Point cloud feature matching method and device, electronic equipment and storage medium

Abstract

The application provides a point cloud feature matching method and device, electronic equipment and storage medium. The method comprises the following steps: dividing point cloud data collected by a laser sensor into ground point cloud data and above-ground point cloud data; extracting a road marking feature from the ground point cloud data based on reflectivity; extracting a three-dimensional geometric feature from the above-ground point cloud data; matching the road marking feature and the three-dimensional geometric feature with map features of a preset map to obtain semantic matching features. The application improves the performance of front-end feature extraction and correlation matching to achieve accurate positioning.

Description

Technical Field

[0001] This invention relates to the field of lidar data processing, and in particular to a point cloud feature matching method, apparatus, electronic device, and storage medium. Background Technology

[0002] Among commonly used vehicle positioning methods, differential GNSS (Global Navigation Satellite System) is difficult to operate in scenarios with strong electromagnetic interference and multipath effects, making it unsuitable for port environments; while positioning methods based on dense point cloud matching have poor positioning performance due to the large number of large movable objects in port environments.

[0003] Therefore, semantic feature matching, which offers higher robustness and lower system load, can be used for vehicle localization. Semantic features refer to environmental features such as geometric structures and traffic signs that can be categorized and described as road signs. In semantic feature localization, semantic features are extracted from the environment by sensors, and these features are then correlated and matched with pre-stored semantic features in a map to obtain the localization observation. In this method, the performance of the front-end feature extraction and correlation matching parts plays a decisive role in localization.

[0004] Therefore, how to improve the performance of front-end feature extraction and association matching to achieve accurate positioning is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] In order to overcome the shortcomings of the prior art, the present invention provides a point cloud feature matching method, device, electronic device, and storage medium to improve the performance of front-end feature extraction and association matching, so as to achieve accurate positioning.

[0006] According to one aspect of the present invention, a point cloud feature matching method is provided, comprising:

[0007] The point cloud data collected by the laser sensor is divided into ground point cloud data and ground point cloud data;

[0008] Road marking features are extracted from the ground point cloud data based on reflectivity;

[0009] Extract three-dimensional geometric features from the above ground point cloud data;

[0010] The road marking features and three-dimensional geometric features are matched with the map features of a preset map to obtain semantic matching features.

[0011] In some embodiments of this application, the extraction of road marking features from the ground point cloud data based on reflectivity includes:

[0012] The reflectivity compensation coefficient is determined based on environmental parameters and / or laser sensor parameters;

[0013] The reflectivity of the ground point cloud data is adjusted based on the reflectivity compensation coefficient.

[0014] In some embodiments of this application, the extraction of road marking features from the ground point cloud data based on reflectivity includes:

[0015] Meshized ground point cloud data;

[0016] The reflectivity of the ground point cloud data within each grid is compensated based on the positional relationship between each grid and the laser sensor.

[0017] In some embodiments of this application, the extraction of road marking features from the ground point cloud data based on reflectivity includes:

[0018] The mincut algorithm is used to obtain candidate road surface marker point cloud data with reflectivity greater than a set reflectivity threshold.

[0019] Lane line features are extracted from the candidate road surface marker point cloud data using a random consistency algorithm.

[0020] In some embodiments of this application, the extraction of road marking features from the ground point cloud data based on reflectivity includes:

[0021] Using the reflectivity of the ground point cloud data as height information, three-dimensional clustering is performed on the ground point cloud data to obtain clustered point clusters;

[0022] Obtain the minimum convex circumference of each cluster of points;

[0023] The pre-classification landmark type of the cluster point cluster is determined based on the minimum circumscribed convex polygon;

[0024] The ground traffic markers for the clustered point clusters are determined based on the pre-classified road sign types.

[0025] In some embodiments of this application, determining the ground traffic signs of the clustered point clusters based on the pre-classified road sign type includes:

[0026] Based on other clusters within a set distance range of the cluster and their pre-classified landmark types, a first descriptor for the cluster is generated;

[0027] Based on other map features and their corresponding ground traffic signs within a set distance range of the pre-classified road sign type map features in the preset map, a second descriptor of the map features is generated;

[0028] Based on the similarity between the first descriptor and the second descriptor, the ground traffic markers of the clustered point clusters are determined.

[0029] In some embodiments of this application, the extraction of three-dimensional geometric features from the ground point cloud data includes:

[0030] Principal component analysis is performed on the ground point cloud data to extract the feature vectors and feature values ​​of the ground point cloud data, so as to classify the ground point cloud data into different geometric feature categories;

[0031] Clustering of ground point cloud data of the same geometric feature category yields three-dimensional geometric features.

[0032] In some embodiments of this application, performing principal component analysis on the ground point cloud data to extract feature vectors and feature values ​​of the ground point cloud data, and classifying the ground point cloud data into different geometric feature categories, includes:

[0033] Based on the above ground point cloud data, construct a point cloud kd tree;

[0034] Traverse the kd-tree of the point cloud and construct the covariance matrix of the coordinate distribution of each point in the neighborhood of the kd-tree.

[0035] Obtain the eigenvalues ​​and eigenvectors of the covariance matrix;

[0036] The geometric feature category to which the points in the ground point cloud data belong is determined based on the feature values ​​and feature vectors.

[0037] In some embodiments of this application, clustering ground point cloud data of the same geometric feature category to obtain three-dimensional geometric features includes:

[0038] A conditional Euclidean space clustering algorithm is used to cluster ground point cloud data of the same geometric feature category to obtain three-dimensional geometric features.

[0039] According to another aspect of the present invention, a point cloud feature matching apparatus is also provided, comprising:

[0040] The point cloud segmentation module is used to divide the point cloud data collected by the laser sensor into ground point cloud data and ground point cloud data;

[0041] The ground feature extraction module is used to extract road marking features from the ground point cloud data based on reflectance;

[0042] The ground feature extraction module is used to extract three-dimensional geometric features from the ground point cloud data;

[0043] The semantic matching module is used to match the road marking features and three-dimensional geometric features with the map features of a preset map to obtain semantic matching features.

[0044] According to another aspect of the present invention, an electronic device is also provided, the electronic device comprising: a processor; and a storage medium having a computer program stored thereon, the computer program being executed by the processor to perform the steps described above.

[0045] According to another aspect of the present invention, a storage medium is also provided, on which a computer program is stored, the computer program being executed by a processor to perform the steps described above.

[0046] Compared with the prior art, the advantages of this invention are:

[0047] By dividing point cloud data into ground point cloud data and above-ground point cloud data, and extracting road marking features from the ground point cloud data and three-dimensional geometric features from the above-ground point cloud data based on reflectivity, the road marking features and three-dimensional geometric features are matched with map features of a preset map to obtain semantic matching features. Thus, feature extraction can be performed on road marking features and three-dimensional geometric features based on the divided point cloud data, improving feature extraction performance. At the same time, extracting road marking features based on reflectivity improves the accuracy of road feature extraction, thereby improving the accuracy of vehicle positioning, and is particularly applicable to port scenarios. Attached Figure Description

[0048] The above and other features and advantages of the present invention will become more apparent from a detailed description of exemplary embodiments thereof with reference to the accompanying drawings.

[0049] Figure 1 A flowchart of a point cloud feature matching method according to an embodiment of the present invention is shown;

[0050] Figure 2 A flowchart illustrating a method for compensating point cloud reflectance according to an embodiment of the present invention is shown;

[0051] Figure 3 A flowchart illustrating another method for compensating point cloud reflectance according to an embodiment of the present invention is shown;

[0052] Figure 4 A flowchart illustrating the extraction of lane line features according to an embodiment of the present invention is shown;

[0053] Figure 5 A flowchart of ground traffic markings according to an embodiment of the present invention is shown;

[0054] Figure 6 A flowchart illustrating the association between ground traffic signs and map features according to an embodiment of the present invention is shown;

[0055] Figure 7 A schematic diagram of a descriptor according to an embodiment of the present invention is shown;

[0056] Figure 8 A flowchart illustrating the extraction of three-dimensional geometric features from the ground point cloud data according to an embodiment of the present invention is shown;

[0057] Figure 9 A flowchart is shown below illustrating a process for performing principal component analysis on the ground point cloud data according to an embodiment of the present invention to extract the feature vectors and feature values ​​of the ground point cloud data, so as to classify the ground point cloud data into different geometric feature categories.

[0058] Figure 10 A block diagram of a point cloud feature matching apparatus according to an embodiment of the present invention is shown;

[0059] Figure 11 This schematic diagram illustrates a computer-readable storage medium according to an exemplary embodiment of the present disclosure;

[0060] Figure 12 A schematic diagram of an electronic device is shown in accordance with an exemplary embodiment of the present disclosure. Detailed Implementation

[0061] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that this disclosure will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0062] Furthermore, the accompanying drawings are merely illustrative of this disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0063] To address the shortcomings of existing technologies, this invention provides a point cloud feature matching method. See below. Figure 1 , Figure 1 A flowchart of a point cloud feature matching method according to an embodiment of the present invention is shown. Figure 1 The steps are as follows:

[0064] Step S110: Divide the point cloud data collected by the laser sensor into ground point cloud data and ground point cloud data;

[0065] Step S120: Extract road marking features from the ground point cloud data based on reflectivity;

[0066] Step S130: Extract three-dimensional geometric features from the ground point cloud data;

[0067] Step S140: Match the road marking features and three-dimensional geometric features with the map features of the preset map to obtain semantic matching features.

[0068] In the point cloud feature matching method provided by this invention, point cloud data is divided into ground point cloud data and above-ground point cloud data. Road marking features are extracted from the ground point cloud data based on reflectivity, and three-dimensional geometric features are extracted from the above-ground point cloud data. The road marking features and three-dimensional geometric features are then matched with map features of a preset map to obtain semantic matching features. Thus, feature extraction can be performed on road marking features and three-dimensional geometric features based on the divided point cloud data, improving feature extraction performance. At the same time, extracting road marking features based on reflectivity improves the accuracy of road feature extraction, thereby improving the accuracy of vehicle positioning. This method is particularly applicable to port scenarios.

[0069] Specifically, before step S110, a step of acquiring point cloud data is included. Due to the laser sparsity of some laser sensors, distant objects often suffer from insufficient information because their neighborhood has too few points. Therefore, after acquiring one frame of point cloud data from the laser sensor, multiple frames of point cloud data can be stitched together to increase point cloud density by constructing a local point cloud map. Specifically, since the temporal variation of distant objects corresponding to adjacent frames of point cloud data is small, temporal alignment can be used to stitch adjacent frames of point cloud data together to increase point cloud density. Furthermore, point cloud data collected by multiple laser sensors with different perspectives can be spatially stitched together to obtain more complete point cloud data. This application can implement many more variations, which will not be elaborated here.

[0070] Specifically, since the markings of the port environment can be divided into ground markings (road surface marking features) and ground geometric features (three-dimensional geometric features), the point cloud can be pre-segmented in step S110 to improve the efficiency of subsequent feature extraction. Specifically, point cloud segmentation can be achieved through one or more of the following methods: principal component analysis algorithm, random consistency algorithm, and segmentation results provided by other perception modules. This application is not limited to these methods.

[0071] Specifically, due to various factors such as reflectance calibration, ground material, rainwater accumulation, and incident angle, the absolute value of reflectance of the same object in point clouds varies significantly under different environments. Therefore, when extracting features based on reflectance, reflectance compensation can be performed first.

[0072] See below. Figure 2 , Figure 2 A flowchart illustrating a method for compensating point cloud reflectance according to an embodiment of the present invention is shown. Figure 2 The following steps are shown:

[0073] Step S121: Determine the reflectivity compensation coefficient based on environmental parameters and / or laser sensor parameters.

[0074] Specifically, environmental parameters may include one or more of the following: whether it is raining, whether there is standing water on the ground, the current light intensity, and the ground material. Laser sensor parameters may include the installation location of the laser sensor, the viewing angle of the laser sensor, etc., which can be used to determine the incident angle.

[0075] In some embodiments, a compensation mapping table can be maintained to pre-set the mapping relationship between reflectivity compensation coefficient and environmental parameters and / or laser sensor parameters, so that the reflectivity compensation coefficient can be determined by querying the compensation mapping table.

[0076] In other embodiments, environmental parameters and / or laser sensor parameters may be input into a pre-trained reflectivity compensation coefficient prediction model to obtain the reflectivity compensation coefficient predicted by the reflectivity compensation coefficient prediction model.

[0077] Step S122: Adjust the reflectivity of the ground point cloud data based on the reflectivity compensation coefficient.

[0078] Specifically, the reflectivity compensation coefficient can be multiplied by the original reflectivity to compensate for the reflectivity. This application is not limited to this; adding the reflectivity compensation coefficient to the original reflectivity or other compensation calculation methods are all within the scope of protection of this application.

[0079] Therefore, the preprocessing for reflectivity compensation can be completed through the above steps S121 and S122 to ensure the accuracy of road marking feature extraction.

[0080] See below. Figure 3 , Figure 3 A flowchart illustrating another method for compensating point cloud reflectance according to an embodiment of the present invention is shown. Figure 3 The following steps are shown:

[0081] Step S123: Grid the ground point cloud data.

[0082] Step S124: Compensate the reflectivity of the ground point cloud data within each grid according to the positional relationship between each grid and the laser sensor.

[0083] Specifically, in this embodiment, after obtaining ground point cloud data, it can be meshed according to the region of interest or a preset grid. Neighborhoods within the same grid have similar incident angles and distances, requiring no compensation. However, grids located at different positions have different incident angles and distances. Therefore, based on the positional relationship between the grid and the laser sensor, the laser incident angle and the distance between the grid and the laser sensor can be determined. Consequently, reflectivity compensation can be applied to different grids based on different laser incident angles and distances between the grid and the laser sensor.

[0084] Furthermore, Figure 2 and Figure 3 The embodiments can also be combined. For example, a reflectivity compensation coefficient can be determined based on environmental parameters; the ground point cloud data can be gridded; the reflectivity compensation coefficient of the ground point cloud data within each grid can be adjusted according to the positional relationship between each grid and the laser sensor; and the reflectivity of the ground point cloud data can be adjusted based on the reflectivity compensation coefficient. This allows for reflectivity compensation while improving its efficiency.

[0085] Specifically, road marking features can include lane line features and ground traffic markings.

[0086] Regarding lane line features, such as Figure 4 , Figure 4 A flowchart illustrating the extraction of lane line features according to an embodiment of the present invention is shown. Figure 4 The following steps are shown:

[0087] Step S125: Use the mincut algorithm to obtain candidate road surface marker point cloud data with reflectivity greater than the set reflectivity threshold.

[0088] Specifically, image segmentation can be viewed as a series of partitioning problems, and graphs can be partitioned in various ways. After mapping an image to a graph, the segmentation problem can be solved using graph theory methods such as minimum cut. The minimum cut algorithm can segment candidate road surface marker point cloud data (high reflectivity point clusters) with high reflectivity thresholds.

[0089] Step S126: Use the random consensus algorithm to extract lane line features from the candidate road surface marker point cloud data.

[0090] Therefore, lane line features can be extracted by combining reflectivity with the minimum cut algorithm and the random consensus algorithm.

[0091] For ground traffic signs, since they are located on a plane, their height information is invalid. Therefore, this embodiment can use reflectivity to replace height information, thereby realizing the three-dimensional distance of ground point cloud data. As a result, points that are close in distance and have similar reflectivity will be clustered together and can be identified as a ground traffic sign.

[0092] like Figure 5 , Figure 5 A flowchart of ground traffic markings according to an embodiment of the present invention is shown. Figure 5 The following steps are shown:

[0093] Step S127: Using the reflectivity of the ground point cloud data as height information, perform three-dimensional clustering on the ground point cloud data to obtain clustered point clusters.

[0094] Step S128: Obtain the minimum circumscribed convex polygon of each cluster of points.

[0095] Step S129: Determine the pre-classification landmark type of the clustered point clusters based on the minimum circumscribed convex polygon.

[0096] Specifically, the smallest circumscribed convex polygon of a cluster of points reflects its shape information. By comparing it with the shape information of prior landmarks, it can be classified into a preset landmark type.

[0097] Step S1210: Determine the ground traffic markers of the clustered point clusters based on the pre-classified road sign types.

[0098] Specifically, step S1210 can be achieved through descriptor matching, such as... Figure 6 , Figure 6 A flowchart illustrating the association between ground traffic signs and map features according to an embodiment of the present invention is shown. Figure 6 The following steps are shown:

[0099] Step S1211: Generate the first descriptor of the cluster based on other clusters within the set distance range of the cluster and their pre-classified landmark types.

[0100] Step S1212: Generate a second descriptor for the map feature based on other map features within a set distance range of the pre-classified road sign type map feature in the preset map and their corresponding ground traffic signs.

[0101] Step S1213: Determine the ground traffic markers of the clustered point clusters based on the similarity between the first descriptor and the second descriptor.

[0102] Specifically, step S1211 above can generate multiple descriptors for a cluster of points based on a cluster of points and its surrounding clusters. For example, multiple descriptors can be encoded according to the orientation and distance of the cluster of points and its surrounding clusters, in a clockwise / counterclockwise (or other direction) order. Figure 7 As shown, the clusters 202 surrounding the cluster 201 are encoded in a clockwise direction to obtain the first descriptors (descriptor 1 to descriptor 5).

[0103] Correspondingly, in step S1212, the second descriptor of the map feature can be obtained in the preset map in a similar manner to step S1211 for the map feature of the same landmark type as the cluster point cluster (such as label 201).

[0104] When the similarity between the first descriptor and the second descriptor is greater than a set threshold, it can be considered that the cluster of points in step S1211 corresponds to the map feature, and thus the map feature can be identified as the ground traffic mark of the map feature.

[0105] Therefore, by going through steps S1211 to S1213, the association between clustered point clusters and map features can be established to improve the efficiency of subsequent semantic matching.

[0106] Specifically, due to calibration errors, high reflectivity noise, and stitching errors, geometric features in point clouds often exhibit ghosting and noise-encircled appearances. These effects reduce the accuracy of geometric feature extraction. Therefore, when extracting 3D geometric features from the ground point cloud data in step S130, preprocessing of the ground point cloud data is performed first. Preprocessing may include, but is not limited to, smoothing, denoising, and downsampling operations on the ground point cloud data. Furthermore, the Laplacian smoothing algorithm can be used to utilize the local features of the ground point cloud data to achieve skeletalization of the ground point cloud data based on denoising. Skeletalization is used to identify objects using important lines, reducing redundant data.

[0107] See below. Figure 8 , Figure 8 A flowchart illustrating the extraction of three-dimensional geometric features from the ground point cloud data according to an embodiment of the present invention is shown. Figure 8 The following steps are shown:

[0108] Step S131: Perform principal component analysis on the ground point cloud data to extract the feature vectors and feature values ​​of the ground point cloud data, so as to classify the ground point cloud data into different geometric feature categories.

[0109] Specifically, the principal component analysis algorithm is applied to the neighborhood of each point in the ground point cloud data to analyze its spatial distribution characteristics. This allows us to find the direction with the largest / smallest variance, remove it, and then iterate to find several orthogonal eigenvectors and their corresponding eigenvalues.

[0110] Step S132: Cluster the ground point cloud data of the same geometric feature category to obtain three-dimensional geometric features.

[0111] Specifically, step S132 can employ a conditional Euclidean space clustering algorithm to cluster ground point cloud data of the same geometric feature category to obtain three-dimensional geometric features. Thus, for different geometric feature types, the growth direction of their points can be restricted.

[0112] See below. Figure 9 , Figure 9 A flowchart illustrating the process of performing principal component analysis on the ground point cloud data according to an embodiment of the present invention to extract the feature vectors and feature values ​​of the ground point cloud data, and to classify the ground point cloud data into different geometric feature categories. Figure 9 The following steps are shown:

[0113] Step S133: Based on the above ground point cloud data, construct a point cloud kd-tree ((k-dimensional tree));

[0114] Step S134: Traverse the point cloud kd tree and construct the covariance matrix of the coordinate distribution of each point in the neighborhood of the point cloud kd tree;

[0115] Step S135: Obtain the eigenvalues ​​and eigenvectors of the covariance matrix;

[0116] Step S136: Determine the geometric feature category of the points in the ground point cloud data based on the feature values ​​and feature vectors (or discard them).

[0117] Therefore, the above steps can improve the efficiency of determining the geometric feature category to which points in ground point cloud data belong.

[0118] The above are merely several specific implementations of the point cloud feature matching method of the present invention. Each implementation can be implemented independently or in combination, and the present invention is not intended to limit it. Furthermore, the flowchart of the present invention is merely illustrative, and the execution order between the steps is not limited thereto. The splitting, merging, sequential exchange, and other synchronous or asynchronous execution methods of the steps are all within the protection scope of the present invention.

[0119] The present invention also provides a point cloud feature matching device. Figure 10A block diagram of a point cloud feature matching device according to an embodiment of the present invention is shown. The point cloud feature matching device 300 includes a point cloud segmentation module 310, a ground feature extraction module 320, a ground feature extraction module 330, and a semantic matching module 340.

[0120] The point cloud segmentation module 310 is used to divide the point cloud data collected by the laser sensor into ground point cloud data and ground point cloud data;

[0121] Ground feature extraction module 320 is used to extract road marking features from the ground point cloud data based on reflectance;

[0122] The ground feature extraction module 330 is used to extract three-dimensional geometric features from the ground point cloud data;

[0123] The semantic matching module 340 is used to match the road marking features and three-dimensional geometric features with the map features of the preset map to obtain semantic matching features.

[0124] In the point cloud feature matching device provided by the present invention, point cloud data is divided into ground point cloud data and above-ground point cloud data. Road marking features are extracted from the ground point cloud data based on reflectivity, and three-dimensional geometric features are extracted from the above-ground point cloud data. The road marking features and three-dimensional geometric features are then matched with map features of a preset map to obtain semantic matching features. Thus, feature extraction of road marking features and three-dimensional geometric features can be performed based on the divided point cloud data, improving feature extraction performance. At the same time, extracting road marking features based on reflectivity improves the accuracy of road feature extraction, thereby improving the accuracy of vehicle positioning. This device is particularly suitable for port scenarios.

[0125] Figure 10 The point cloud feature matching device 300 provided by this invention is merely illustrated schematically. Without departing from the inventive concept, the splitting, merging, and addition of modules are all within the scope of protection of this invention. The point cloud feature matching device 300 provided by this invention can be implemented by software, hardware, firmware, plugins, and any combination thereof; this invention is not limited thereto.

[0126] In exemplary embodiments of this disclosure, a computer-readable storage medium is also provided, having stored thereon a computer program that, when executed by, for example, a processor, can implement the steps of the point cloud feature matching method described in any of the above embodiments. In some possible implementations, various aspects of the invention can also be implemented as a program product comprising program code that, when run on a terminal device, causes the terminal device to perform the steps of the various exemplary embodiments of the invention described in the point cloud feature matching method section of this specification.

[0127] refer to Figure 11 As shown, a program product 800 for implementing the above-described method according to an embodiment of the present invention is described. This product may employ a portable compact disc read-only memory (CD-ROM) and include program code, and may run on a terminal device, such as a personal computer. However, the program product of the present invention is not limited thereto. In this document, the readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0128] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0129] The computer-readable storage medium may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The readable storage medium may also be any readable medium other than a readable storage medium, capable of transmitting, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0130] Program code for performing the operations of this invention can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the tenant's computing device, partially on the tenant's device, as a standalone software package, partially on the tenant's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the tenant's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0131] In exemplary embodiments of this disclosure, an electronic device is also provided, which may include a processor and a memory for storing executable instructions of the processor. The processor is configured to perform the steps of the point cloud feature matching method described in any of the above embodiments by executing the executable instructions.

[0132] Those skilled in the art will understand that various aspects of the present invention can be implemented as systems, methods, or program products. Therefore, various aspects of the present invention can be specifically implemented in the following forms: entirely hardware implementations, entirely software implementations (including firmware, microcode, etc.), or implementations combining hardware and software aspects, collectively referred to herein as “circuits,” “modules,” or “systems.”

[0133] The following reference Figure 12 To describe an electronic device 600 according to this embodiment of the present invention. Figure 12 The electronic device 600 shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0134] like Figure 12 As shown, the electronic device 600 is presented in the form of a general-purpose computing device. The components of the electronic device 600 may include, but are not limited to: at least one processing unit 610, at least one storage unit 620, a bus 630 connecting different system components (including storage unit 620 and processing unit 610), a display unit 640, etc.

[0135] The storage unit stores program code that can be executed by the processing unit 610, causing the processing unit 610 to perform the steps described in the point cloud feature matching method section of this specification, according to various exemplary embodiments of the present invention. For example, the processing unit 610 can perform actions such as... Figure 1The steps are shown in the figure.

[0136] The storage unit 620 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 6201 and / or a cache storage unit 6202, and may further include a read-only memory unit (ROM) 6203.

[0137] The storage unit 620 may also include a program / utility 6204 having a set (at least one) program module 6205, such program module 6205 including but not limited to: an operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0138] Bus 630 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0139] Electronic device 600 can also communicate with one or more external devices 700 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable tenants to interact with electronic device 600, and / or with any device that enables electronic device 600 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 650. Furthermore, electronic device 600 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 660. Network adapter 660 can communicate with other modules of electronic device 600 via bus 630. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0140] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, or network device, etc.) to execute the above-described point cloud feature matching method according to the embodiments of this disclosure.

[0141] Compared with the prior art, the advantages of this invention are:

[0142] By dividing point cloud data into ground point cloud data and above-ground point cloud data, and extracting road marking features from the ground point cloud data and three-dimensional geometric features from the above-ground point cloud data based on reflectivity, the road marking features and three-dimensional geometric features are matched with map features of a preset map to obtain semantic matching features. Thus, feature extraction can be performed on road marking features and three-dimensional geometric features based on the divided point cloud data, improving feature extraction performance. At the same time, extracting road marking features based on reflectivity improves the accuracy of road feature extraction, thereby improving the accuracy of vehicle positioning, and is particularly applicable to port scenarios.

[0143] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the appended claims.

Claims

1. A point cloud feature matching method, characterized in that, include: The point cloud data collected by the laser sensor is divided into ground point cloud data and ground point cloud data; Extracting road marking features from ground point cloud data based on reflectivity includes: determining a reflectivity compensation coefficient based on environmental parameters and / or laser sensor parameters; adjusting the reflectivity of the ground point cloud data based on the reflectivity compensation coefficient; using a min-cut algorithm to obtain candidate road marking point cloud data with reflectivity greater than a set reflectivity threshold; using a random consistency algorithm to extract lane line features from the candidate road marking point cloud data; using the reflectivity of the ground point cloud data as height information to perform three-dimensional clustering on the ground point cloud data to obtain clusters; obtaining the minimum bounding convex polygon of each cluster; determining the pre-classified road sign type of the cluster based on the minimum bounding convex polygon; generating a first descriptor for the cluster based on other clusters within a set distance range of the cluster and their pre-classified road sign types; generating a second descriptor for the map features within a set distance range of the map features of the pre-classified road sign type in a preset map and their corresponding ground traffic marks; and determining the ground traffic marks of the cluster based on the similarity between the first descriptor and the second descriptor. Extracting three-dimensional geometric features from the ground point cloud data includes: constructing a kd-tree for the point cloud based on the ground point cloud data; traversing the kd-tree to construct a covariance matrix of the coordinate distribution of each point in the neighborhood of the kd-tree; obtaining the eigenvalues ​​and eigenvectors of the covariance matrix; determining the geometric feature category to which the points in the ground point cloud data belong based on the eigenvalues ​​and eigenvectors; and clustering ground point cloud data of the same geometric feature category to obtain three-dimensional geometric features. The road marking features and three-dimensional geometric features are matched with the map features of a preset map to obtain semantic matching features.

2. The point cloud feature matching method as described in claim 1, characterized in that, The extraction of road marking features from the ground point cloud data based on reflectivity includes: Meshized ground point cloud data; The reflectivity of the ground point cloud data within each grid is compensated based on the positional relationship between each grid and the laser sensor.

3. The point cloud feature matching method as described in claim 1, characterized in that, The process of clustering ground point cloud data of the same geometric feature category to obtain three-dimensional geometric features includes: A conditional Euclidean space clustering algorithm is used to cluster ground point cloud data of the same geometric feature category to obtain three-dimensional geometric features.

4. A point cloud feature matching device, characterized in that, include: The point cloud segmentation module is used to divide the point cloud data collected by the laser sensor into ground point cloud data and ground point cloud data; The ground feature extraction module extracts road marking features from the ground point cloud data based on reflectance, including: determining a reflectance compensation coefficient based on environmental parameters and / or laser sensor parameters; adjusting the reflectance of the ground point cloud data based on the reflectance compensation coefficient; using a min-cut algorithm to obtain candidate road marking point cloud data with reflectance greater than a set reflectance threshold; using a random consistency algorithm to extract lane line features from the candidate road marking point cloud data; using the reflectance of the ground point cloud data as height information to perform three-dimensional clustering on the ground point cloud data to obtain clusters; obtaining the minimum bounding convex polygon of each cluster; determining the pre-classified road sign type of the cluster based on the minimum bounding convex polygon; generating a first descriptor for the cluster based on other clusters within a set distance range of the cluster and their pre-classified road sign types; generating a second descriptor for the map features within a set distance range of the map features of the pre-classified road sign type in a preset map and their corresponding ground traffic signs; and determining the ground traffic signs of the cluster based on the similarity between the first descriptor and the second descriptor. The ground feature extraction module extracts three-dimensional geometric features from the ground point cloud data, including: constructing a point cloud kd-tree based on the ground point cloud data; traversing the point cloud kd-tree; constructing a covariance matrix of the coordinate distribution of each point in the neighborhood of the point cloud kd-tree; obtaining the eigenvalues ​​and eigenvectors of the covariance matrix; determining the geometric feature category to which the points in the ground point cloud data belong based on the eigenvalues ​​and eigenvectors; and clustering ground point cloud data of the same geometric feature category to obtain three-dimensional geometric features. The semantic matching module is used to match the road marking features and three-dimensional geometric features with the map features of a preset map to obtain semantic matching features.

5. An electronic device, characterized in that, The electronic device includes: processor; A storage medium having a computer program stored thereon, the computer program being executed by the processor to perform the point cloud feature matching method as described in any one of claims 1 to 3.

6. A storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, performs the point cloud feature matching method as described in any one of claims 1 to 3.

Images