requirements. In the target space, the point cloud data in the first point cloud dataset that is located within the point cloud removal area can be removed. The remaining point cloud data after removal forms the second point cloud dataset.

[0089] For example, in a scenario where compliance requirements necessitate the removal of point cloud data located outside a 15m radius on both sides of the road surface area, the area outside the 15m radius on both sides of the road surface area in the target space can be defined as the aforementioned point cloud removal area. Point cloud data located within this removal area in the target space can then be removed, leaving only the point cloud data located within a 15m radius on both sides of the road surface area as the aforementioned second point cloud dataset.

[0090] It should be noted that the point cloud data in the first point cloud dataset that matches the road surface area can also be point cloud data that is below a preset height (e.g., 3m) in the target space and located outside the point cloud removal area.

[0091] In this embodiment, a target image including at least one line element of the target road is generated by collecting a first point cloud dataset from the target device for the target road. Then, based on the target image, the road width of the target road, and the target trajectory points collected by the target device while collecting the first point cloud dataset, the road surface area of ​​the target road is generated. That is, by integrating multi-source data including trajectory points, road width, and point cloud data, the road surface area of ​​the target road traversed by the target device is accurately generated. Furthermore, based on the accurate road surface area, a second point cloud dataset that accurately matches the target road is determined from the original first point cloud dataset, thereby improving the accuracy of compliant cropping of the point cloud dataset.

[0092] In some embodiments, before generating the road surface area of ​​the target road in the target space based on the target trajectory points, the target image, and the road width of the target road, the method further includes:

[0093] Project at least two trajectory points onto the road network map corresponding to the target map, where the road network map includes at least two road segments;

[0094] Based on the distance between each target trajectory point and each road segment in the road network map, the candidate road segments corresponding to each target trajectory point are determined;

[0095] Based on the candidate road segments corresponding to each target trajectory point, the target road is determined;

[0096] Extract the road width of the target road from the target map.

[0097] In this embodiment, the electronic device can determine the candidate road segments corresponding to each target trajectory point based on the distance between each target trajectory point and each road segment in the road network map of the target map, and determine the target road based on the distance between each target trajectory point. This allows the target device to accurately determine the target road it travels on in the target map, thereby extracting the accurate road width and ensuring the accuracy of subsequent determination of the road surface area.

[0098] The aforementioned road network map can be a topology map generated based on map data from the target map. This topology map can be a weighted directed graph or a weighted undirected graph. Nodes in the road network map represent intersections, etc., while edges in the road network map represent road segments.

[0099] The target map mentioned above can be a map of a region. For example, it can be high-precision map data extracted within a 500-meter radius of the above at least two trajectory points as the target map.

[0100] The above method of determining the candidate road segment corresponding to each target trajectory point based on the distance between each target trajectory point and each road segment in the road network map can be achieved by projecting the target trajectory point onto each road segment (edge) located around the target trajectory point in the road network map; obtaining the distance (also called "projection distance") between the target trajectory point and the projected point on each road segment; and determining the road segment where the projected point with the closest projection distance to the target trajectory point is located as the candidate road segment corresponding to the target trajectory point.

[0101] It should be noted that the electronic device can determine the candidate road segment corresponding to each of the above at least two target trajectory points. The candidate road segments corresponding to different target trajectory points among the at least two target trajectory points may be different, or the candidate road segments corresponding to two or more trajectory points among the at least two target trajectory points may be the same. For example, when the acquisition interval of the target trajectory points is short, two or more consecutively acquired target trajectory points may have the shortest projection distance on the same candidate road segment.

[0102] For example, such as Figure 3 As shown, the electronic device acquires at least two target trajectory points, including trajectory point N1, trajectory point N2, trajectory point N3, and trajectory point N4 (i.e., the yellow dots represent target trajectory points). The electronic device can project trajectory point N1 onto road segments (including road segments R1, R4, ...) located around trajectory point N1 in the road network map. The electronic device determines that the projection point (i.e., the red dot) with the closest projection distance to trajectory point N1 is located on road segment R1, that is, road segment R1 is the candidate road segment corresponding to trajectory point N1. Similarly, road segment R2 can be determined as the candidate road segment corresponding to trajectory point N2, and road segment R5 can be determined as the candidate road segments corresponding to trajectory points N3 and N4.

[0103] The above method of determining the target road based on the candidate road segments corresponding to each target trajectory point can be achieved by performing path planning based on the candidate road segments corresponding to at least two target trajectory points to generate at least two candidate paths; based on the collection order of at least two target trajectory points (which can be determined by the collection timestamp of the target trajectory points), adjacent trajectory points are connected sequentially by straight line segments to form a trajectory point connection line; among the at least two candidate paths, the candidate path whose path shape matches the shape of the trajectory point connection line is selected as the above target road.

[0104] The above-mentioned path planning is based on the candidate road segments corresponding to at least two target trajectory points to generate at least two candidate paths. This can be achieved by using a preset path planning algorithm to generate at least two candidate paths, and each candidate path includes at least one candidate road segment corresponding to a target trajectory point.

[0105] For example, such as Figure 3 As shown, given the candidate road segments R1, R2, and R5, path planning can determine candidate path 1 (i.e., R1+R2) and candidate path 2 (i.e., R1+R4+R5). Since the shape of candidate path 2 best matches the shape of the line connecting trajectory points N1, N2, N3, and N4, candidate path 2 is determined as the target road.

[0106] In some embodiments, determining the target road based on the candidate road segments corresponding to each target trajectory point includes:

[0107] Path planning is performed based on the candidate road segments corresponding to each target trajectory point to generate at least two candidate paths, each candidate path including at least one candidate road segment.

[0108] Among at least two candidate paths, the candidate path whose path length is closest to the trajectory point migration distance is selected as the target road, where the trajectory point migration distance is the sum of the distances between any two adjacent target trajectory points.

[0109] In this embodiment, the target road is determined more accurately by selecting the candidate path whose path length is closest to the migration distance of at least two trajectory points from among at least two candidate paths.

[0110] For example, such as Figure 3 As shown, given the candidate road segments R1, R2, and R5, path planning can determine candidate path 1 (R1+R2) and candidate path 2 (R1+R4+R5). Since the path length of candidate path 2 is closest to the migration distance of trajectory points N1, N2, N3, and N4, candidate path 2 is determined as the target road.

[0111] The above trajectory point migration distance is the sum of the distances between two adjacent target trajectory points in each target trajectory point. It can be understood as calculating the straight-line distance between two adjacent target trajectory points in the collection order of at least two target trajectory points, and determining the sum of the straight-line distances between all two adjacent trajectory points as the above trajectory point migration distance.

[0112] For example, such as Figure 3 As shown, since the target device sequentially collects trajectory points N1, N2, N3 and N4, the electronic device obtains the straight-line distance 1 between trajectory point N1 and trajectory point N2, the straight-line distance 2 between trajectory point N2 and trajectory point N3, and the straight-line distance 3 between trajectory point N3 and trajectory point N4, and determines the sum of straight-line distance 1, straight-line distance 2 and straight-line distance 3 as the trajectory point migration distance of trajectory points N1, N2, N3 and N4.

[0113] In some embodiments, generating a target image based on a first point cloud dataset collected by the target device for the target road includes:

[0114] The point cloud data of the first point cloud dataset is divided into regions to form at least two point cloud data region blocks;

[0115] Based on each point cloud data region block, generate a point cloud pseudo-map corresponding to each point cloud data region block;

[0116] Based on the spatial coordinate information of the line elements in each point cloud pseudo-image, the line elements in each point cloud pseudo-image are fused to form the target image.

[0117] In this embodiment, at least two point cloud data region blocks are formed by dividing the point cloud data of the first point cloud dataset into regions, and point cloud pseudo-images corresponding to each point cloud data region block are generated. Then, based on the spatial coordinate information of the line elements in each point cloud pseudo-image, the line elements in each point cloud pseudo-image are fused to form a target image, thereby reducing the complexity of generating the target image and making the road edges in the target image more complete.

[0118] The above-mentioned division of the point cloud data of the first point cloud dataset into regions to form at least two point cloud data region blocks can be achieved by expanding outward from the center of the first point cloud dataset, dividing the point cloud data of the first point cloud dataset into point cloud data blocks of a preset size, and forming at least two point cloud data region blocks.

[0119] For example, the point cloud data of at least two first point cloud datasets can be divided into regions according to a preset size of 3.5m*3.5m to form a structure like... Figure 4 At least two point cloud data regions are shown (each dashed square represents a point cloud data region).

[0120] Before generating a pseudo-point cloud image based on at least two point cloud data region blocks, the electronic device can also remove point cloud data region blocks containing too few or too sparse point cloud points, and generate a pseudo-point cloud image based on the remaining point cloud data region blocks.

[0121] The aforementioned point cloud pseudo-image can be a three-channel image, and the three channels of this three-channel image can reflect the reflection intensity, height variation, and absolute height of the point cloud.

[0122] The above method generates a pseudo-point cloud image based on each point cloud data region block, which can extract features such as height and reflection intensity of each point cloud in the point cloud data region block; then, by selecting orthographic projection, depth projection or other methods, the three-dimensional point cloud is mapped onto a two-dimensional plane; then, the point cloud features are encoded into pixel values ​​to form a pseudo-point cloud image with corresponding RGB three channels.

[0123] In some embodiments, generating a pseudo-point cloud map corresponding to each point cloud data region block based on each point cloud data region block includes:

[0124] The following processing is performed on each point cloud data region block:

[0125] Assign the point cloud data of the point cloud data region block to the preset pseudo-map grid;

[0126] Background filling is applied to the preset pseudo-map grid after the point cloud data is allocated.

[0127] The preset pseudo-image mesh after background filling is image-based composited to form the point cloud pseudo-image corresponding to the point cloud data region block.

[0128] In this embodiment, the point cloud data of each point cloud data region block is allocated to a preset pseudo-image grid, and the filled preset pseudo-image grid is subjected to background filling processing. Finally, the filled preset pseudo-image grid is used for image synthesis to form a corresponding sub-pseudo-image, thereby making the target image of the formed higher quality.

[0129] The size of the preset pseudo-image grid can be set according to the size of the point cloud data region block. For example, if the cloud data region block is 3.5m*3.5m, the preset pseudo-image grid can be set to 3.5m*3.5m.

[0130] The above-mentioned background filling processing of the preset pseudo-map mesh after allocating point cloud data can be based on fitting the ground plane equation (such as RANSAC fitting the ground plane equation) to fill the blank areas of the preset pseudo-map mesh after allocating point cloud data.

[0131] Before allocating the point cloud data of the point cloud data region block to the preset pseudo-map grid, the electronic device can also remove point cloud data that exceeds a preset height (for example, the preset height is 3 meters) within the point cloud data region block.

[0132] It should be noted that for each of the above at least two point cloud data regions, the point cloud data can be allocated to a preset pseudo-image grid, the preset pseudo-image grid can be filled with background, and the image can be synthesized to obtain a point cloud pseudo-image corresponding to each point cloud data region, thereby obtaining at least two point cloud pseudo-images corresponding to at least two point cloud data regions. That is, the above target image includes the at least two point cloud pseudo-images.

[0133] In the aforementioned at least two point cloud pseudo-images, there may be at least two line features. These at least two line features may be distributed in at least a portion of the point cloud pseudo-images, and the different line features among these at least two line features are discontinuous or at least partially overlap. These at least two line features may include road lane lines, boundary lines, and center lines, etc.

[0134] The above-mentioned method of fusing line elements in each point cloud pseudo-image to form a target image based on the spatial coordinate information of line elements in each point cloud pseudo-image may include: determining, based on the spatial coordinate information of at least two line elements in the above-mentioned at least two point cloud pseudo-images, the overlapping line elements and adjacent line elements with an endpoint spacing of less than or equal to a preset distance (e.g., 0.5 meters) among the at least two line elements; connecting the adjacent line elements with an endpoint spacing of less than or equal to the preset distance to form a single line element; and deleting the overlapping portion of the overlapping line elements to obtain the target image.

[0135] The spatial coordinate information of the aforementioned line elements can be the three-dimensional spatial coordinate position information including the corresponding point cloud data.

[0136] In some embodiments, generating the road surface area of ​​the target road in the target space based on the target trajectory points, the target image, and the road width of the target road includes:

[0137] In the target space, connect adjacent trajectory points of at least two trajectory points to form at least two trajectory point segments;

[0138] For each of the at least two trajectory point segments, the following processing is performed to obtain a first boundary point set and a second boundary point set: the perpendicular line of the trajectory point segment is determined, and the intersection of the perpendicular line and the target line feature in the target image is added to the first boundary point set as the first boundary point; based on the first boundary point and the road width of the target road, a second boundary point is determined on the perpendicular line and added to the second boundary point set; the target line feature is the line feature in the target image that meets the preset conditions;

[0139] The road surface area of ​​the target road is determined based on the first boundary point set and the second boundary point set.

[0140] In this embodiment, at least two trajectory point segments are formed by connecting adjacent trajectory points in pairs. Based on the line elements and road width in the target image, a first boundary point and a second boundary point are determined on the perpendicular line of each trajectory point segment to obtain a first boundary point set and a second boundary point set. Then, the first boundary point set and the second boundary point set determine the road surface area, making it easier to determine the road surface area.

[0141] The adjacent trajectory points among the above-mentioned at least two target trajectory points can be two trajectory points collected in succession according to the collection order of the at least two target trajectory points, and the two trajectory points are connected by a straight line to form a trajectory point line segment.

[0142] For each of the above trajectory point segments, the electronic device can perform the following processing to obtain the first boundary point and the second boundary point corresponding to the trajectory point segment: In the target space, determine the perpendicular line of the trajectory point segment; take the intersection of the perpendicular line and the target line element in the target image as the first boundary point; and determine the point on the perpendicular line whose distance from the first boundary point is the width of the target road as the second boundary point.

[0143] The aforementioned target line features can be line features determined in the target image based on user operations; or they can be line features determined in the target image based on preset rules.

[0144] The perpendicular line to the trajectory point segment described above can be generated at any position on the trajectory point segment. Specifically, it can be generated at the position of the target trajectory point on the trajectory point segment. For example, as... Figure 5 As shown, the target trajectory points collected by the aforementioned target device include trajectory points {A, B, C, D, E, F, G, H, I}. Connecting adjacent trajectory points of these points forms trajectory segments {AB, BC, CD, DE, EF, FG, GH, HI}. For trajectory segment AB, a perpendicular line (represented by a dashed line) passing through trajectory point A can be generated; similarly, for trajectory segment BC, a perpendicular line passing through trajectory point B can be generated; for trajectory segment CD, a perpendicular line passing through trajectory point C can be generated; for trajectory segment DE, a perpendicular line passing through trajectory point D can be generated; for trajectory segment EF, a perpendicular line passing through trajectory point E can be generated; for trajectory segment FG, a perpendicular line passing through trajectory point F can be generated; for trajectory segment GH, a perpendicular line passing through trajectory point G can be generated; and for trajectory segment HI, a perpendicular line passing through trajectory point H can be generated.

[0145] After determining the perpendicular line of the trajectory point segment, the first boundary point can be determined based on the intersection of the perpendicular line and the target line feature in the target image. Then, the second boundary point can be determined based on the first boundary point and the road width. Of course, since the line features in the target image are discontinuous, it is possible that the perpendicular line of some trajectory point segments does not intersect with the line features. In this case, the trajectory point segment does not have a corresponding first boundary point and second boundary point.

[0146] Specifically, the aforementioned target line elements are line elements in the target image that match the traffic rules followed by the target device, thereby ensuring that the determined road surface area matches the actual traffic rules.

[0147] In some embodiments, when the target device follows a right-hand driving rule, the target line feature is the line feature in the target image located to the right of the trajectory point line segment; or,

[0148] When the target device follows the left-hand driving rule, the target line element is the line element located to the left of the trajectory point line segment in the target image.

[0149] In this embodiment, different target road edges can be selected for vehicles based on different driving rules, thereby making the application scenarios more extensive.

[0150] For example, such as Figure 5 As shown, after determining the perpendicular lines of the trajectory point line segments {AB, BC, CD, DE, EF, FG, GH, HI}, if the target device is a vehicle that follows the right-hand driving rule, the electronic device can select the line features (i.e., the target line features) located to the right of the trajectory point line segments. Figure 5 The intersection of the orange line segment (representing a line element) with the perpendicular line of each trajectory point line segment is determined as the first boundary point. That is, the first boundary point corresponding to trajectory point line segment AB is A1, the first boundary point corresponding to trajectory point line segment BC is B1, the first boundary point corresponding to trajectory point line segment CD is C1, the first boundary point corresponding to trajectory point line segment EF is E1, the first boundary point corresponding to trajectory point line segment FG is F1, the first boundary point corresponding to trajectory point line segment GH is G1, and the first boundary point corresponding to trajectory point line segment HI is H1. However, since the perpendicular line of trajectory point line segment DE does not intersect with any line element, trajectory point line segment DE does not have a corresponding first boundary point.

[0151] After determining the first boundary points {A1, B1, C1, E1, F1, G1, H1}, the electronic device can determine the second boundary points {A2, B2, C2, E2, F2, G2, H2} on the vertical line of each first boundary point, based on the road width w of the target road.

[0152] If the target line element is a line element located to the right of the trajectory point line segment in the target image, and if there are at least two line elements located to the right of the trajectory point line segment that intersect with the perpendicular line of the trajectory point line segment, then the intersection point farthest from the trajectory point line segment can be determined as the first boundary point.

[0153] Similarly, if the target line element is a line element located to the left of the trajectory point line segment in the target image, and there are at least two line elements located to the left of the trajectory point line segment that intersect with the perpendicular line of the trajectory point line segment, then the intersection point farthest from the trajectory point line segment can also be determined as the first boundary point.

[0154] In some embodiments, determining the road surface area based on the first boundary point set and the second boundary point set includes:

[0155] Connect the first boundary points in the first boundary point set sequentially to form the first boundary; and connect the second boundary points in the second boundary point set sequentially to form the second boundary;

[0156] The area between the first boundary and the second boundary is defined as the road surface area.

[0157] In this embodiment, the first boundary is formed by sequentially connecting the first boundary points in the first boundary point set, and the second boundary is formed by sequentially connecting the second boundary points in the second boundary point set. The area in the target space located between the first boundary and the second boundary is determined as the road surface area, thereby making the method of determining the road surface area simpler.

[0158] The first boundary points in the aforementioned first boundary point set are sequentially connected to form the first boundary. This can be achieved by connecting adjacent first boundary points, such that all first boundary points in the first boundary point set are connected to form the first boundary. Similarly, the second boundary points in the aforementioned second boundary point set are sequentially connected to form the second boundary. This can be achieved by connecting adjacent second boundary points, such that all second boundary points in the second boundary point set are connected to form the first boundary. Since the direction of travel of the target device (which can be understood as the extension direction of at least two trajectory point line segments) matches (is similar to or the same as) the extension direction of the target road, the extension directions of the first and second boundaries match the extension directions of at least two trajectory point line segments.

[0159] For example, such as Figure 5As shown, the first boundary points {A1, B1, C1, E1, F1, G1, H1} are connected sequentially to form the first boundary A1-B1-C1-E1-F1-G1-H1 (represented by the red line); and the second boundary points {A2, B2, C2, E2, F2, G2, H2} are connected sequentially to form the second boundary A2-B2-C2-E2-F2-G2-H2 (represented by the purple line). The area between the first boundary A1-B1-C1-E1-F1-G1-H1 and the second boundary A2-B2-C2-E2-F2-G2-H2 is the aforementioned road surface area.

[0160] In some embodiments, the above-mentioned determination of the point cloud data in the first point cloud dataset that matches the road surface area as the second point cloud dataset of the target road based on the target space includes:

[0161] Project the point cloud data in the first point cloud dataset onto the target space to obtain the point cloud projection points corresponding to each point cloud data.

[0162] The point cloud projection points that match the road surface area among all point cloud projection points are determined as the target projection points;

[0163] A second point cloud dataset is generated based on the point cloud data corresponding to each target projection point in the first point cloud dataset.

[0164] In this embodiment, by projecting the point cloud data in the first point cloud dataset into the target space, and generating the second point cloud dataset based on the point cloud data that matches the corresponding point cloud projection points with the road surface area, the coordinate system of the point cloud data is converted into the coordinate system of the road surface area, thereby achieving compliant processing of point cloud data when the coordinate systems of the point cloud data and the road surface area are different.

[0165] The above-mentioned determination of the point cloud projection points that match the road surface area among the point cloud projection points as target projection points can be understood as determining the point cloud projection points located outside the point cloud removal area as target projection points.

[0166] For example, such as Figure 6 As shown, assuming the green area (i.e. the area between the two green boundaries) is the aforementioned road surface area, in the scenario where compliance requirements require the removal of point cloud data located within a 15m range on both sides of the road surface area, it can be determined that the point cloud removal area is the area other than the area between the two red dashed lines. The electronic device can determine the projection point of the area between the two red dashed lines (i.e. the area other than the point cloud removal area) as the target projection point.

[0167] To facilitate understanding of the point cloud data compliance processing method in this application embodiment, a specific application example of the point cloud data compliance processing method is provided here for illustration. The specific processing flow is as follows:

[0168] Step S1: Extract road attributes

[0169] Based on the trajectory information, the trajectory points (coordinate points or GPS points output when the vehicle collects the trajectory) are matched with map data from a standard accuracy (SD) map to find prior knowledge such as the road width of this section of road (i.e., the target road). The core processing logic of this step can be as follows:

[0170] Step S11: Preprocess the road network data to generate a road network map (such as a weighted directed graph of points and lines).

[0171] Step S12: Extract a road network map including road data within a 500m radius of the trajectory point;

[0172] Step S13: For each trajectory point in the trajectory (e.g., Figure 3 Projecting the yellow dots (as shown in Figure 3) onto the road network map, the road with the smallest projection distance (the distance between the yellow and red dots as shown in Figure 3) is selected as the candidate road set (e.g., ...). Figure 3 (R1, R2, R4, and R5 are shown).

[0173] Step S14: Calculate the length of the pathway (i.e., candidate path) composed of all candidate roads, and extract the path whose migration distance is closest to that of the trajectory point (hereinafter). Figure 3 For example, the migration distance is D1+D2+D3; there are two sets of pathways, one set is R1+R2, and the other set is R1+R4+R5. The road attributes (including road width) in the pathways are used in step S3.

[0174] Step S2: Identify road edges based on point cloud data.

[0175] This process involves preprocessing the input point cloud data into an image, then fitting road edge objects based on reflection intensity, and finally generating the road surface area by connecting at least two road edges. The specific processing steps are as follows:

[0176] Step S21, Point Cloud Segmentation:

[0177] The point cloud data is divided into 30cm*30cm (configurable) regions. To maximize point cloud coverage, the division starts from the center of the point cloud and expands outwards until the point cloud boundary, resulting in the following: Figure 3 The diagram shows at least two sub-blocks (i.e., ). After slicing, remove unrecognizable sub-blocks (i.e., point cloud data region blocks) that contain too few or too sparse point cloud points.

[0178] Step S22: Generate a pseudo-image from the BEV perspective:

[0179] For each sub-block, the point cloud is filtered based on its height value (Z value less than 3m) to obtain point cloud data within the ground area. Then, the point cloud is assigned to a predefined BEV grid (i.e., a preset pseudo-map grid) (30cm x 30cm grid). RANSAC fitting is performed on the points within the grid to fill blank areas using the ground plane equation (areas with higher reflection intensity in the laser point cloud data have a higher probability of being fitted as vector features). Finally, the BEV grid images are synthesized into a three-channel BEV image, with the three channels representing intensity, height variation, and absolute height. The resulting BEV image effectively displays road area features (including road edges).

[0180] Step S23: Identify lane lines from the BEV perspective based on pseudo-point cloud images:

[0181] Lane detection is performed on the generated BEV images for each sub-image, and the lane line positions are detected. For the detected lane line positions, the 3D spatial coordinates of the lane lines in the original point cloud are obtained by referring to the original point cloud information.

[0182] Step S24: Generate road area:

[0183] The lane lines of each submap are merged (overlapping areas are deleted, and two lane lines with a length of less than 0.5m are automatically connected) to generate a global road region for use in S3.

[0184] Step S3: Generate the target road surface area

[0185] By combining the trajectory, raw laser point cloud data, and standard-precision map data (SD map), the road surface area for the uphill (or downhill) direction is calculated. The specific steps are as follows:

[0186] Step S31: Combine consecutive points in the trajectory into at least two line segments (i.e., form a line segment from each pair of adjacent points in the trajectory), such as... Figure 5 As shown;

[0187] Step S32: Except for the first point, generate a perpendicular line for each of the remaining trajectory points based on the line segment from the previous step;

[0188] Step S33: Based on the global road area map generated in S2, prioritize finding the line segment with the closest distance on the "right side" to obtain the "right perpendicular point" (i.e., the first boundary point) and the "right distance value";

[0189] Step S34: Calculate the "left edge point" (i.e., the second boundary point) for each trajectory point by subtracting the nearest distance to the right from the road width; the road width is extracted from the most similar road in S1 and the SD map (the road width is an attribute value in the SD map and does not need to be calculated);

[0190] Repeat steps S33 and S34 above to calculate the "right foot point" and "left edge point" of all trajectory points;

[0191] Step S35: Connect all "right perpendicular foot points" and "left edge points" to generate the target road surface area (e.g., Figure 5 As shown, due to the high density of trajectory points, some trajectory points were selected and their corresponding "right perpendicular foot point" and "left edge point" were marked for explanation. The "left edge point" can distinguish the boundaries between the upward and downward directions.

[0192] Step S35: Based on the compliance threshold specified by the business (i.e., the preset spacing), extend the road to both sides by a threshold distance, forming a structure like... Figure 6 The target area shown is the region between the two red dashed lines.

[0193] Step S4, Point Cloud Cropping: Cropping the original point cloud data.

[0194] After coordinate system alignment, point cloud clipping is performed. The specific coding process is as follows:

[0195] The vast majority of laser point cloud data is expressed in the local coordinate system of ENU, while SD map data is expressed in the coordinate system of GCJ02. Therefore, it is necessary to first unify the coordinate systems of the point cloud points in the ENU coordinate system and the vector points in the GCJ02 coordinate system to prepare for the final cropping.

[0196] Based on the road surface area map finally generated by S3, determine the target range for clipping (e.g., in cases where point cloud data outside the 15m range on both sides of the road specified by the business is non-compliant). Then, refer to the XY dimensions to delete point cloud points that fall outside the target range to meet compliance requirements.

[0197] The point cloud data compliance processing method provided in this application can be executed by a point cloud data processing device. This application uses the execution of the point cloud data compliance processing method by a point cloud data processing device as an example to illustrate the point cloud data processing device provided in this application.

[0198] Figure 7 This is a schematic diagram of the point cloud data compliance processing device provided in an embodiment of this application. Figure 7 As shown, the point cloud data compliance processing device 700 includes:

[0199] Image generation module 701 is used to generate a target image based on a first point cloud dataset collected by the target device for the target road; the target image includes at least one line feature of the target road;

[0200] The road surface area determination module 702 is used to generate the road surface area of ​​the target road in the target space based on the target trajectory points, the target image, and the road width of the target road; the target trajectory points include the trajectory points collected during the first point cloud dataset collection process;

[0201] 703 is used to determine, based on the target space, the point cloud data in the first point cloud dataset that matches the road surface area as the second point cloud dataset of the target road.

[0202] In some embodiments, the road surface area determination module 702 is specifically used for:

[0203] In the target space, at least two adjacent trajectory points are connected to form at least two trajectory point segments;

[0204] For each of the at least two trajectory point segments, the following processing is performed to obtain a first boundary point set and a second boundary point set: The perpendicular line of the trajectory point segment is determined, and the intersection of the perpendicular line and the target line element in the target image is added to the first boundary point set as a first boundary point; based on the first boundary point and the road width of the target road, a second boundary point is determined on the perpendicular line and added to the second boundary point set; the target line element is a line element in the target image that meets preset conditions.

[0205] The road surface area of ​​the target road is determined based on the first set of boundary points and the second set of boundary points.

[0206] In some embodiments, the road surface area determination module 702 is specifically used for:

[0207] Connect the first boundary points in the first boundary point set sequentially to form a first boundary; and connect the second boundary points in the second boundary point set sequentially to form a second boundary;

[0208] The area between the first boundary and the second boundary is defined as the road surface area.

[0209] In some embodiments, the target line feature is a line feature in the target image that matches the traffic rules followed by the target device.

[0210] In some embodiments, when the target device follows a right-hand driving rule, the target line feature is a line feature in the target image located to the right of the trajectory point line segment; or,

[0211] When the target device follows the left-hand driving rule, the target line element is the line element located to the left of the trajectory point line segment in the target image.

[0212] In some embodiments, the image generation module 701 is specifically used for:

[0213] The point cloud data of the first point cloud dataset is divided into regions to form at least two point cloud data region blocks;

[0214] Based on each of the point cloud data region blocks, generate a point cloud pseudo-image corresponding to each of the point cloud data region blocks;

[0215] Based on the spatial coordinate information of the line elements in each of the point cloud pseudo-images, the line elements in each of the point cloud pseudo-images are fused to form the target image.

[0216] In some embodiments, the image generation module 701 is specifically used for:

[0217] The following processing is performed on each of the aforementioned point cloud data regions:

[0218] The point cloud data of the point cloud data region block is allocated to a preset pseudo-map grid;

[0219] The preset pseudo-image mesh after the point cloud data is allocated is filled with background.

[0220] The preset pseudo-image mesh after background filling is image-synthesized to form the point cloud pseudo-image corresponding to the point cloud data region block.

[0221] In some embodiments, the device 700 further includes:

[0222] A trajectory point projection module is used to project at least two target trajectory points onto a road network map corresponding to the target map, wherein the road network map includes at least two road segments;

[0223] The alternative road segment determination module is used to determine the alternative road segment corresponding to each target trajectory point based on the distance between each target trajectory point and each road segment in the road network map;

[0224] The target road determination module is used to determine the target road based on the candidate road segments corresponding to each target trajectory point;

[0225] The road width extraction module is used to extract the road width of the target road from the target map.

[0226] In some embodiments, the target road determination module is specifically used for:

[0227] Based on the candidate road segments corresponding to each target trajectory point, path planning is performed to generate at least two candidate paths, and each candidate path includes at least one of the candidate road segments.

[0228] Among the at least two candidate paths, the candidate path whose path length is closest to the trajectory point migration distance is the target road, wherein the trajectory point migration distance is the sum of the distances between any two adjacent target trajectory points.

[0229] In some embodiments, the compliance processing module 703 is specifically used for:

[0230] Project the point cloud data in the first point cloud dataset onto the target space to obtain the point cloud projection points corresponding to each point cloud data.

[0231] The point cloud projection points that match the road surface area among the point cloud projection points are determined as the target projection points;

[0232] A second point cloud dataset is generated based on the point cloud data corresponding to each of the target projection points in the first point cloud dataset.

[0233] The point cloud data compliance processing device 700 provided in this application embodiment can execute the technical solution shown in the above-described point cloud data compliance processing method embodiment. Its implementation principle and beneficial effects are similar, and will not be described again here.

[0234] Figure 8 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application.

[0235] An electronic device may include a processor 801 and a memory 802 storing computer program instructions.

[0236] Specifically, the processor 801 may include a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or may be configured to implement one or at least two integrated circuits of the embodiments of this application.

[0237] Memory 802 may include mass storage for data or instructions. For example, and not limitingly, memory 802 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 802 may include removable or non-removable (or fixed) media. Where appropriate, memory 802 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 802 is non-volatile solid-state memory.

[0238] In some embodiments, memory 802 may include read-only memory (ROM), random access memory (RAM), disk storage media device, optical storage media device, flash memory device, electrical, optical, or other physical / tangible memory storage device. Therefore, typically, memory includes one or at least two tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or at least two processors), it is operable to perform the operations described with reference to the point cloud data compliance processing method according to this application.

[0239] The processor 801 reads and executes computer program instructions stored in the memory 802 to implement the point cloud data compliance processing method in the above embodiments.

[0240] In one example, the electronic device may also include a communication interface 803 and a bus 810. For example, Figure 8 As shown, the processor 801, memory 802, and communication interface 803 are connected through bus 810 and complete communication with each other.

[0241] The communication interface 803 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0242] Bus 810 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 810 may include one or at least two buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.

[0243] The electronic device can execute the point cloud data compliance processing method in the embodiments of this application, thereby achieving a combination of Figures 1 to 7 The described point cloud data compliance processing method and apparatus.

[0244] In addition, in conjunction with the point cloud data compliance processing method in the above embodiments, this application also provides a computer storage medium for implementation. The computer storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement the point cloud data compliance processing method in the above embodiments.

[0245] In conjunction with the point cloud data compliance processing method in the above embodiments, this application also provides a computer program product, wherein the instructions in the computer program product, when executed by the processor of an electronic device, enable the electronic device to implement the point cloud data compliance processing method in the above embodiments.

[0246] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0247] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0248] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0249] The aspects of this application have been described above with reference to flowchart illustrations and / or block diagrams of methods (systems) and computer program products according to embodiments of this application. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the function / action specified in one or at least two blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by dedicated hardware performing the specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0250] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A method for compliant processing of point cloud data, characterized in that, include: The point cloud data of the first point cloud dataset collected by the target device for the target road is divided into regions to form at least two point cloud data region blocks; Based on each of the point cloud data region blocks, generate a point cloud pseudo-image corresponding to each of the point cloud data region blocks; Based on the spatial coordinate information of the line elements in each point cloud pseudo-image, the line elements in each point cloud pseudo-image are fused to form a target image. In the target space, at least two adjacent trajectory points are connected to form at least two trajectory point line segments, wherein the target space is a global coordinate system; For each of the at least two trajectory point segments, the following processing is performed to obtain a first boundary point set and a second boundary point set: The perpendicular line of the trajectory point segment is determined, and the intersection of the perpendicular line and the target line element in the target image is added to the first boundary point set as a first boundary point; based on the first boundary point and the road width of the target road, a second boundary point is determined on the perpendicular line and added to the second boundary point set; the target line element is a line element in the target image that meets preset conditions. The road surface area of ​​the target road is determined based on the first set of boundary points and the second set of boundary points. Based on the target space, the point cloud data in the first point cloud dataset that matches the road surface area is determined as the second point cloud dataset of the target road.

2. The method according to claim 1, characterized in that, Determining the road surface area based on the first boundary point set and the second boundary point set includes: Connect the first boundary points in the first boundary point set sequentially to form a first boundary; and connect the second boundary points in the second boundary point set sequentially to form a second boundary; The area between the first boundary and the second boundary is defined as the road surface area.

3. The method according to claim 1, characterized in that, The target line feature is a line feature in the target image that matches the traffic rules followed by the target device.

4. The method according to claim 3, characterized in that, When the target device follows the right-hand driving rule, the target line element is the line element in the target image located to the right of the trajectory point line segment; or, When the target device follows the left-hand driving rule, the target line element is the line element located to the left of the trajectory point line segment in the target image.

5. The method according to claim 1, characterized in that, The step of generating a pseudo-point cloud map corresponding to each point cloud data region block based on each point cloud data region block includes: The following processing is performed on each of the aforementioned point cloud data regions: The point cloud data of the point cloud data region block is allocated to a preset pseudo-map grid; The preset pseudo-image mesh after the point cloud data is allocated is filled with background. The preset pseudo-image mesh after background filling is image-synthesized to form the point cloud pseudo-image corresponding to the point cloud data region block.

6. The method according to any one of claims 1-5, characterized in that, Also includes: At least two target trajectory points are projected onto the road network map corresponding to the target map, wherein the road network map includes at least two road segments; Based on the distance between each target trajectory point and each road segment in the road network map, the candidate road segment corresponding to each target trajectory point is determined; The target road is determined based on the candidate road segments corresponding to each target trajectory point; Extract the road width of the target road from the target map.

7. The method according to claim 6, characterized in that, The step of determining the target road based on the candidate road segments corresponding to each target trajectory point includes: Based on the candidate road segments corresponding to each target trajectory point, path planning is performed to generate at least two candidate paths, and each candidate path includes at least one of the candidate road segments. Among the at least two candidate paths, the candidate path whose path length is closest to the trajectory point migration distance is the target road, wherein the trajectory point migration distance is the sum of the distances between any two adjacent target trajectory points.

8. The method according to any one of claims 1-5, characterized in that, The step of determining the point cloud data in the first point cloud dataset that matches the road surface area as the second point cloud dataset of the target road, based on the target space, includes: Project the point cloud data in the first point cloud dataset onto the target space to obtain the point cloud projection points corresponding to each point cloud data. The point cloud projection points that match the road surface area among the point cloud projection points are determined as the target projection points; A second point cloud dataset is generated based on the point cloud data corresponding to each of the target projection points in the first point cloud dataset.

9. An electronic device, characterized in that, The electronic device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the point cloud data compliance processing method as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions; When the computer program instructions are executed by the processor, they implement the point cloud data compliance processing method as described in any one of claims 1-8.

11. A computer program product, characterized in that, When the instructions in the computer program product are executed by the processor of the electronic device, the electronic device performs the point cloud data compliance processing method as described in any one of claims 1-8.

Images