Point cloud data processing method and apparatus, vehicle, and storage medium

Abstract

The application relates to a point cloud data processing method and device, a vehicle and a storage medium, which comprise the following steps: converting a point cloud file into a rosbag file, labeling the rosbag file according to preset labeling requirements to obtain a labeled rosbag file; merging the labeled rosbag files with the same theme in the labeled rosbag file into a label file, and generating a new rosbag file from the point cloud file and the label file; extracting three-dimensional coordinates and point cloud intensity from the new rosbag file, writing the three-dimensional coordinates and the point cloud intensity into a bin file according to a first preset frame number, writing the labeling information of the label file into a txt file according to a second preset frame number in sequence, outputting the bin file and the txt file, and displaying the point cloud file through the bin file and the txt file. The application solves the problems in the related art that the point cloud data processing method cannot be used for laser radar point cloud deep learning training, the data result is not accurate, and the like, and improves the accuracy of data.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, and in particular to a point cloud data processing method, apparatus, vehicle, and storage medium. Background Technology

[0002] In autonomous driving systems, environmental perception is a crucial component, providing reliable environmental information for prediction, decision-making, and control, and forming the foundation for autonomous driving. Therefore, in the field of LiDAR perception, accurately processing LiDAR point cloud data and providing the 3D coordinates of each point, the vehicle's heading angle, and the point cloud reflection intensity is paramount. Accurate point cloud information is essential for subsequent point cloud detection and recognition.

[0003] However, in related technologies, point cloud data processing methods cannot be used for deep learning training of LiDAR point clouds, and the data results are inaccurate, which urgently needs to be addressed. Summary of the Invention

[0004] This application provides a point cloud data processing method, apparatus, vehicle, and storage medium to solve the problems in related technologies, such as the inability of point cloud data processing methods to be used for deep learning training of LiDAR point clouds and the inaccuracy of data results, thereby improving the accuracy of data processing.

[0005] The first aspect of this application provides a point cloud data processing method, comprising the following steps: acquiring a point cloud file collected by a target radar, converting the point cloud file into a rosbag file, and annotating the rosbag file according to preset annotation requirements to obtain multiple annotated rosbag files; merging the annotated rosbag files with the same theme into a tag file, and generating a new rosbag file based on the point cloud file and the tag file; based on the new rosbag file, extracting the three-dimensional coordinates and point cloud intensity of each point cloud data in the point cloud file, writing the three-dimensional coordinates and point cloud intensity of each point cloud data into a bin file according to a first preset frame number, and writing the annotation information of the tag file into a txt file according to a preset order according to a second preset frame number, and outputting the bin file and the txt file to display the point cloud file through the bin file and the txt file.

[0006] Based on the above technical means, the problems that point cloud data processing methods in related technologies cannot be used for deep learning training of LiDAR point clouds and the data results are inaccurate have been solved, thereby improving the accuracy of data processing.

[0007] Furthermore, before outputting the bin file and the txt file, the method further includes: reading the bin file to obtain the three-dimensional coordinates and point cloud intensity to be verified for each point cloud data in the bin file; reading the txt file and calculating the coordinates of multiple vertices corresponding to the annotation box to be verified based on a preset transformation matrix and the length, width, and height of a preset annotation box, and drawing the annotation box to be verified based on the multiple vertices coordinates; verifying the three-dimensional coordinates to be verified, the point cloud intensity to be verified, and the annotation box to be verified respectively, and generating the verification results of the bin file and the txt file.

[0008] Based on the aforementioned technical means, vertex coordinates are obtained by reading bin and txt files and analyzing the data from the annotation boxes. This facilitates input into visualization software for data comparison and verification of the accuracy of the converted data.

[0009] Furthermore, the step of writing the annotation information of the tag file into the txt file according to a preset order and a second preset number of frames includes: writing the annotation information of the tag file into the txt file according to the annotation type, object truncation degree, object occlusion degree, object viewing angle, position of 2D annotation box, size of annotation box, three-dimensional coordinates of the center point of annotation box, heading of annotation box and confidence level of annotation box, according to the second preset number of frames.

[0010] Based on the above technical means, by writing the annotation information of the label file into a txt file, it is convenient to calculate the vertex coordinates of the annotation box through the txt file.

[0011] Further, generating a new rosbag file based on the point cloud file and the tag file includes: obtaining a first topic of the point cloud file and a second topic of the tag file; obtaining topic information and a first timestamp of the first topic, topic information and a second timestamp of the second topic; and merging the point cloud file and the tag file into the new rosbag file based on the topic information of the first topic, the first timestamp, the topic information of the second topic and the second timestamp.

[0012] Using the aforementioned technical methods, point cloud data is converted into rosbag files, which facilitates merging with annotation files.

[0013] A second aspect of this application provides a point cloud data processing apparatus, comprising: an annotation module, configured to acquire point cloud files collected by a target radar, convert the point cloud files into rosbag files, and annotate the rosbag files according to preset annotation requirements to obtain multiple annotated rosbag files; a generation module, configured to merge annotated rosbag files with the same theme into a tag file, and generate a new rosbag file based on the point cloud files and the tag files; and an output module, configured to extract the three-dimensional coordinates and point cloud intensity of each point cloud data in the point cloud files based on the new rosbag files, write the three-dimensional coordinates and point cloud intensity of each point cloud data into a bin file according to a first preset frame number, write the annotation information of the tag files into a txt file according to a preset order according to a second preset frame number, and output the bin file and the txt file to display the point cloud files through the bin file and the txt file.

[0014] Further, before outputting the bin file and the txt file, the output module is specifically used for: reading the bin file to obtain the three-dimensional coordinates and point cloud intensity to be verified for each point cloud data in the bin file; reading the txt file and calculating the coordinates of multiple vertices corresponding to the annotation box to be verified according to a preset transformation matrix and the length, width, and height of a preset annotation box, and drawing the annotation box to be verified according to the multiple vertices coordinates; verifying the three-dimensional coordinates to be verified, the point cloud intensity to be verified, and the annotation box to be verified respectively, and generating the verification results of the bin file and the txt file.

[0015] Furthermore, the output module is also used to: write the annotation information of the tag file into the txt file according to the annotation type, object truncation degree, object occlusion degree, object viewing angle, position of 2D annotation box, size of annotation box, three-dimensional coordinates of the center point of annotation box, heading of annotation box and confidence level of annotation box according to the second preset frame number.

[0016] Further, the generation module is specifically used to: obtain a first topic of the point cloud file and a second topic of the tag file; obtain topic information and a first timestamp of the first topic, topic information and a second timestamp of the second topic; and merge the point cloud file and the tag file into the new rosbag file according to the topic information of the first topic, the first timestamp, the topic information of the second topic and the second timestamp.

[0017] A third aspect of this application provides a vehicle, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the point cloud data processing method as described in the above embodiments.

[0018] A fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement the point cloud data processing method as described in the above embodiments.

[0019] Therefore, by acquiring point cloud files collected by the target radar, converting them into ROSBA files, and annotating them according to preset annotation requirements, multiple annotated ROSBA files are obtained. Annotated ROSBA files with the same theme are merged into a label file. A new ROSBA file is generated based on the point cloud file and the label file. Based on the new ROSBA file, the 3D coordinates and point cloud intensity of each point cloud data point are extracted. The 3D coordinates and point cloud intensity of each point cloud data point are written to a BIN file according to a first preset frame number. The annotation information from the label file is written to a TXT file according to a second preset frame number, and the BIN and TXT files are output to display the point cloud file. This solves the problems in related technologies where point cloud data processing cannot be used for deep learning training of LiDAR point clouds, and the data results are inaccurate, thus improving the accuracy of data processing.

[0020] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0021] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0022] Figure 1 This is a flowchart of a point cloud data processing method provided according to an embodiment of this application;

[0023] Figure 2 This is a schematic diagram of a lidar point cloud according to an embodiment of this application;

[0024] Figure 3 This is a schematic diagram of a lidar point cloud and point cloud annotation according to an embodiment of this application;

[0025] Figure 4 This is a schematic diagram illustrating a label file format according to an embodiment of this application;

[0026] Figure 5 This is a schematic diagram of label file annotation information according to an embodiment of this application;

[0027] Figure 6 This is a schematic diagram of a heading angle according to an embodiment of this application;

[0028] Figure 7 This is a flowchart of a point cloud data processing method according to an embodiment of this application;

[0029] Figure 8 This is a block diagram of a point cloud data processing apparatus according to an embodiment of this application;

[0030] Figure 9 This is a structural schematic diagram of a vehicle according to an embodiment of this application.

[0031] Explanation of reference numerals in the attached figures: 10 - Point cloud data processing method device, 100 - Annotation module, 200 - Generation module, 300 - Output module. Detailed Implementation

[0032] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0033] The following describes a point cloud data processing method, apparatus, vehicle, and storage medium according to embodiments of this application with reference to the accompanying drawings. Addressing the issues mentioned in the background art, such as the inability of point cloud data processing to be used for deep learning training of LiDAR point clouds and the inaccuracy of data results, this application provides a point cloud data processing method. In this method, point cloud files collected by a target radar are acquired and converted into rosbag files. These rosbag files are then annotated according to preset annotation requirements, resulting in multiple annotated rosbag files. Annotated rosbag files with the same theme are merged into a tag file. A new rosbag file is generated based on the point cloud file and the tag file. Based on the new rosbag file, the 3D coordinates and point cloud intensity of each point cloud data point are extracted from the point cloud file. The 3D coordinates and point cloud intensity of each point cloud data point are written to a bin file at a first preset frame count. The annotation information of the tag file is written to a txt file at a second preset frame count according to a preset order. The bin file and txt file are then output to display the point cloud file. This solves the problems in related technologies, such as the inability of point cloud data processing to be used for deep learning training of LiDAR point clouds and the inaccuracy of data results, thus improving the accuracy of data processing.

[0034] Specifically, Figure 1 This is a flowchart illustrating a point cloud data processing method provided in an embodiment of this application.

[0035] like Figure 1 As shown, the point cloud data processing method includes the following steps:

[0036] In step S101, the point cloud file collected by the target radar is acquired, and the point cloud file is converted into a rosbag file. The rosbag file is then annotated according to the preset annotation requirements to obtain multiple annotated rosbag files. The rosbag file is mainly used for recording, playback, and analysis of data.

[0037] Specifically, such as Figure 2 The diagram illustrates the LiDAR point cloud process. Point cloud files are collected from the target radar, and the target radar .idc point cloud files are converted into corresponding ROSBACK files every 500 frames. Point cloud files corresponding to the relevant scenes are selected and labeled according to the collected scene and training requirements for subsequent deep learning training. A schematic diagram of the LiDAR point cloud and its annotation is shown below. Figure 3 As shown.

[0038] In step S102, multiple rosbag files with the same theme are merged into a label file, and a new rosbag file is generated based on the point cloud file and the label file.

[0039] Optionally, in some embodiments, generating a new rosbag file based on the point cloud file and the tag file includes: obtaining a first topic from the point cloud file and a second topic from the tag file; obtaining topic information and a first timestamp of the first topic, topic information and a second timestamp of the second topic; and merging the point cloud file and the tag file into a new rosbag file based on the topic information, the first timestamp, the topic information and the second timestamp of the second topic.

[0040] Specifically, based on the rosbag folder directory provided in the configuration file, the rosbag.Bag function reads the contents of the rosbag files sequentially. Simultaneously, it retrieves point cloud information, including msg type and topics, using get_type_and_topic_info(). Then, it reads the contents of the bag using read_messages(), which returns an iterator. Each iteration returns three values: topic, msg, and t, where msg is the actual data and t is the timestamp. Multiple labeled rosbag files with the same topic are merged into a single label file. The label file format is described below. Figure 4 As shown.

[0041] Further, the point cloud file and tag file are read. Based on the topics in the point cloud file and tag file, the two topics are merged into a rosbag file by reading the topic information bag.read_messages() and reading the timestamp resBag.write(), and a new tag rosbag file is generated.

[0042] In step S103, based on the new rosbag file, the three-dimensional coordinates and point cloud intensity of each point cloud data in the point cloud file are extracted, and the three-dimensional coordinates and point cloud intensity of each point cloud data are written into the bin file according to the first preset frame number. The annotation information of the label file is written into the txt file according to the second preset frame number in the preset order, and the bin file and txt file are output to display the point cloud file through the bin file and txt file.

[0043] Specifically, the X, Y, and Z 3D coordinates and intensity of each point in the point cloud file and the new label rosbag file are extracted. The merged point cloud file and label file are displayed using the visualization software RViz. When the topic is / ibeo_pointcloud, the information under the topic is read, the data type is converted to np.float32, the height of the vehicle body (i.e., the installation height of the target radar) is subtracted from the Z coordinate of the point cloud, and then the intensity is normalized. Finally, the data is written to the bin file frame by frame.

[0044] Furthermore, in some embodiments, the annotation information of the label file is written to a txt file according to a preset order and a second preset number of frames, including: writing the annotation information of the label file to a txt file according to the annotation type, object truncation degree, object occlusion degree, object viewing angle, position of 2D annotation box, size of annotation box, three-dimensional coordinates of the center point of annotation box, heading of annotation box and confidence level of annotation box, according to a second preset number of frames.

[0045] Specifically, the label file annotation information is as follows: Figure 5 As shown, the annotation information is written to a txt file frame by frame according to the annotation type, object truncation degree, object occlusion degree, object viewing angle alpha, 2D annotation box position bbox, annotation box size, 3D coordinates of the annotation box center point location, annotation box yaw, and annotation box confidence score.

[0046] Furthermore, the merged rosbag file is read. When the topic is / ld_object_lists, the label boxes are classified according to their size. The label box types are: cone (traffic cone), pedestrian (pedestrian), bicycle (bicycle), electric bicycle (electric bicycle), motorcycle (motorcycle), tricycle (tricycle), vehicle (car), big vehicle (large vehicle), and huge vehicle (oversized vehicle).

[0047] The center point of the annotation box is the rear axle center point of the vehicle. The yaw angle is also calculated; it is defined as the yaw angle of the annotation box. The vehicle's forward direction relative to the X-axis of the point cloud coordinate system is positive counterclockwise and negative clockwise. A schematic diagram of the yaw angle is shown below. Figure 6 As shown.

[0048] Furthermore, in some embodiments, before outputting the bin file and txt file, the method further includes: reading the bin file to obtain the three-dimensional coordinates and point cloud intensity to be verified for each point cloud data in the bin file; reading the txt file and calculating the coordinates of multiple vertices corresponding to the annotation box to be verified based on a preset transformation matrix and the length, width, and height of a preset annotation box, and drawing the annotation box to be verified based on the multiple vertex coordinates; verifying the three-dimensional coordinates to be verified, the point cloud intensity to be verified, and the annotation box to be verified respectively, and generating the verification results of the bin file and txt file.

[0049] Specifically, by reading the bin file, the 3D coordinates and point cloud intensity of points in the point cloud are obtained; by reading the txt point cloud annotation file, the transformation matrix is ​​used to... The coordinates of the eight vertices corresponding to the annotation box are calculated from the length, width, and height of the annotation box. This coordinates are needed for drawing the annotation box in 3D visualization software. At the same time, different colors are assigned to the annotation box according to its type for easy differentiation. The results are compared with the data before the conversion to verify the accuracy of the converted data.

[0050] To enable those skilled in the art to further understand the point cloud data processing method of the embodiments of this application, the following is combined with... Figure 7 Specific embodiments will be described in detail.

[0051] Step S701: Input the point cloud file acquired by the target radar.

[0052] Step S702: Convert the target radar .idc point cloud file into rosbag format.

[0053] Step S703: Select point cloud scenarios and annotate the point cloud files according to actual needs.

[0054] Step S704: Merge multiple annotated rosbag files into one file based on the same topic.

[0055] Step S705: Merge the point cloud file and the label file into a new rosbag file.

[0056] Step S706: Extract the X, Y, Z 3D coordinates and intensity of each point in the point cloud file, and write them to the bin file frame by frame.

[0057] Step S707: Extract the annotation information from the annotation file and write it to a txt file frame by frame according to information such as annotation type, object truncation degree, and object occlusion degree.

[0058] Step S708: Write a visualization program to display the point cloud and annotation information, and verify the accuracy of the converted data.

[0059] Step S709: Output the converted point cloud bin file and the annotation txt file.

[0060] Therefore, the point cloud data processing method proposed in this application involves acquiring point cloud files collected by the target radar, converting the point cloud files into rosbag files, and annotating the rosbag files according to preset annotation requirements to obtain multiple annotated rosbag files. Annotated rosbag files with the same theme are merged into a tag file. A new rosbag file is generated based on the point cloud file and the tag file. Based on the new rosbag file, the 3D coordinates and point cloud intensity of each point cloud data point are extracted from the point cloud file. The 3D coordinates and point cloud intensity of each point cloud data point are written to a bin file according to a first preset frame number. The annotation information of the tag file is written to a txt file according to a second preset frame number in a preset order. The bin file and txt file are then output to display the point cloud file. This solves the problems in related technologies where point cloud data processing cannot be used for deep learning training of target radar point clouds and the data results are inaccurate, thus improving the accuracy of data processing.

[0061] Next, the point cloud data processing apparatus proposed according to the embodiments of this application is described with reference to the accompanying drawings.

[0062] Figure 8 This is a block diagram of a point cloud data processing device according to an embodiment of this application.

[0063] like Figure 8 As shown, the point cloud data processing device 10 includes: a labeling module 100, a generation module, and an output module 300.

[0064] The annotation module 100 is used to acquire point cloud files collected by the target radar, convert the point cloud files into rosbag files, and annotate the rosbag files according to preset annotation requirements to obtain multiple annotated rosbag files; the generation module 200 is used to merge the annotated rosbag files with the same theme into a tag file, and generate a new rosbag file based on the point cloud file and the tag file; the output module 300 is used to extract the three-dimensional coordinates and point cloud intensity of each point cloud data in the point cloud file based on the new rosbag file, write the three-dimensional coordinates and point cloud intensity of each point cloud data into a bin file according to a first preset frame number, and write the annotation information of the tag file into a txt file according to a second preset frame number in a preset order, and output the bin file and txt file to display the point cloud file through the bin file and txt file.

[0065] Optionally, in some embodiments, before outputting the bin file and txt file, the output module 300 is specifically used to: read the bin file to obtain the three-dimensional coordinates to be verified and the intensity of the point cloud data to be verified for each point cloud data in the bin file; read the txt file and calculate the coordinates of multiple vertices corresponding to the annotation box to be verified according to the preset transformation matrix and the preset length, width and height of the annotation box, and draw the annotation box to be verified according to the multiple vertices coordinates; verify the three-dimensional coordinates to be verified, the intensity of the point cloud to be verified and the annotation box to be verified respectively, and generate the verification results of the bin file and txt file.

[0066] Optionally, in some embodiments, the output module 300 is further configured to: write the annotation information of the label file into a txt file according to the annotation type, object truncation degree, object occlusion degree, object viewing angle, position of 2D annotation box, size of annotation box, three-dimensional coordinates of the center point of annotation box, heading of annotation box and confidence level of annotation box according to a second preset number of frames.

[0067] Optionally, in some embodiments, the generation module 100 is specifically used to: obtain a first topic of the point cloud file and a second topic of the tag file; obtain topic information and a first timestamp of the first topic, topic information and a second timestamp of the second topic; and merge the point cloud file and the tag file into a new rosbag file according to the topic information, the first timestamp, the topic information and the second timestamp of the second topic.

[0068] It should be noted that the foregoing explanation of the point cloud data processing method embodiment also applies to the point cloud data processing device of this embodiment, and will not be repeated here.

[0069] The point cloud data processing apparatus proposed in this application acquires point cloud files collected by a target radar, converts the point cloud files into rosbag files, and annotates the rosbag files according to preset annotation requirements to obtain multiple annotated rosbag files. Annotated rosbag files with the same theme are merged into a tag file. A new rosbag file is generated based on the point cloud file and the tag file. Based on the new rosbag file, the 3D coordinates and point cloud intensity of each point cloud data in the point cloud file are extracted. The 3D coordinates and point cloud intensity of each point cloud data are written to a bin file according to a first preset frame number. The annotation information of the tag file is written to a txt file according to a second preset frame number in a preset order. The bin file and txt file are then output to display the point cloud file. This solves the problems in related technologies where point cloud data processing cannot be used for deep learning training of target radar point clouds and the data results are inaccurate, thus improving the accuracy of data processing.

[0070] Figure 9 A schematic diagram of the structure of a vehicle provided in an embodiment of this application. The vehicle may include:

[0071] The memory 901, the processor 902, and the computer program stored on the memory 901 and capable of running on the processor 902.

[0072] When the processor 902 executes the program, it implements the point cloud data processing method provided in the above embodiments.

[0073] Furthermore, the vehicle also includes:

[0074] Communication interface 903 is used for communication between memory 901 and processor 902.

[0075] The memory 901 is used to store computer programs that can run on the processor 902.

[0076] The memory 901 may include high-speed RAM (Random Access Memory) memory, and may also include non-volatile memory, such as at least one disk storage.

[0077] If the memory 901, processor 902, and communication interface 903 are implemented independently, then the communication interface 903, memory 901, and processor 902 can be interconnected via a bus to complete communication between them. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 9 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0078] Optionally, in a specific implementation, if the memory 901, processor 902, and communication interface 903 are integrated on a single chip, then the memory 901, processor 902, and communication interface 903 can communicate with each other through an internal interface.

[0079] The processor 902 may be a CPU (Central Processing Unit), an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of this application.

[0080] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the point cloud data processing method described above.

[0081] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0082] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0083] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0084] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (FPGAs), field-programmable gate arrays (FPGAs), etc.

[0085] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0086] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A point cloud data processing method, characterized in that, Includes the following steps: The point cloud file collected by the target radar is acquired, and the point cloud file is converted into a rosbag file. The rosbag file is then annotated according to the preset annotation requirements to obtain multiple annotated rosbag files. Multiple annotated rosbag files with the same theme are merged into a tag file, and a new rosbag file is generated based on the point cloud file and the tag file. The merged rosbag file is read, and when the topic is / ld_object_lists, the tag boxes are classified according to their size. Based on the new rosbag file, the 3D coordinates and point cloud intensity of each point cloud data in the point cloud file are extracted. The height of the vehicle body, i.e., the installation height of the target radar, is subtracted from the Z coordinate of the point cloud, and then the intensity of the point cloud is normalized. The 3D coordinates and point cloud intensity of each point cloud data are written to a bin file according to a first preset frame number, and the annotation information of the tag file is written to a txt file according to a second preset frame number in a preset order. The center point of the annotation box is the rear axle center point of the vehicle. The yaw angle is calculated, and the yaw angle is defined as the yaw angle of the annotation box. The forward direction of the vehicle relative to the X-axis of the point cloud coordinate system is positive counterclockwise and negative clockwise. The bin file and the txt file are output to display the point cloud file. Before outputting the bin file and the txt file, the following is also included: Read the bin file to obtain the three-dimensional coordinates and point cloud intensity to be verified for each point cloud data in the bin file; Read the txt file, calculate the coordinates of multiple vertices corresponding to the annotation box to be verified according to the preset transformation matrix and the preset length, width and height of the annotation box, and draw the annotation box to be verified according to the multiple vertex coordinates; The three-dimensional coordinates to be verified, the point cloud intensity to be verified, and the bounding box to be verified are verified respectively, and the verification results of the bin file and the txt file are generated. The step of generating a new rosbag file based on the point cloud file and the label file includes: Obtain the first topic of the point cloud file and the second topic of the tag file; Obtain the topic information and first timestamp of the first topic, and the topic information and second timestamp of the second topic; The point cloud file and the tag file are merged into a new rosbag file based on the topic information of the first topic, the first timestamp, the topic information of the second topic, and the second timestamp.

2. The method according to claim 1, characterized in that, The step of writing the annotation information of the tag file into a txt file according to a preset order and a second preset number of frames includes: The annotation information of the label file is written into the txt file according to the annotation type, object truncation degree, object occlusion degree, object viewing angle, position of 2D annotation box, size of annotation box, three-dimensional coordinates of the center point of annotation box, heading of annotation box and confidence level of annotation box, according to the second preset frame number.

3. A point cloud data processing device, characterized in that, For performing the point cloud data processing method as described in any one of claims 1-2, comprising: The annotation module is used to acquire point cloud files collected by the target radar, convert the point cloud files into rosbag files, and annotate the rosbag files according to preset annotation requirements to obtain multiple annotated rosbag files. The generation module is used to merge the multiple labeled rosbag files with the same theme into a label file, and generate a new rosbag file based on the point cloud file and the label file; The output module is used to extract the three-dimensional coordinates and point cloud intensity of each point cloud data in the point cloud file based on the new rosbag file, write the three-dimensional coordinates and point cloud intensity of each point cloud data into a bin file according to a first preset frame number, write the annotation information of the label file into a txt file according to a second preset frame number in a preset order, and output the bin file and the txt file to display the point cloud file through the bin file and the txt file; Before outputting the bin file and the txt file, the output module is specifically used for: Read the bin file to obtain the three-dimensional coordinates and point cloud intensity to be verified for each point cloud data in the bin file; Read the txt file, calculate the coordinates of multiple vertices corresponding to the annotation box to be verified according to the preset transformation matrix and the preset length, width and height of the annotation box, and draw the annotation box to be verified according to the multiple vertex coordinates; The three-dimensional coordinates to be verified, the point cloud intensity to be verified, and the bounding box to be verified are verified respectively, and the verification results of the bin file and the txt file are generated. The generation module is specifically used for: Obtain the first topic of the point cloud file and the second topic of the tag file; Obtain the topic information and first timestamp of the first topic, and the topic information and second timestamp of the second topic; The point cloud file and the tag file are merged into a new rosbag file based on the topic information of the first topic, the first timestamp, the topic information of the second topic, and the second timestamp.

4. The apparatus according to claim 3, characterized in that, The output module is also used for: The annotation information of the label file is written into the txt file according to the annotation type, object truncation degree, object occlusion degree, object viewing angle, position of 2D annotation box, size of annotation box, three-dimensional coordinates of the center point of annotation box, heading of annotation box and confidence level of annotation box, according to the second preset frame number.

5. A vehicle, characterized in that, Including memory and processor; The processor reads executable program code stored in the memory to run a program corresponding to the executable program code, so as to implement the self-point cloud data processing method as described in any one of claims 1-2.

6. A computer-readable storage medium storing a computer program, characterized in that, When executed by a processor, the program implements the point cloud data processing method as described in any one of claims 1-2.

Images