Method and device for processing point cloud
By resampling and registering point clouds and images in autonomous driving cars, a target point cloud with true colors is generated, which solves the problem of point cloud color deviation and improves the driver's operation refinement and traffic safety.
Patent Information
- Application Number
- CN201980008350.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2019-07-30
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2039-07-30
AI Technical Summary
In existing autonomous vehicles, the color presentation of point clouds is largely different from the color of the real environment, making it difficult for drivers to intuitively understand the surrounding environment through visual means, affecting the refined decisions of operations such as reversing, lane change and overtaking.
By acquiring the point cloud generated by the lidar and the images captured by the camera, resampling and registration, the target point cloud containing colors is generated based on the image processing, ensuring that the colors reflect the real driving environment.
It provides a more intuitive visual point cloud, helping drivers better perform operations such as reversing, lane change and overtaking, and improves drivers' decision-making ability and traffic safety.
Smart Images

Figure CN111630520B_ABST
Abstract
Description
[0001] Copyright Notice
[0002] The disclosure of this patent document contains material that is subject to copyright protection. The copyright is reserved by the copyright owner. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure as it exists in the official records and files of the Patent and Trademark Office. Technical Field
[0003] The present application relates to the field of autonomous driving, and more specifically, to a method and device for processing point clouds. Background Art
[0004] Point cloud is a representation of a three-dimensional object or a three-dimensional scene. It is composed of a group of irregularly distributed discrete points in space. The discrete points are point cloud data. Point cloud data is used to characterize the spatial structure of a three-dimensional object or a three-dimensional scene. A point cloud data usually consists of location information.
[0005] In order to help the driver understand the surrounding environment, the self-driving car can present the point cloud to the driver, for example, by displaying the point cloud on a screen or projecting the point cloud on the car window. In order to enable the driver to understand the surrounding environment more intuitively, one method of presenting the point cloud is to color the point cloud before presenting it to the driver. However, the color of the point cloud obtained by directly coloring the point cloud has a large deviation from the color of the real environment. Summary of the invention
[0006] The present application provides a method for processing point clouds, which can provide drivers with more intuitive visual point clouds.
[0007] In a first aspect, a method for processing a point cloud is provided, comprising: obtaining a first point cloud; obtaining an image corresponding to the same environment as the first point cloud; resampling the first point cloud to obtain a second point cloud, wherein the density of the second point cloud is less than the density of the first point cloud; registering the second point cloud and the image; and processing the second point cloud according to the image to generate a target point cloud containing color.
[0008] Since the color of the image taken by the camera can reflect the real driving environment, the above scheme can provide the driver with a more intuitive visual point cloud, which is convenient for the driver to better perform refined operations such as reversing, changing lanes and overtaking according to the target point cloud.
[0009] In a second aspect, a device is provided, wherein the device is used to execute the method in the first aspect.
[0010] According to a third aspect, a device is provided, comprising a memory and a processor, wherein the memory is used to store instructions, and the processor is used to execute the instructions stored in the memory, and the execution of the instructions stored in the memory causes the processor to execute the method of the first aspect.
[0011] In a fourth aspect, a chip is provided, comprising a processing module and a communication interface, wherein the processing module is used to control the communication interface to communicate with the outside, and the processing module is also used to implement the method of the first aspect.
[0012] In a fifth aspect, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a computer, the computer implements the method in the first aspect. Specifically, the computer may be the above-mentioned device.
[0013] In a sixth aspect, a computer program product comprising instructions is provided, wherein when the instructions are executed by a computer, the computer implements the method in the first aspect. Specifically, the computer may be the above-mentioned device. BRIEF DESCRIPTION OF THE DRAWINGS
[0014] Figure 1 is a schematic diagram of a method for processing point clouds provided by the present application;
[0015] Figure 2 is a schematic diagram of another method for processing point clouds provided by the present application;
[0016] Figure 3 It is a schematic diagram of a device for processing point clouds provided in the present application. DETAILED DESCRIPTION
[0017] The technical solutions in the embodiments of the present application will be described below in conjunction with the accompanying drawings.
[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application.
[0019] The point cloud processing method provided in this application can be applied to a mobile platform, including but not limited to an autonomous vehicle. An autonomous vehicle is also called an unmanned vehicle. It can obtain environmental information through sensors such as radar and control the vehicle according to the environmental information, reducing the driver's control time. Therefore, an autonomous vehicle has the advantages of reducing traffic violations.
[0020] However, in the evolution of autonomous driving technology, drivers need to participate in driving to varying degrees. Therefore, presenting the driving environment to the driver through visualization means will help the driver make better decisions such as reversing, changing lanes and overtaking.
[0021] It should be noted that in the present application, the driver can drive the movable platform inside the movable platform, for example, the driver controls the autonomous driving car from inside the autonomous driving car; the driver can also drive the movable platform outside the movable platform, for example, the driver controls a drone or an unmanned submersible through a remote control device.
[0022] To reduce the cost of visualization, the existing sensors of the autonomous vehicle can be used to present the driving environment. For example, the point cloud generated by the lidar can be processed and presented to the driver. Figure 1 This application provides a method for processing point clouds. The method 100 includes:
[0023] S110, obtaining a first point cloud.
[0024] S110 may be executed by a processor, such as a central processing unit (CPU). The CPU may obtain the first point cloud photographed by the laser radar through a communication interface, or the CPU may generate the first point cloud after processing the point cloud photographed by the laser radar. The present application does not limit the specific method of obtaining the first point cloud.
[0025] S110 can also be performed by an autonomous vehicle that includes a processor and a lidar. After the point cloud is captured by the lidar, it is transmitted to the processor in real time so that the processor can perform subsequent steps.
[0026] The above-mentioned generation of the first point cloud by laser radar is only an example, and the present application does not limit the device for generating the point cloud. For example, the first point cloud can also be generated by millimeter wave radar.
[0027] The processor may obtain the first point cloud from one laser radar or from multiple laser radars. For example, the processor obtains multiple point clouds from multiple laser radars and splices the multiple point clouds into the first point cloud; optionally, the processor may register the multiple point clouds (i.e., splice the multiple point clouds) by an iterative closest point (ICP) algorithm to obtain the first point cloud.
[0028] The processor may stitch the multiple point clouds into a first point cloud according to the external parameters of the multiple laser radars. The above external parameters are used to indicate the positional relationship of the coordinate system of each laser radar in the multiple laser radars relative to the coordinate system of the movable platform. For example, the above external parameters may include:
[0029] Rotation parameters of the coordinate system of each of the plurality of laser radars relative to the coordinate system of the movable platform; and / or,
[0030] Translation parameters of the coordinate system of each of the multiple laser radars relative to the coordinate system of the movable platform.
[0031] The processor may execute S120 after executing S110 , or the processor may execute S110 and S120 .
[0032] S120: Acquire an image corresponding to the same environment as the first point cloud.
[0033] For example, the CPU obtains an image captured by a camera through a communication interface. S120 may also be performed by an autonomous driving vehicle including a camera of a processor. The camera may be a color camera or a grayscale camera. Accordingly, the image obtained by the processor may be a color image or a grayscale image.
[0034] The same environment corresponding to the image and the first point cloud can be interpreted as: two results obtained by the camera and the lidar shooting the same scene at the same time; or, two results obtained by the camera and the lidar shooting similar scenes at the same time; or, two results obtained by the camera and the lidar shooting the same scene at adjacent times; or, two results obtained by the camera and the lidar shooting similar scenes at adjacent times.
[0035] For example, the camera captures the lane directly in front of the vehicle at time A to generate an image; the lidar captures the lane directly in front of the vehicle at time A to generate a first point cloud; if the scene corresponding to the image is exactly the same as the scene corresponding to the first point cloud, then the image and the first point cloud correspond to the same environment.
[0036] For another example, a camera captures the lane directly in front of the vehicle at time A to generate an image; a lidar captures the lane in front of the left side of the vehicle at time A to generate a first point cloud; if the scene corresponding to the image is partially identical to the scene corresponding to the first point cloud, then the image and the first point cloud correspond to the same environment.
[0037] For another example, a camera photographs the lane directly in front of a vehicle at time A to generate an image; a lidar photographs the lane directly in front of the vehicle at time B to generate a first point cloud; time A is adjacent to time B, and the scene corresponding to the image is partially identical to the scene corresponding to the first point cloud, then the image and the first point cloud correspond to the same environment.
[0038] For another example, a camera captures the lane directly in front of the vehicle at time A to generate an image; a lidar captures the lane in front of the left side of the vehicle at time B to generate a first point cloud; time A is adjacent to time B, and the scene corresponding to the image is partially identical to the scene corresponding to the first point cloud, then the image and the first point cloud correspond to the same environment.
[0039] In order to obtain images and the first point cloud with high scene similarity, the processor can control the camera and lidar to shoot at the closest possible time and at the same possible angle.
[0040] S130: Resample the first point cloud to obtain a second point cloud, where the density of the second point cloud is less than the density of the first point cloud.
[0041] In order to reduce the load of the processor, the processor may resample the first point cloud to reduce the density of the point cloud blocks with higher density in the first point cloud. The first point cloud and the second point cloud are both colorless point clouds, and the "colorless point cloud" refers to a point cloud without grayscale information and RGB information.
[0042] When the processor resamples the first point cloud, the resampling may be performed according to the following steps:
[0043] Determine a point cloud block in the first point cloud whose point cloud density is greater than a density threshold;
[0044] The point cloud block is resampled.
[0045] There may be some point cloud blocks with high density in the first point cloud. The processor can first identify these point cloud blocks and then resample these point cloud blocks without resampling the entire first point cloud, thereby reducing the burden on the processor.
[0046] S140: Register the second point cloud and the image.
[0047] The processor may align the second point cloud and the image using calibration results of the camera and the lidar, where the calibration results may be preconfigured information, for example, the calibration results may be external parameters between the preconfigured camera and the lidar.
[0048] The processor may also register the second point cloud and the image in other ways. For example, the processor may determine the same objects in the second point cloud and the image and make the same objects overlap, thereby registering the second point cloud and the image.
[0049] S150: Process the second point cloud according to the image to generate a target point cloud containing colors.
[0050] After registering the second point cloud and the image, the processor can process the second point cloud according to the image, that is, color the second point cloud based on the color of the image to generate a target point cloud containing colors.
[0051] For example, the processor may determine the overlap between the image and the second point cloud, and copy the color of the overlap in the image to the overlap in the second point cloud. Alternatively, the processor may determine the same object in the image and the second point cloud, and assign the color of the object in the image to the corresponding object in the second point cloud.
[0052] Through the above steps, the processor finally generates a point cloud containing colors, that is, a target point cloud. Since the color of the image taken by the camera can reflect the real driving environment, the above solution can provide the driver with a more intuitive visual point cloud, which is convenient for the driver to better perform refined operations such as reversing, changing lanes and overtaking according to the target point cloud.
[0053] Since the density of the second point cloud obtained after resampling is relatively sparse, in order to improve the display effect after visualization of the second point cloud, when processing the second point cloud according to the image, the processor may process the second point cloud according to the following steps:
[0054] determining edges of the object in the second point cloud based on the image;
[0055] Performing filtering processing on the second point cloud according to the edge to obtain a second point cloud with increased density, wherein the second point cloud with increased density includes the edge;
[0056] The color of the second point cloud with increased density is determined from the image.
[0057] The processor may process the second point cloud using an edge preserving filter algorithm to determine the edge of the object in the second point cloud. For example, the processor may use a guided filter algorithm to determine the point cloud blocks in the second point cloud corresponding to the object in the above image, thereby determining the edge of the object in the second point cloud; then filter the point cloud within the edge to increase the density of the point cloud within the edge to obtain a second point cloud with increased density; then, based on the correspondence between the image and the second point cloud, determine the color of the second point cloud with increased density. Finally, a target point cloud with a better visualization effect (i.e., a second point cloud with increased density including color) is obtained.
[0058] The above-mentioned objects are, for example, objects in the driving environment, such as adjacent cars, obstacles, etc. The above-mentioned visualization effect may be better when the details of the objects are richer.
[0059] The processor may directly present the target point cloud to the driver, or may process the target point cloud based on the method described below before presenting it to the driver.
[0060] For example, when the environment area corresponding to the target point cloud is larger than the preset environment area, the processor may further delete redundant point cloud blocks in the target point cloud, where the environment area corresponding to the redundant point cloud blocks is outside the preset environment area.
[0061] The preset environment area is, for example, an area within 200 meters in front of the movable platform and within 5 meters in height. An area that is too large does not make much sense to the driver and may distract the driver's attention. Therefore, the above solution can focus the driver's attention on the driving environment near the movable platform and improve the safety of the movable platform. In addition, displaying part of the content in the target point cloud can also reduce the burden on the processor.
[0062] In addition, in order to further improve the visualization effect of the target point cloud, the processor can also identify objects in the target point cloud and replace the objects in the target point cloud with a 3D model. For example, the processor can identify people, cars and traffic lights in the target point cloud, determine the positions of these objects, and replace these point cloud objects with more detailed 3D models.
[0063] The processor may also process the target point cloud to generate one or more mesh grids, where the mesh grids may be composed of lines connecting three adjacent points in the target point cloud.
[0064] After the processor obtains the target point cloud, it can directly render the target point cloud, or further process the target point cloud according to the above method and then render the processed target point cloud. For example, the processor can render the target point cloud according to the observation angle set by the driver or the observation angle preset by the program, and display the rendered target point cloud on the screen, thereby improving the user experience.
[0065] Figure 2 Another schematic diagram of the method for processing point clouds provided by the present application is shown.
[0066] like Figure 2 As shown, after the processor obtains multiple point cloud data from the laser radar, it merges the multiple point cloud data into one point cloud (ie, the first point cloud in method 100), and resamples the point cloud to obtain a sparse point cloud.
[0067] Subsequently, the processor acquires the image from the camera and uses the calibration results of the camera and the lidar to project the sparse point cloud onto the image to obtain a sparse point cloud of the superimposed image.
[0068] Since the visualization effect of the sparse point cloud of the superimposed image is poor, the edge-preserving filtering algorithm can be used to process the sparse point cloud of the superimposed image, and the 3D coordinates of the pixels around the sparse point cloud can be guessed by using the consistency between the image and the sparse point cloud, and the sparse point cloud can be filled with pixels to obtain a dense point cloud with color. For each point in the dense point cloud, its color information is the pixel value of the pixel corresponding to the point.
[0069] The processor can detect objects in the dense point cloud, for example, detect people, cars, traffic lights and location information of these objects in the dense point cloud, and replace these objects with 3D models to achieve better visualization effect.
[0070] Finally, the processor can render the point cloud according to the viewing angle set by the driver or the viewing angle automatically set by the program, and present it on the screen.
[0071] The above describes in detail an example of a method for processing point clouds provided by the present application. It is understandable that in order to achieve the above functions, the device for processing point clouds includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should easily realize that, in combination with the units and algorithm steps of each example described in the embodiments disclosed herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in hardware or in a computer software-driven hardware manner depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.
[0072] The present application can divide the processing point cloud device into functional units according to the above method example. For example, each function can be divided into each functional unit, or two or more functions can be integrated into one functional unit. The above functional units can be implemented in the form of hardware or software. It should be noted that the division of units in the present application is schematic and is only a logical functional division. There may be other division methods in actual implementation.
[0073] Figure 3 A schematic structural diagram of a point cloud processing device provided in the present application is shown. Figure 3 The dotted line in the figure indicates that the unit is an optional unit. The device 300 can be used to implement the method described in the above method embodiment. The device 300 can be a software module, a chip, a terminal device or other electronic device.
[0074] The device 300 includes one or more processing units 301, which can support the device 300 to implement Figure 1The method in the corresponding method embodiment. The processing unit 301 can be a software processing unit, a general-purpose processor, or a dedicated processor. The processing unit 301 can be used to control the device 300, execute a software program (for example, a software program including the method 100), and process data (for example, process the first point cloud). The device 300 may also include a communication unit 305 to implement signal input (reception) and output (transmission).
[0075] For example, the device 300 may be a software module, and the communication unit 305 may be an interface function of the software module. The software module may run on a processor or a control circuit.
[0076] For another example, the apparatus 300 may be a chip, the communication unit 305 may be an input and / or output circuit of the chip, or the communication unit 305 may be a communication interface of the chip, and the chip may be a component of a terminal device or other electronic device.
[0077] In the apparatus 300, the communication unit 305 may execute: acquiring a first point cloud; acquiring an image corresponding to the same environment as the first point cloud;
[0078] The processing unit 301 can execute: resampling the first point cloud to obtain a second point cloud, wherein the density of the second point cloud is less than the density of the first point cloud; aligning the second point cloud and the image; processing the second point cloud according to the image to generate a target point cloud containing color.
[0079] It should be noted that the processing unit 301 may also include a resampling unit and a configuration unit, wherein the resampling unit is used to resample the first point cloud, and the configuration unit is used to register the second point cloud and the image.
[0080] Optionally, the processing unit 301 is specifically used to: determine a point cloud block in the first point cloud whose point cloud density is greater than a density threshold; and resample the point cloud block.
[0081] Optionally, the processing unit 301 is specifically used to: determine the edge of the object in the second point cloud based on the image; filter the second point cloud based on the edge to obtain a second point cloud with increased density, wherein the second point cloud with increased density includes the edge; and determine the color of the second point cloud with increased density based on the image.
[0082] Optionally, the processing unit 301 is specifically used to: when the environment area corresponding to the target point cloud is larger than the preset environment area, delete redundant point cloud blocks in the target point cloud, and the environment area corresponding to the redundant point cloud blocks is outside the preset environment area.
[0083] Optionally, the processing unit 301 is specifically used to: identify an object in the target point cloud; and replace the object with a three-dimensional model.
[0084] Optionally, the processing unit 301 is further used to: process the target point cloud to generate a surface patch.
[0085] Optionally, the patch consists of three adjacent points in the target point cloud.
[0086] Optionally, the processing unit 301 is further used to: render the target point cloud.
[0087] Optionally, the communication unit 305 is specifically used to: obtain multiple point clouds from multiple laser radars; the processing unit 301 is also used to: splice the multiple point clouds into the first point cloud.
[0088] Optionally, the processing unit 301 is specifically used to: splice the multiple point clouds into the first point cloud according to the external parameters of the multiple laser radars.
[0089] Optionally, the external parameters are used to indicate the positional relationship of the coordinate system of each laser radar among the multiple laser radars relative to the coordinate system of the movable platform.
[0090] Optionally, the external parameters include:
[0091] The rotation parameters of the coordinate system of each laser radar in the plurality of laser radars relative to the coordinate system of the movable platform; and / or,
[0092] The translation parameters of the coordinate system of each laser radar among the multiple laser radars relative to the coordinate system of the movable platform.
[0093] Optionally, the processing unit 301 is specifically configured to: stitch the multiple point clouds together using an ICP algorithm to obtain the first point cloud.
[0094] Those skilled in the art will clearly understand that for the convenience and brevity of description, the specific working process and effects of the above devices and units can be found in Figure 1 For the sake of brevity, it will not be repeated here.
[0095] As an optional implementation, the above steps can be completed by hardware-based logic circuits or software-based instructions. For example, the processing unit 301 can be a central processing unit (CPU), a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, such as discrete gates, transistor logic devices, or discrete hardware components.
[0096] The device 300 may include one or more storage units 302, in which a program 304 (e.g., a software program including the method 100) is stored. The program 304 can be executed by the processing unit 301 to generate instructions 303, so that the processing unit 301 performs the method described in the above method embodiment according to the instructions 303. Optionally, data (e.g., the first point cloud) may also be stored in the storage unit 302. Optionally, the processing unit 301 may also read the data stored in the storage unit 302, and the data may be stored at the same storage address as the program 304, or the data may be stored at a different storage address from the program 304.
[0097] The processing unit 301 and the storage unit 302 may be provided separately or integrated together, for example, integrated on a single board or a system on chip (SOC).
[0098] The present application also provides a computer program product, which, when executed by the processing unit 301, implements the method described in any embodiment of the present application.
[0099] The computer program product may be stored in the storage unit 302 , for example, a program 304 , which is finally converted into an executable target file that can be executed by the processing unit 301 after preprocessing, compiling, assembling, and linking.
[0100] The computer program product can be transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, from one website site, computer, server or data center to another website site, computer, server or data center via wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means.
[0101] The present application also provides a computer-readable storage medium (e.g., storage unit 302) on which a computer program is stored, and when the computer program is executed by a computer, the method described in any embodiment of the present application is implemented. The computer program can be a high-level language program or an executable target program.
[0102] The computer-readable storage medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a magnetic tape), an optical medium (e.g., a digital video disc (DVD)), or a semiconductor medium (e.g., a solid state disk (SSD)). For example, the computer-readable storage medium may be a volatile memory or a non-volatile memory, or the computer-readable storage medium may include both a volatile memory and a non-volatile memory. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).
[0103] It should be understood that in the various embodiments of the present application, the size of the serial number of each process does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.
[0104] The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.
[0105] The systems, devices and methods disclosed in the embodiments provided in this application may be implemented in other ways. For example, some features of the method embodiments described above may be ignored or not performed. The device embodiments described above are merely schematic, and the division of units is only a logical function division. There may be other division methods in actual implementation, and multiple units or components may be combined or integrated into another system. In addition, the coupling between the units or the coupling between the components may be direct coupling or indirect coupling, and the coupling may include electrical, mechanical or other forms of connection.
[0106] In short, the above description is only a partial embodiment of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims
1. A method for processing point clouds, It is characterized in that Applied to a movable platform, the movable platform includes a point cloud sensor and a visual sensor, the method includes: Acquire a first point cloud; the first point cloud may be obtained by splicing a plurality of point clouds using external parameters, the plurality of point clouds being obtained by a plurality of laser radars, the external parameters comprising: a rotation parameter of a coordinate system of each of the plurality of laser radars relative to a coordinate system of the movable platform; and / or a translation parameter of a coordinate system of each of the plurality of laser radars relative to a coordinate system of the movable platform; Acquire an image corresponding to the same environment as the first point cloud; Determine a point cloud block in the first point cloud having a point cloud density greater than a density threshold; Resampling the point cloud block to obtain a second point cloud, wherein the density of the second point cloud is less than the density of the first point cloud; registering the second point cloud and the image; determining an edge of an object in the second point cloud based on the image; Performing filtering processing on the point cloud within the edge to increase the density of the point cloud within the edge to obtain a second point cloud with increased density; The color of the second point cloud with increased density is determined according to the image, and a target point cloud containing color is generated.
2. The method according to claim 1, It is characterized in that Also includes: When the environment area corresponding to the target point cloud is larger than the preset environment area, redundant point cloud blocks in the target point cloud are deleted, and the environment area corresponding to the redundant point cloud blocks is outside the preset environment area.
3. The method according to claim 1, It is characterized in that Also includes: identifying an object in the target point cloud; The object is replaced with a three-dimensional model.
4. The method according to claim 1, It is characterized in that Also includes: The target point cloud is processed to generate a surface patch.
5. The method according to claim 4, It is characterized in that The patch is composed of three adjacent points in the target point cloud.
6. The method according to claim 1, It is characterized in that Also includes: Rendering the target point cloud.
7. The method according to any one of claims 1 to 6, It is characterized in that The obtaining of the first point cloud comprises: Get multiple point clouds from multiple lidars; The multiple point clouds are stitched into the first point cloud.
8. The method according to claim 7, It is characterized in that The step of stitching the plurality of point clouds into the first point cloud comprises: The multiple point clouds are spliced into the first point cloud according to the external parameters of the multiple laser radars.
9. The method according to claim 8, It is characterized in that The external parameters are used to indicate the positional relationship of the coordinate system of each laser radar among the multiple laser radars relative to the coordinate system of the movable platform.
10. The method according to claim 7, It is characterized in that The step of stitching the plurality of point clouds into the first point cloud comprises: The multiple point clouds are spliced by an iterative closest point ICP algorithm to obtain the first point cloud.
11. A device for processing point clouds, It is characterized in that The invention is applied to a movable platform, wherein the movable platform comprises a point cloud sensor and a visual sensor, and the device comprises a communication unit, a resampling unit, a registration unit and a processing unit. The communication unit is used to: obtain a first point cloud; obtain an image corresponding to the same environment as the first point cloud; the first point cloud is obtained by splicing multiple point clouds using external parameters, and the multiple point clouds are obtained by multiple laser radars, and the external parameters include: rotation parameters of the coordinate system of each laser radar in the multiple laser radars relative to the coordinate system of the movable platform; and / or, translation parameters of the coordinate system of each laser radar in the multiple laser radars relative to the coordinate system of the movable platform; The resampling unit is used to: resample the first point cloud to obtain a second point cloud, wherein the density of the second point cloud is less than the density of the first point cloud; The registration unit is used to: register the second point cloud and the image; The processing unit is used to: process the second point cloud according to the image to generate a target point cloud containing colors; The processing unit is specifically used for: determining an edge of an object in the second point cloud based on the image; Performing filtering processing on the second point cloud according to the edge to obtain a second point cloud with increased density, wherein the second point cloud with increased density includes the edge; Filtering the second point cloud according to the edge to obtain a second point cloud with increased density includes: filtering the point cloud within the edge to increase the density of the point cloud within the edge to obtain a second point cloud with increased density; determining the color of the second point cloud with increased density according to the image; The resampling unit is specifically used for: Determine a point cloud block in the first point cloud having a point cloud density greater than a density threshold; The point cloud block is resampled.
12. The device according to claim 11, It is characterized in that The processing unit is also used for: When the environment area corresponding to the target point cloud is larger than the preset environment area, redundant point cloud blocks in the target point cloud are deleted, and the environment area corresponding to the redundant point cloud blocks is outside the preset environment area.
13. The device according to claim 11, It is characterized in that The processing unit is also used for: identifying an object in the target point cloud; The object is replaced with a three-dimensional model.
14. The device according to claim 11, It is characterized in that The processing unit is also used for: The target point cloud is processed to generate a surface patch.
15. The device according to claim 14, It is characterized in that The patch is composed of three adjacent points in the target point cloud.
16. The device according to claim 11, It is characterized in that The processing unit is also used for: Rendering the target point cloud.
17. The device according to any one of claims 11 to 16, It is characterized in that The communication unit is specifically used to: obtain multiple point clouds from multiple laser radars; The processing unit is further used to: splice the multiple point clouds into the first point cloud.
18. The device according to claim 17, It is characterized in that The processing unit is specifically used for: The multiple point clouds are spliced into the first point cloud according to the external parameters of the multiple laser radars.
19. The device according to claim 18, It is characterized in that The external parameters are used to indicate the positional relationship of the coordinate system of each laser radar among the multiple laser radars relative to the coordinate system of the movable platform.
20. The device according to claim 17, It is characterized in that The processing unit is specifically used for: The multiple point clouds are spliced by an iterative closest point ICP algorithm to obtain the first point cloud.
21. A movable platform, It is characterized in that include: A memory, a processor, a point cloud sensor and a visual sensor, wherein the memory is used to store instructions, the processor is used to execute the instructions stored in the memory, and control the point cloud sensor to acquire a first point cloud according to the instructions; and control the visual sensor to acquire an image corresponding to the same environment as the first point cloud; The processor is further configured to execute according to the instructions: Resampling the first point cloud to obtain a second point cloud, wherein the density of the second point cloud is less than the density of the first point cloud; The first point cloud may be obtained by splicing a plurality of point clouds using external parameters, the plurality of point clouds being obtained by a plurality of laser radars, the external parameters comprising: a rotation parameter of a coordinate system of each of the plurality of laser radars relative to a coordinate system of the movable platform; and / or a translation parameter of a coordinate system of each of the plurality of laser radars relative to a coordinate system of the movable platform; registering the second point cloud and the image; Processing the second point cloud according to the image to generate a target point cloud containing colors; The processor is further configured to execute according to the instructions: determining an edge of an object in the second point cloud based on the image; Performing filtering processing on the second point cloud according to the edge to obtain a second point cloud with increased density, wherein the second point cloud with increased density includes the edge; Filtering the second point cloud according to the edge to obtain a second point cloud with increased density includes: filtering the point cloud within the edge to increase the density of the point cloud within the edge to obtain a second point cloud with increased density; determining the color of the second point cloud with increased density according to the image; The processor is further configured to execute according to the instructions: Determine a point cloud block in the first point cloud having a point cloud density greater than a density threshold; The point cloud block is resampled.
22. The movable platform according to claim 21, It is characterized in that The processor is configured to execute according to the instruction: When the environment area corresponding to the target point cloud is larger than the preset environment area, redundant point cloud blocks in the target point cloud are deleted, and the environment area corresponding to the redundant point cloud blocks is outside the preset environment area.
23. The movable platform according to claim 21, It is characterized in that The processor is configured to execute according to the instruction: identifying an object in the target point cloud; The object is replaced with a three-dimensional model.
24. The movable platform according to claim 21, It is characterized in that The processor is configured to execute according to the instruction: The target point cloud is processed to generate a surface patch.
25. The movable platform according to claim 24, It is characterized in that The patch is composed of three adjacent points in the target point cloud.
26. The movable platform according to claim 21, It is characterized in that The processor is configured to execute according to the instruction: Rendering the target point cloud.
27. A movable platform according to any one of claims 21 to 26, It is characterized in that The processor is configured to execute according to the instruction: Acquire multiple point clouds from multiple point cloud sensors; The multiple point clouds are stitched into the first point cloud.
28. The movable platform according to claim 27, It is characterized in that The processor is configured to execute according to the instruction: The multiple point clouds are spliced into the first point cloud according to the external parameters of the multiple laser radars.
29. The movable platform according to claim 28, It is characterized in that The external parameters are used to indicate the positional relationship of the coordinate system of each laser radar among the multiple laser radars relative to the coordinate system of the movable platform.
30. The movable platform according to claim 27, It is characterized in that The processor is configured to execute according to the instruction: The multiple point clouds are spliced by an iterative closest point ICP algorithm to obtain the first point cloud.
31. The movable platform according to claim 28, It is characterized in that The point cloud sensor is a laser radar, and the visual sensor is a camera.
32. A computer storage medium, It is characterized in that A computer program is stored thereon, and when the computer program is executed by a computer, the computer is caused to execute the method according to any one of claims 1 to 10.
33. A computer program product comprising instructions, It is characterized in that When the instructions are executed by a computer, the computer is caused to perform the method according to any one of claims 1 to 10.
