4D millimeter wave radar point cloud miscellaneous point filtering method, device, medium and product

Abstract

The embodiment of the invention relates to the field of radar point cloud data processing, and discloses a 4D millimeter wave radar point cloud miscellaneous point filtering method and device, a medium and a product. Preprocessing each frame of 4D millimeter wave radar point cloud, and clustering to obtain a point cloud cluster; carrying out feature extraction on the point cloud cluster, and respectively extracting a plane point cloud feature vector and a columnar point cloud feature vector; according to a Manhattan world hypothesis, forced xyz orientation correction is carried out on the planar point cloud feature vector and the columnar point cloud feature vector, and the planar point cloud feature vector and the columnar point cloud feature vector are aligned with coordinate axes in the Manhattan world hypothesis; and screening the corrected plane point cloud features and columnar point cloud features, and reserving effective point cloud features for map construction. The technical problem that the point cloud of the millimeter wave radar generates miscellaneous points due to the garage environment can be solved at least.

Description

Technical Field

[0001] This application relates to the field of radar point cloud data processing, and particularly to a method, device, medium, and product for filtering out spurious points in 4D millimeter-wave radar point clouds. Background Art

[0002] In enclosed environments such as underground parking lots, due to the presence of a large number of static or dynamic objects (such as walls, pipes, fire hydrants, electric vehicles, pedestrians, etc.), the surface characteristics and materials of these objects are different, and their reflection and refraction characteristics of radar waves are also different. Especially for metal objects such as cars and pipes, they often generate strong reflection signals, while non-metal objects (such as walls and glass) generate weak reflection signals. The difference in reflection intensity and scattering effect of different objects makes the 4D millimeter-wave radar likely to receive a large number of incorrect echo signals in the basement environment. These incorrect echo signals not only cause misjudgment of multiple echoes in the point cloud data (i.e., the same object is misjudged as multiple targets), but also may cause the radar system to misjudge a static object as a dynamic object, resulting in incorrect perception results.

[0003] In addition, due to the particularity of environments such as underground parking lots, the radar system usually cannot obtain sufficient position information. For example, in an underground garage, due to the lack of GPS signals, positioning and mapping face greater challenges. The 4D millimeter-wave radar can provide basic data for positioning and mapping for the autonomous driving system without the support of an external positioning system by acquiring environmental point cloud data. However, a large number of spurious points in the radar point cloud data will greatly affect the accuracy of the map, making subsequent positioning and path planning tasks difficult.

[0004] In the prior art, for this spurious point problem, methods based on point cloud clustering, feature extraction, and filtering are usually adopted, but these methods mainly focus on eliminating unnecessary noise and lack in-depth optimization for specific environments (such as underground garages). The method of point cloud feature extraction based on the Manhattan world hypothesis has not been effectively applied, and there is still a large room for improvement in the accuracy and efficiency of filtering out radar spurious points in the prior art. Summary of the Invention

[0005] An object of this application is to provide a method, device, medium, and product for filtering out spurious points in 4D millimeter-wave radar point clouds, at least to solve the technical problem that the point clouds of millimeter-wave radar are generated with spurious points due to the garage environment.

[0006] To achieve the above object, some embodiments of this application provide the following aspects:

[0007] In a first aspect, some embodiments of the present application also provide a method for filtering out noise points from 4D millimeter-wave radar point clouds, including preprocessing each frame of 4D millimeter-wave radar point clouds to cluster point cloud clusters; extracting features from the point cloud clusters, respectively extracting planar point cloud feature vectors and columnar point cloud feature vectors; according to the Manhattan world hypothesis, performing forced xyz orientation correction on the planar point cloud feature vectors and columnar point cloud feature vectors to align with the coordinate axes in the Manhattan world hypothesis; screening the corrected planar point cloud feature vectors and columnar point cloud feature vectors, and retaining valid point cloud features for map construction.

[0008] In a second aspect, some embodiments of the present application also provide an electronic device, which includes: one or more processors; and a memory storing computer program instructions, and when the computer program instructions are executed, the processors execute the steps of the method as described above.

[0009] In a third aspect, some embodiments of the present application also provide a computer-readable medium, on which computer program instructions are stored, and the computer program instructions can be executed by a processor to implement the method as described above.

[0010] In a fourth aspect, some embodiments of the present application also provide a computer program product, including computer programs / instructions, and when the computer programs / instructions are executed by a processor, the steps of the method as described above are implemented.

[0011] Compared with the related art, in the solution provided by the embodiments of the present application, the point cloud processing method based on the Manhattan world hypothesis effectively filters out radar noise points in the basement environment through precise feature extraction and orientation correction. In the basement scenario of autonomous driving, the filtered point cloud data can better reflect the environmental features, improving the accuracy and stability of local maps and global mapping. Simplifying the feature extraction and filtering steps can significantly reduce the amount of calculation and improve the efficiency of data processing, which is suitable for real-time environmental perception requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] One or more embodiments are exemplarily illustrated by pictures in the corresponding drawings. These exemplary illustrations do not limit the embodiments. Elements with the same reference numerals in the drawings are represented as similar elements, unless otherwise stated, and the drawings in the drawings do not constitute a proportional limitation.

[0013] Figure 1 It is a schematic flowchart of a method for filtering out noise points from 4D millimeter-wave radar point clouds provided by an embodiment of the present application;

[0014] Figure 2 It is a schematic flowchart of another method for filtering out noise points from 4D millimeter-wave radar point clouds provided by an embodiment of the present application;

[0015] Figure 3 It is a schematic structural diagram of an electronic device provided according to an embodiment of the present application. Specific embodiments

[0016] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Apparently, the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the scope of protection of the present application.

[0017] The first embodiment

[0018] The first embodiment of the present application relates to a method for filtering out noise points in 4D millimeter-wave radar point clouds. As Figure 1 shown, the method may include the following steps:

[0019] S101, preprocess each frame of 4D millimeter-wave radar point cloud and cluster to obtain point cloud clusters;

[0020] S102, extract features from the point cloud clusters, and extract a planar point cloud feature vector and a columnar point cloud feature vector respectively;

[0021] S103, according to the Manhattan world hypothesis, perform forced xyz orientation correction on the planar point cloud feature vector and the columnar point cloud feature vector, and align them with the coordinate axes in the Manhattan world hypothesis;

[0022] S104, screen the corrected planar point cloud feature vector and columnar point cloud feature vector, and retain the effective point cloud features for subsequent map construction.

[0023] The following will separately elaborate on each of the above steps in detail.

[0024] For step S101, there are many static or dynamic objects in the underground parking lot, such as walls, pipes, fire hydrants, moving people, animals, electric vehicles, etc. These objects generate a large amount of refraction and reflection interference on the electromagnetic waves emitted and received by the radar, resulting in echoes of the same target received by the radar in multiple directions and being misjudged as multiple targets; at the same time, metal objects such as cars and pipes reflect strongly, while non-metal objects have a small reflection intensity. This difference leads to inconsistent echo intensities received by the radar, increasing the complexity of data analysis; therefore, it is necessary to preprocess the 4D millimeter-wave radar point cloud data in the basement environment.

[0025] For step S102, after the point cloud clusters are clustered, the feature extraction stage is entered. Specifically, for each point cloud cluster, its structural features are extracted through geometric analysis. Plane structural features are extracted by fitting methods. For large-area plane structures (such as walls, floors, etc.), the random sampling consensus algorithm (RANSAC) can be used to extract plane features. The algorithm calculates and screens out qualified plane point clouds by repeatedly randomly selecting points and fitting plane models. These plane feature point clouds represent important plane information in the scene.

[0026] For columnar structure features, the principal component analysis (PCA) method is used to extract features. Columnar structures usually appear in the shape of long strips, and the main axis direction of the point cloud is relatively obvious. Through the PCA method, the main component direction of the point cloud cluster can be identified and the features of the columnar point cloud can be extracted. If the main eigenvalue ratio of the point cloud cluster is large, it can usually be determined to be a columnar structure. At this time, the columnar point cloud features will be recorded for subsequent processing steps.

[0027] For step S103, after the feature extraction is completed, the extracted plane and column features are oriented. Since the structures in scenes such as underground parking lots are usually vertical or horizontal, the Manhattan world hypothesis is used to correct the feature orientation. According to this hypothesis, the ground is usually horizontal, while the walls and columns are perpendicular to the ground.

[0028] If the feature vector of a plane point cloud (such as a wall, ground, etc.) has a large deviation from the ground, it needs to be rotated and adjusted to align its direction with the ground. Similarly, if the feature vector of a columnar structure has a large deviation from the vertical axis, it also needs to be adjusted to align it with the vertical axis. Through this correction step, the orientation of the plane point cloud and the columnar point cloud will be consistent with the actual geographical environment, reducing the structural deviation caused by point cloud data errors.

[0029] For step S104, after completing the orientation correction, the corrected point cloud features are then screened. The screening process is completed by calculating the distance between the point cloud and its features. Specifically, the distance from each point cloud to the corrected planar or columnar features is calculated, and the points are screened according to a preset distance threshold. Only those point clouds whose distance is less than the set threshold will be retained, while those farther away will be considered as noise and removed. After screening, the remaining point cloud features represent the valid structural information in the scene, such as walls, columns, ground, etc. These valid point cloud features will serve as the basis for subsequent map construction.

[0030] It is not difficult to find that, compared with the related technologies, in the solution provided by the embodiments of the present application, through operations such as clustering, feature extraction, orientation correction, and secondary screening of point cloud data, radar clutter in environments such as underground parking lots is successfully removed, representative environmental features are extracted, and accurate point cloud data support can be provided for subsequent local map building, global mapping, and SLAM tasks. This method not only improves the reliability of the data but also reduces the computational complexity, and is applicable to the applications of autonomous driving systems and intelligent transportation systems in complex environments.

[0031] Second Embodiment

[0032] The second embodiment of the present application relates to a method for filtering 4D millimeter-wave radar point cloud clutter. The second implementation is an improvement based on the first embodiment. The specific improvement lies in:

[0033] Further, the preprocessing of each frame of 4D millimeter-wave radar point cloud and clustering to obtain point cloud clusters includes: performing voxel filtering on the 4D millimeter-wave radar point cloud, processing the dense overlapping small cluster point clouds obtained from strong reflections, and performing clustering to obtain the point cloud clusters.

[0034] In complex environments such as underground parking lots, due to the reflection characteristics of radar echoes, the point cloud data received by the radar may contain a large amount of noise and clutter. These noise point clouds are usually caused by strong reflections of metal objects or multipath effects, resulting in a high degree of overlap of point cloud data in space, forming "dense overlapping small cluster point clouds". Therefore, effective preprocessing and clustering processing are performed on these noise point clouds to extract effective feature information for subsequent steps.

[0035] Voxel filtering is a spatial downsampling technique aimed at reducing noise and redundant points in point cloud data and improving data processing efficiency. According to the resolution of the radar and the characteristics of environments such as underground parking lots, a suitable voxel size is set. The size of the voxel determines the degree of downsampling of the point cloud data. Usually, the voxel size will be set according to the density of the point cloud data and processing requirements.

[0036] The point cloud is divided into columnar and planar point clouds to reduce the influence of clutter in the basement environment; the point cloud clusters after clustering usually represent effective object or structural features (such as walls, columns, etc.) in the environment. These point cloud clusters can be used for subsequent feature extraction, orientation correction, and screening steps. If the size of some point cloud clusters is small, the shape is irregular, or the distance from other point cloud clusters is too far, further screening or removal is performed to avoid misjudgment as effective features.

[0037] Furthermore, the feature extraction of the point cloud cluster includes: statistically distinguishing the inliers and outliers in the point cloud cluster by the RANSAC method, specifically including randomly selecting points from the point cloud cluster, constructing a plane model, calculating the distance from the points to the plane according to a preset threshold, screening out the eligible plane models, so as to obtain the plane point cloud features.

[0038] First, randomly select a point from the point cloud cluster as the initial seed point p0=(x0,y0,z0) and add it to the plane model M = A*(x - x0)+B*(y - y0)+C*(z - z0)=0. Then randomly select two other points p1=(x1,y1,z1) and p2=(x2,y2,z2) and also add them to the plane model to form a plane M0 with p0. At the same time, calculate the distance d from other points to M0 through the formula. According to the characteristics of the radar noise value (the farther away, the greater the noise and the greater the measurement error), design the threshold t. Count the points with d less than the threshold. If the number of points exceeds the preset minimum point number threshold n, it is considered that the plane model meets the conditions, and at the same time, the vector feature of the plane model is obtained as r=(A,B,C). Repeat the above iterative process, and select the plane model with the largest number of support points as the final segmentation result.

[0039] Furthermore, the feature extraction of the point cloud cluster includes: extracting the columnar point cloud features by the PCA method, specifically including screening the eigenvector corresponding to the columnar point cloud features from the point cloud cluster, and obtaining the columnar point cloud features according to the eigenvector.

[0040] Using the point cloud cluster obtained after the above processing to accelerate the point cloud processing flow. Due to the characteristics of the radar, except for the large-area flat point cloud such as the wall surface, the angular appearance will be "smoothed", so the wall corners are generally shown as rounded corners. In this case, the principal component analysis (PCA) can be used to obtain the eigenvector corresponding to the columnar point cloud. When one of the two eigenvalues of Λ is much larger than the other (set this multiple ratio as n according to experience), it can be determined that the point cloud cluster is a columnar point cloud. At this time, let the eigenvector corresponding to the largest eigenvalue be the vector feature λ of the current columnar point cloud, and the x and y means of the point cloud cluster are respectively and Finally, estimate the radius of the cylinder as r by calculating the average distance from the point cloud to the main axis. c .

[0041] Furthermore, the Manhattan world hypothesis includes: the columnar point cloud feature vectors are parallel to each other; the intersection angles between the plane point cloud feature vectors and the columnar point cloud feature vectors are all right angles.

[0042] Based on the obtained planar feature vector r = (A, B, C), the feature vector c = (J, K, L) of the columnar point cloud can also be obtained. Perform forced xyz orientation correction on these two related point cloud results. According to the Manhattan world assumption, the feature vectors c of the columnar point clouds are all parallel, and the intersection angles between the feature vector r of the plane and c are all right angles. Thus, process according to the following method:

[0043] In the Manhattan world assumption, it is considered that the ground is completely horizontal, so for perform processing perpendicular to the horizontal ground. The columnar point cloud can be represented in the following way:

[0044]

[0045] where z0 is the z coordinate of a specific point of the line on the z-axis, and t is a real number parameter. The point cloud after forced orientation correction is secondarily screened by the distance from the point to the line. The point cloud with a distance less than r c is retained, and at the same time, the features of this columnar point cloud are retained for subsequent local map building;

[0046] Process the planar point cloud and its features. Similarly, in the Manhattan world assumption, it can be considered that the wall feature vector and the horizontal ground vector are nearly orthogonal. So perform forced orientation processing on the planar features to obtain:

[0047]

[0048] where p0 is the center point of the point cloud. The point cloud after forced orientation correction is secondarily screened by the distance from the point to the plane, and the points with a distance less than the threshold d are retained. The features of this planar point cloud are retained for subsequent local map building.

[0049] Furthermore, the method is applied to the processing of 4D millimeter-wave radar point cloud data in the basement environment.

[0050] Underground parking lots usually have a relatively enclosed space structure. Objects such as surrounding walls, columns, parking spaces, and doors are relatively fixed, and the radar is prone to being interfered by the reflected waves of stationary and moving objects (such as cars, pedestrians, electric vehicles, etc.) in this environment, resulting in a large number of invalid clutter points in the point cloud data. The point cloud data output by the radar in the basement has a high density, especially when close to metal objects or other strong reflection objects, the density of the point cloud points is higher, causing misjudgment and incorrect classification.

[0051] Furthermore, the method is applied to local map building or SLAM global mapping.

[0052] The construction of the local map is to obtain the point cloud data of the surrounding environment in real time by a 4D millimeter-wave radar, and use this data to construct an accurate local map. This map is used for tasks such as the instant positioning, obstacle avoidance, and path planning of the system. By accurately extracting the plane and columnar features in the basement environment and correcting the directions and performing secondary screening on these features, it can help to construct a local map with high precision. These local maps have fewer miscellaneous points and can provide reliable data support for subsequent positioning and control.

[0053] SLAM global mapping is widely used in the field of autonomous driving, aiming to simultaneously perform positioning and mapping in real time in an unknown environment. By continuously collecting sensor data and updating the map, the SLAM system can construct a global map of the entire environment.

[0054] In the SLAM task, the point cloud data of the 4D millimeter-wave radar plays a key role. However, the original point cloud data usually contains a large number of miscellaneous points, which will interfere with the mapping process and reduce the accuracy and reliability of the map. Through the application of this application, in the process of SLAM global mapping, it is possible to first remove the miscellaneous points through feature extraction and orientation correction, ensure that the constructed map has higher accuracy and stability, and can also operate stably in a dynamic environment. Especially in complex and enclosed environments such as underground parking lots, it can effectively filter the noise in the environment, so as to generate a global map more accurately.

[0055] As Figure 2 shown, single-frame 4D millimeter-wave radar cloud: Obtain the point cloud data from the 4D millimeter-wave radar. These data contain information in four dimensions: distance, speed, horizontal angle, and pitch angle, and are used to describe the surrounding environment.

[0056] Point cloud clustering: Group the point cloud data through a clustering algorithm, divide the point cloud into multiple point cloud clusters, and each point cloud cluster usually represents an object or a structural part in the environment.

[0057] The xyz-axis orientation of each point cloud cluster: Calculate the direction of each point cloud cluster in space and record its orientation information in the xyz coordinate system. This step prepares for subsequent feature extraction and orientation correction. Force the orientation to the xyz direction: Under the guidance of the Manhattan world hypothesis, forcefully adjust the direction of the point cloud features to ensure that the point cloud features are consistent with the actual orientations of structures such as the ground and walls. The eigenvector of the columnar point cloud is aligned with the vertical axis, and the normal vector of the planar point cloud is perpendicular to the ground.

[0058] Extraction of columnar point cloud and planar point cloud: Extract columnar and planar point cloud features from the point cloud clusters respectively. The columnar point cloud usually represents structures such as columns and pipes, and the planar point cloud represents large-area flat structures such as walls and the ground.

[0059] Point cloud filtering: The processed point cloud is screened again. By setting a distance threshold, those point clouds that are too far away and do not conform to the feature structure are removed, and finally the effective point cloud data is retained, preparing for local mapping or global mapping.

[0060] It is not difficult to find that, compared with the related technology, in the solution provided by the embodiments of the present application, through steps such as voxel filtering, clustering, feature extraction, and forced orientation correction, the miscellaneous points in the point cloud are significantly reduced, and the efficiency and accuracy of data processing are improved. It is especially suitable for closed environments such as underground parking lots, where radar signals are easily affected by strong reflections and multipath effects. By extracting the plane and columnar point cloud features, the structures such as walls and columns in the environment can be accurately described, providing accurate data for subsequent map construction and positioning.

[0061] In addition, some embodiments of the present application also provide an electronic device. The electronic device can be various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and so on. The electronic device can also be various forms of mobile devices, such as personal digital processors, cellular phones, smart phones, wearable devices, and other similar computing devices.

[0062] The electronic device includes: one or more processors; and a memory storing computer program instructions, and when the computer program instructions are executed, the processors execute the steps of the method provided by any one or more of the above embodiments. Figure 3 An exemplary structural diagram of the electronic device is disclosed. As Figure 3 shown, the electronic device includes: one or more processors 1101, a memory 1102, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Each component is interconnected using different buses and can be installed on a common motherboard or installed in other ways as needed. The processor can process instructions executed within the electronic device, including instructions stored in the memory or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some other embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple electronic devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Among them, the components shown herein, their connections and relationships, and their functions are only examples and are not intended to limit the implementation of the present application described and / or claimed herein.

[0063] The electronic device may further include: an input device 1103 and an output device 1104. The processor 1101, the memory 1102, the input device 1103, and the output device 1104 may be connected through a bus or other means. Figure 3 Taking the connection through the bus as an example.

[0064] The input device 1103 can receive input digital or character information, and generate key signal inputs related to the user settings and function controls of the electronic device, such as input devices like touchscreens, keypads, mice, trackpads, touchpads, pointing sticks, one or more mouse buttons, trackballs, joysticks, etc. The output device 1104 may include a display device, an auxiliary lighting device (e.g., LED), and a haptic feedback device (e.g., a vibration motor), etc. The display device may include, but is not limited to, a liquid crystal display (LCD), a light-emitting diode (LED) display, and a plasma display. In some embodiments, the display device may be a touchscreen.

[0065] To provide interaction with the user, the electronic device may be a computer. The computer has: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or an LCD (liquid crystal display) monitor); and a keyboard and a pointing device (e.g., a mouse or a trackball), through which the user can provide input to the computer. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensing feedback (e.g., visual feedback, auditory feedback, or haptic feedback); and the input from the user can be received in any form (including sound input, voice input, or haptic input).

[0066] In the embodiments of the present application, a computer program / instructions is stored on a computer-readable medium. When the computer program / instructions are executed by the processor, the steps of the method provided in any one or more of the above embodiments are implemented. The computer-readable medium may be included in the electronic device described in the above embodiments; or it may exist separately without being assembled into the device. The above computer-readable medium carries one or more computer-readable instructions.

[0067] The memory 1102 may be used as a non-transitory computer-readable storage medium, and may be used to store non-transitory software programs, non-transitory computer-executable programs, and modules. The processor 1101 executes various functional applications and data processing of the server by running the non-transitory software programs, instructions, and modules stored in the memory 1102, so as to implement the program instructions / modules corresponding to the method provided in any one or more of the above embodiments of the present application.

[0068] The memory 1102 may include a program storage area and a data storage area. The program storage area may store an operating system and application programs required for at least one function. The data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 1102 may include a high-speed random access memory and may also include non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory 1102 may optionally include a memory remotely disposed relative to the processor 1101, and these remote memories may be connected to the electronic device through a network. Examples of the above network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0069] It should be noted that the computer-readable medium described in this application may be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. The computer-readable medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this application, the computer-readable medium may be any tangible medium that contains or stores a program, and this program may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0070] The computer-readable medium includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. The information may be computer-readable instructions, data structures, program modules, or other data. Examples of the computer's storage medium include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory, or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD), or other optical storage, magnetic cassette tapes, magnetic disk storage, or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible by a computing device.

[0071] Computer program code for performing the operations of this application can be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, by using an Internet service provider to connect through the Internet).

[0072] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. For example, an application-specific integrated circuit (ASIC), a general-purpose computer, or any other similar hardware device can be used. In some embodiments, the software program of this application can be executed by a processor to implement the above steps or functions. Similarly, the software program of this application (including related data structures) can be stored in a computer-readable recording medium, for example, a RAM memory, a magnetic or optical drive, or a floppy disk and similar devices. In addition, some steps or functions of this application can be implemented by hardware, for example, as a circuit that cooperates with a processor to execute each step or function.

[0073] The computer program product provided by the embodiments of this application includes one or more computer programs / instructions. When the computer programs / instructions are executed by a processor, they wholly or partly generate the processes or functions described in the embodiments of this application. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wirelessly (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more integrated available media. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid state disk (SSD)), etc.

[0074] The flowcharts or block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of devices, methods, and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a part of code, which contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and the combinations of blocks in the block diagram and / or flowchart, may be implemented by a dedicated hardware-based system that performs the specified functions or operations, or may be implemented by a combination of dedicated hardware and computer instructions.

[0075] The scope of the present application is defined by the appended claims rather than the above description. Therefore, all changes that fall within the meaning and scope of the equivalent elements of the claims are intended to be included in the present application. Any reference signs in the claims should not be construed as limiting the claims involved. In addition, it is obvious that the word "comprising" does not exclude other elements or steps, and the singular does not exclude the plural. The multiple elements or devices recited in the apparatus claims may also be implemented by one element or device through software or hardware. The terms "first", "second", etc. are only used for descriptive distinction and do not represent any specific order, nor can they be construed as indicating or implying relative importance.

[0076] As described above, the above are only specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed in the present application can easily mention changes or substitutions, which should all be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims, and the above embodiments should be regarded as exemplary and non-limiting.

Claims

1. A 4D millimeter wave radar point cloud noise filtering method, characterized in that: The method comprises: Pre-process each frame of 4D millimeter-wave radar point cloud and cluster them to obtain point cloud clusters; Performing feature extraction on the point cloud cluster, respectively extracting a plane point cloud feature vector and a columnar point cloud feature vector; According to the Manhattan world hypothesis, the planar point cloud feature vector and the cylindrical point cloud feature vector are forced to be xyz oriented and aligned with the coordinate axes in the Manhattan world hypothesis; The corrected planar point cloud feature vector and cylindrical point cloud feature vector are screened to retain valid point cloud features for map construction.

2. The method according to claim 1, characterized in that The preprocessing of each frame of 4D millimeter wave radar point cloud and clustering of point cloud clusters include: The 4D millimeter-wave radar point cloud is subjected to voxel filtering, and the densely overlapping small cluster point cloud obtained by strong reflection is processed, and clustering is performed to obtain the point cloud cluster.

3. The method according to claim 2, characterized in that Extracting features from the point cloud cluster includes: The RANSAC method is used to count and distinguish the internal and external points in the point cloud group, specifically including randomly selecting points from the point cloud group, constructing a plane model, and calculating the distance from the point to the plane according to a preset threshold, screening out plane models that meet the conditions, thereby obtaining the plane point cloud feature vector.

4. The method according to claim 3, characterized in that Extracting features from the point cloud cluster includes: The cylindrical point cloud feature vector is extracted by the PCA method, specifically including selecting the eigenvalue vector that conforms to the cylindrical point cloud from the point cloud cluster, and obtaining the cylindrical point cloud feature vector according to the eigenvalue vector.

5. The method according to claim 4, characterized in that The Manhattan world assumption includes: the cylindrical point cloud feature vectors are parallel to each other; and the intersection angles of the plane point cloud feature vector and the cylindrical point cloud feature vector are both right angles.

6. The method according to any one of claims 1 to 5, characterized in that: The method is applied to 4D millimeter wave radar point cloud data processing in a basement environment.

7. The method according to any one of claims 1 to 5, characterized in that: The method is applied to local map building or SLAM global map building.

8. An electronic device, characterized in that: The electronic device comprises: one or more processors; and A memory storing computer program instructions, which, when executed, cause the processor to perform the steps of the method as claimed in any one of claims 1 to 7.

9. A computer readable medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images