Dynamic point cloud-based mapping method and device, computer equipment and storage medium
By combining the camera object detection algorithm and filtering model, the dynamic point clouds in robot mapping are accurately detected and processed, which solves the problem of poor effect of dynamic object interference and clustering algorithms, and improves the mapping accuracy and efficiency of static point clouds.
Patent Information
- Application Number
- CN202411989963.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-31
- Publication Date
- 2025-05-02
AI Technical Summary
During the robot mapping process, moving objects in the dynamic environment will interfere with the mapping/positioning accuracy of point clouds. The existing technology has the problem of inaccurate labeling of dynamic objects and accidentally delete useful point clouds, and the clustering algorithm is not effective in high-density environments.
The detection box range of the target camera is determined through the camera object detection algorithm and preset objects, the initial point cloud data set within the detection box range is extracted, the target point cloud cluster and its cluster type are determined, and the initial point cloud strategy is determined based on the cluster type. Based on the preset filtering model, the target point cloud cluster is dynamically tracked, the current location is determined, and the target point cloud graph construction strategy is determined based on the current location and the initial point cloud strategy.
It improves the accuracy of detection and removal of dynamic objects, optimizes the mapping effect of static point clouds, reduces the risk of accidentally deleting useful point clouds, and improves the performance of clustering algorithms in high-density environments.
Smart Images
Figure CN119919481A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of robotics technology, and in particular to a mapping method, device, computer equipment and storage medium based on dynamic point cloud. Background Art
[0002] During the robot mapping process, moving objects (such as people and vehicles) in a dynamic environment will interfere with the mapping / positioning accuracy of the point cloud. The commonly used Euclidean clustering method has the problem of inaccurate annotation of dynamic objects and mistaken deletion of useful point clouds. The defects and shortcomings of the existing technology are that the removal of dynamic point clouds depends on the generalization ability of training data, resulting in inaccurate recognition of objects outside the coverage of the data set; the use of simple rules (such as setting the point range to zero) may lead to the loss of useful point clouds; clustering algorithms (such as DBSCAN or Euclidean clustering) are not effective in high-density environments and may generate too many small clusters or interference clusters. Therefore, how to improve the accuracy of detection and removal of dynamic objects, and then optimize the mapping effect of static point clouds has become a technical problem that needs to be solved urgently. Summary of the invention
[0003] The present application provides a mapping method, apparatus, computer equipment and storage medium based on dynamic point clouds, which improve the accuracy of detection and removal of dynamic objects and thus optimize the mapping effect of static point clouds.
[0004] In a first aspect, the present application provides a mapping method based on dynamic point cloud, the method comprising:
[0005] Determine the detection frame range of the target camera through the camera target detection algorithm and the preset object, and determine the initial point cloud data set within the detection frame range;
[0006] Determining a target point cloud cluster and a cluster type of the target point cloud cluster from the initial point cloud data set, and determining an initial point cloud strategy according to the cluster type;
[0007] The target point cloud cluster is dynamically tracked based on a preset filtering model to determine the current position of the target point cloud cluster, and the target point cloud mapping strategy is determined according to the current position and the initial point cloud strategy.
[0008] In a second aspect, the present application further provides a mapping device based on a dynamic point cloud, the device comprising:
[0009] An initial point cloud data set determination module is used to determine the detection frame range of the target camera through a camera target detection algorithm and a preset object, and to determine an initial point cloud data set within the detection frame range;
[0010] An initial point cloud strategy determination module, used to determine a target point cloud cluster and a cluster type of the target point cloud cluster from the initial point cloud data set, and determine an initial point cloud strategy according to the cluster type;
[0011] The target point cloud mapping strategy determination module is used to dynamically track the target point cloud cluster based on a preset filtering model, determine the current position of the target point cloud cluster, and determine the target point cloud mapping strategy according to the current position and the initial point cloud strategy.
[0012] In a third aspect, the present application also provides a computer device, comprising a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and implement the dynamic point cloud-based mapping method as described above when executing the computer program.
[0013] In a fourth aspect, the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the processor implements the dynamic point cloud-based mapping method as described above.
[0014] The present application discloses a mapping method, device, computer equipment and storage medium based on dynamic point clouds, wherein the mapping method based on dynamic point clouds includes determining the detection frame range of the target camera through a camera target detection algorithm and a preset object, and determining an initial point cloud data set within the detection frame range; determining a target point cloud cluster and a cluster type of the target point cloud cluster from the initial point cloud data set, and determining an initial point cloud strategy according to the cluster type; dynamically tracking the target point cloud cluster based on a preset filtering model, determining the current position of the target point cloud cluster, and determining a target point cloud mapping strategy according to the current position and the initial point cloud strategy. In the above manner, the present application combines the camera target detection algorithm to accurately determine the detection frame range and extract point cloud data related to the target object. By determining the target point cloud cluster and its cluster type, the point cloud data can be more effectively classified and processed, and different point cloud strategies can be adopted according to different cluster types, thereby improving the mapping effect of static point clouds. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0016] Figure 1 is a schematic flow chart of a mapping method based on dynamic point cloud provided in the first embodiment of the present application;
[0017] Figure 2 is a schematic flow chart of a mapping method based on dynamic point cloud provided in the second embodiment of the present application;
[0018] Figure 3 A schematic block diagram of a mapping device based on dynamic point cloud provided in an embodiment of the present application;
[0019] Figure 4 A schematic block diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0020] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0021] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.
[0022] It should be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.
[0023] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.
[0024] The embodiments of the present application provide a mapping method, device, computer equipment and storage medium based on dynamic point cloud. The mapping method based on dynamic point cloud can be applied to a server. By combining the camera target detection algorithm, the detection frame range is accurately determined, and the point cloud data related to the target object is extracted. By determining the target point cloud cluster and its cluster type, the point cloud data can be more effectively classified and processed. Different point cloud strategies are adopted according to different cluster types, thereby improving the mapping effect of static point cloud. The server can be an independent server or a server cluster.
[0025] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.
[0026] See also Figure 1 , Figure 1 This is a schematic flow chart of a mapping method based on dynamic point cloud provided by the first embodiment of the present application. The mapping method based on dynamic point cloud can be applied to the server, and is used to accurately determine the detection frame range by combining the camera target detection algorithm, extract point cloud data related to the target object, and more effectively classify and process the point cloud data by determining the target point cloud cluster and its cluster type. Different point cloud strategies are adopted according to different cluster types, thereby improving the mapping effect of static point cloud.
[0027] like Figure 1 As shown, the mapping method based on dynamic point cloud specifically includes steps S10 to S30.
[0028] Step S10: determining a detection frame range of a target camera through a camera target detection algorithm and a preset object, and determining an initial point cloud data set within the detection frame range;
[0029] Specifically, the camera is used to detect the target (such as a person or a vehicle), and the YOLOv5s algorithm is used to generate the detection frame of the dynamic object. The camera target detection algorithm is used to identify the position of the object in the camera field of view, and the point cloud index within the detection frame is marked. The intensity attribute of the point cloud within the detection frame is marked as 1 as the preliminary screening basis for the dynamic point cloud.
[0030] The radar point cloud and the camera detection frame are spatially mapped. According to the calibration relationship between the camera coordinate system and the lidar coordinate system, the detection frame position is converted into the range index of the point cloud data. It is determined whether the center point of the point cloud falls within the detection frame, and the point cloud falling within the frame is marked as a dynamic point cloud.
[0031] Step S20, determining a target point cloud cluster and a cluster type of the target point cloud cluster from the initial point cloud data set, and determining an initial point cloud strategy according to the cluster type;
[0032] Specifically, dynamic object clustering is performed on point cloud data, and the improved DBSCAN algorithm is used to perform secondary clustering on the point cloud in the box to avoid interference from multiple small clusters in the box. The cluster with the largest number of clusters is selected as the main cluster, and the point clouds with close Z axes are added to this cluster.
[0033] According to the different types of dynamic objects, the point cloud clusters are classified and processed. If it is detected as a "person", the point cloud cluster is directly removed; if it is detected as a "vehicle", the motion state of the vehicle is determined. If the vehicle is in motion, the vehicle point cloud cluster is removed; if the vehicle is stationary, the point cloud cluster is retained and waits for subsequent updates.
[0034] Step S30: dynamically track the target point cloud cluster based on a preset filtering model, determine the current position of the target point cloud cluster, and determine the target point cloud mapping strategy according to the current position and the initial point cloud strategy.
[0035] Specifically, for dynamic objects in the detection frame, the Kalman filter tracker is initialized, and its position coordinates in the world coordinate system are calculated through the center point position of the point cloud cluster. The tracker is initialized to record the predicted position and updated position of the dynamic object.
[0036] For each detection point in each frame, the difference between its predicted position and the actual observed position is calculated. If the distance between the observed point and the predicted point is within the set range, the tracker is updated; otherwise, the object is marked as a new target and a new tracker is initialized. Targets that have not been updated for a long time are removed from the tracker to avoid interference from invalid points in the tracker.
[0037] In one embodiment, during the robot mapping process, a camera is usually used to capture an environmental image, and pre-processing operations such as denoising and enhancement are performed on the image to improve the accuracy of target detection. A target detection algorithm (such as YOLO, SSD, etc.) is applied to analyze the image to identify the target object. Based on the output of the target detection algorithm, the detection frame range of the target object is determined in the image.
[0038] Use sensors such as LiDAR or depth cameras to collect point cloud data of the environment, and perform preprocessing operations such as filtering and denoising on the collected point cloud data. Match the preprocessed point cloud data with the camera detection frame range to extract the initial point cloud data set within the detection frame range.
[0039] Apply clustering algorithms (such as DBSCAN, K-means, etc.) to the initial point cloud data set, divide the point cloud into multiple clusters, perform feature analysis on each cluster, determine the type of cluster (such as static cluster, dynamic cluster, etc.), and formulate corresponding initial point cloud processing strategies based on the cluster type, such as mapping static clusters and tracking dynamic clusters.
[0040] The target point cloud cluster is dynamically tracked using a preset filtering model (such as Kalman filtering, particle filtering, etc.), the state of the target point cloud cluster is updated in real time, and the current position of the target point cloud cluster is determined based on the output of the filtering model.
[0041] According to the current position and motion state of the target point cloud cluster, a target point cloud mapping strategy is formulated, and the tracked target point cloud cluster information is updated to the map to form or update the point cloud map. According to environmental changes and the dynamic characteristics of the target point cloud cluster, the mapping strategy is dynamically adjusted to adapt to environmental changes.
[0042] The present embodiment discloses a mapping method, apparatus, computer equipment and storage medium based on dynamic point clouds, wherein the mapping method based on dynamic point clouds includes determining the detection frame range of the target camera through a camera target detection algorithm and a preset object, and determining an initial point cloud data set within the detection frame range; determining a target point cloud cluster and a cluster type of the target point cloud cluster from the initial point cloud data set, and determining an initial point cloud strategy according to the cluster type; dynamically tracking the target point cloud cluster based on a preset filtering model, determining the current position of the target point cloud cluster, and determining a target point cloud mapping strategy according to the current position and the initial point cloud strategy. In the above manner, the present application accurately determines the detection frame range and extracts point cloud data related to the target object by combining a camera target detection algorithm. By determining the target point cloud cluster and its cluster type, the point cloud data can be more effectively classified and processed, and different point cloud strategies can be adopted according to different cluster types, thereby improving the mapping effect of static point clouds.
[0043] See also Figure 2 , Figure 2 This is a schematic flow chart of a mapping method based on dynamic point cloud provided in the second embodiment of the present application. This mapping method based on dynamic point cloud can be applied to a server, and is used to accurately determine the detection frame range by combining the camera target detection algorithm, extract point cloud data related to the target object, and determine the target point cloud cluster and its type from the initial point cloud data set, so that the point cloud data processing is more targeted, and the efficiency and effect of data processing are improved. The Kalman filter model is used to dynamically track the target point cloud cluster, and the position information of the target point cloud cluster can be updated in real time, which improves the mapping effect of the static point cloud.
[0044] based on Figure 1 The embodiment shown, this embodiment Figure 2 As shown, step S30 includes steps S301 to S304.
[0045] Step S301, dynamically tracking the target point cloud cluster through the Kalman filter model to determine the current position of the target point cloud cluster;
[0046] Step S302, predicting the predicted position of the target point cloud cluster;
[0047] Step S303: determining a built-in tracker state of the Kalman filter model according to the current position and the predicted position;
[0048] Step S304: determining the target point cloud mapping strategy according to the state of the built-in tracker.
[0049] In a specific embodiment, the state variables of the Kalman filter model, including position, speed, etc., are predetermined, initial estimated values are set for the state variables, the initial uncertainty (covariance matrix) of the state estimate is set, and the model defines the covariance matrix of process noise and measurement noise.
[0050] The prediction equation of the Kalman filter model is used to predict the next state of the target point cloud cluster and the uncertainty of the state estimation (covariance matrix).
[0051] The preprocessed point cloud data is input into the Kalman filter model as the measurement value, and the Kalman gain is calculated to weigh the predicted state and the measurement value. The state estimate of the target point cloud cluster is updated according to the Kalman gain and the measurement value, the uncertainty of the state estimate is updated, and the current position of the target point cloud cluster is extracted from the updated state estimate.
[0052] The present embodiment discloses a mapping method, apparatus, computer equipment and storage medium based on dynamic point cloud. The mapping method based on dynamic point cloud includes determining the detection frame range of the target camera through a camera target detection algorithm and a preset object, and determining an initial point cloud data set within the detection frame range; determining a target point cloud cluster and a cluster type of the target point cloud cluster from the initial point cloud data set, and determining an initial point cloud strategy according to the cluster type; dynamically tracking the target point cloud cluster through the Kalman filter model to determine the current position of the target point cloud cluster; predicting the predicted position of the target point cloud cluster; determining the built-in tracker state of the Kalman filter model according to the current position and the predicted position; and determining the target point cloud mapping strategy according to the built-in tracker state. Through the above method, the present application combines the camera target detection algorithm to accurately determine the detection frame range, extract the point cloud data related to the target object, determine the target point cloud cluster and its type from the initial point cloud data set, so that the point cloud data processing is more targeted, and the efficiency and effect of data processing are improved. The Kalman filter model is used to dynamically track the target point cloud cluster, and the position information of the target point cloud cluster can be updated in real time, thereby improving the mapping effect of the static point cloud.
[0053] based on Figure 2 In the embodiment shown, in this embodiment, step S303 includes:
[0054] Calculating a position error between the current position and the predicted position;
[0055] If the position error is within a preset error range, updating the built-in tracker state according to the predicted position;
[0056] If the position error exceeds the preset error range, the built-in tracker state is initialized.
[0057] In one embodiment, the current position and predicted position of the target point cloud cluster are extracted from the Kalman filter model, and the difference between the current position and the predicted position, i.e., the position error, is calculated. It is determined whether the position error is within a preset error range, and a preset range of position error is set according to system requirements and environmental characteristics, and the calculated position error is compared with the preset error range.
[0058] If the position error is within the preset error range, it means that the tracker is in good condition. The predicted position can be used to update the built-in tracker state and adjust the parameters of the Kalman filter model, such as state variables and covariance matrix, according to the predicted position.
[0059] If the position error exceeds the preset error range, it means that the tracker state may have deviated and needs to be initialized, that is, resetting the parameters of the Kalman filter model, such as state variables, covariance matrix, etc., to restore the initial state of the tracker.
[0060] The present embodiment discloses a mapping method, apparatus, computer equipment and storage medium based on dynamic point cloud. The mapping method based on dynamic point cloud includes determining the detection frame range of the target camera through a camera target detection algorithm and a preset object, and determining an initial point cloud data set within the detection frame range; determining a target point cloud cluster and a cluster type of the target point cloud cluster from the initial point cloud data set, and determining an initial point cloud strategy according to the cluster type; dynamically tracking the target point cloud cluster through the Kalman filter model to determine the current position of the target point cloud cluster; predicting the predicted position of the target point cloud cluster; calculating the position error between the current position and the predicted position; if the position error is within a preset error range, updating the built-in tracker state according to the predicted position; if the position error exceeds the preset error range, initializing the built-in tracker state; and determining the target point cloud mapping strategy according to the built-in tracker state. Through the above method, the present application combines the camera target detection algorithm to accurately determine the detection frame range and extract the point cloud data related to the target object. By determining the target point cloud cluster and its cluster type, the point cloud data can be more effectively classified and processed. Different point cloud strategies are adopted according to different cluster types, thereby improving the mapping effect of static point clouds.
[0061] based on Figure 1 In the embodiment shown, in this embodiment, step S20 includes:
[0062] Generate at least one dynamic point cloud cluster by using a preset clustering algorithm and the initial point cloud data set and determine the target point cloud cluster from each of the dynamic point cloud clusters;
[0063] A cluster type corresponding to the target point cloud cluster is determined, and the initial point cloud strategy of the target point cloud cluster is set according to the cluster type.
[0064] In one embodiment, according to the characteristics of the point cloud data and application requirements, a suitable clustering algorithm (such as DBSCAN, K-means, Mean Shft, etc.) is selected, and necessary parameters are set for the selected clustering algorithm, such as search radius, minimum number of points, cluster center, etc. The preprocessed point cloud data is input into the clustering algorithm to generate multiple dynamic point cloud clusters.
[0065] The generated dynamic point cloud clusters are evaluated, and which clusters are target point cloud clusters are determined based on their size, shape, density and other characteristics. Point cloud clusters related to the mission objectives are selected from the evaluation results as target point cloud clusters, and feature analysis is performed on the target point cloud clusters, such as speed, acceleration, shape change, etc. According to the analysis results, the target point cloud clusters are classified into different cluster types, such as static clusters, dynamic clusters, pedestrian clusters, vehicle clusters, etc.
[0066] According to the cluster type of the target point cloud cluster, the corresponding initial point cloud strategy is defined. For static clusters, a high-precision mapping strategy is adopted; for dynamic clusters, a dynamic tracking and prediction strategy is adopted. According to the strategy requirements, the sensor parameters, data processing algorithm parameters, etc. are adjusted to adapt to different types of point cloud clusters.
[0067] The present embodiment discloses a mapping method, device, computer equipment and storage medium based on dynamic point cloud, wherein the mapping method based on dynamic point cloud includes determining the detection frame range of the target camera through the camera target detection algorithm and the preset object, and determining the initial point cloud data set within the detection frame range; generating at least one dynamic point cloud cluster and determining the target point cloud cluster from each of the dynamic point cloud clusters through the preset clustering algorithm and the initial point cloud data set; determining the cluster type corresponding to the target point cloud cluster, and setting the initial point cloud strategy of the target point cloud cluster according to the cluster type; dynamically tracking the target point cloud cluster based on the preset filtering model, determining the current position of the target point cloud cluster, and determining the target point cloud mapping strategy according to the current position and the initial point cloud strategy. In the above manner, the present application accurately determines the detection frame range and extracts the point cloud data related to the target object by combining the camera target detection algorithm. By determining the target point cloud cluster and its cluster type, the point cloud data can be more effectively classified and processed, and different point cloud strategies are adopted according to different cluster types, thereby improving the mapping effect of static point cloud.
[0068] In a more preferred embodiment, at least one dynamic point cloud cluster is generated by using a preset clustering algorithm and the initial point cloud data set, and the target point cloud cluster is determined from each of the dynamic point cloud clusters, including:
[0069] Performing dynamic clustering processing on each initial point cloud data in the initial point cloud data set by using the preset clustering algorithm to generate at least one dynamic point cloud cluster;
[0070] The point cloud cluster with the most point cloud data among the dynamic point cloud clusters is determined as the target point cloud cluster.
[0071] Specifically, the preprocessed point cloud data is input into the clustering algorithm to perform a clustering operation. The clustering algorithm groups the point cloud data according to set parameters to generate at least one dynamic point cloud cluster.
[0072] Extract features from each generated dynamic point cloud cluster, such as cluster size, shape, density, etc., and calculate the number of point cloud data in each dynamic point cloud cluster. Compare the number of point cloud data in each dynamic point cloud cluster, and determine the point cloud cluster with the most point cloud data as the target point cloud cluster.
[0073] In a more preferred embodiment, setting the initial point cloud strategy of the target point cloud cluster according to the cluster type includes:
[0074] In a case where the cluster type is a human point cloud cluster, determining the initial point cloud strategy to delete the human point cloud cluster;
[0075] In a case where the cluster type is a vehicle point cloud cluster, the initial point cloud strategy is determined according to a motion state of the vehicle point cloud cluster.
[0076] Specifically, the dynamic point clouds (points with intensity of 1) within the detection box are set to 0 and removed from the point cloud data;
[0077] Dynamic point cloud clusters that have been clustered and classified are also removed to ensure the purity of static point clouds, retain useful point cloud data, optimize map accuracy, and reduce the interference of dynamic objects on mapping results.
[0078] Based on any of the above embodiments, in this embodiment, step S10 includes:
[0079] Identifying the position information of the preset object in the field of view of the target camera by the camera target detection algorithm, and determining the detection frame range based on the position information;
[0080] All point cloud data within the detection frame are determined as the initial point cloud data set.
[0081] See also Figure 3 , Figure 3 The embodiment of the present application provides a schematic block diagram of a mapping device based on a dynamic point cloud, and the mapping device based on a dynamic point cloud is used to execute the aforementioned mapping method based on a dynamic point cloud. The mapping device based on a dynamic point cloud can be configured on a server.
[0082] like Figure 3 As shown, the mapping device 400 based on dynamic point cloud includes:
[0083] An initial point cloud data set determination module 410 is used to determine a detection frame range of a target camera through a camera target detection algorithm and a preset object, and to determine an initial point cloud data set within the detection frame range;
[0084] An initial point cloud strategy determination module 420, configured to determine a target point cloud cluster and a cluster type of the target point cloud cluster from the initial point cloud data set, and determine an initial point cloud strategy according to the cluster type;
[0085] The target point cloud mapping strategy determination module 430 is used to dynamically track the target point cloud cluster based on a preset filtering model, determine the current position of the target point cloud cluster, and determine the target point cloud mapping strategy according to the current position and the initial point cloud strategy.
[0086] Furthermore, the target point cloud mapping strategy determination module 430 includes:
[0087] A current position determination unit, configured to dynamically track the target point cloud cluster through the Kalman filter model to determine the current position of the target point cloud cluster;
[0088] A predicted position determination unit, used to predict the predicted position of the target point cloud cluster;
[0089] a built-in tracker state determination unit, configured to determine a built-in tracker state of the Kalman filter model according to the current position and the predicted position;
[0090] The target point cloud mapping strategy determination unit is used to determine the target point cloud mapping strategy according to the state of the built-in tracker.
[0091] Furthermore, the built-in tracker state determination unit includes:
[0092] a position error determination subunit, configured to calculate a position error between the current position and the predicted position;
[0093] a built-in tracker state updating subunit, configured to update the built-in tracker state according to the predicted position if the position error is within a preset error range;
[0094] The built-in tracker state initialization subunit is used to initialize the built-in tracker state if the position error exceeds the preset error range.
[0095] Furthermore, the initial point cloud strategy determination module 420 includes:
[0096] A target point cloud cluster determination unit, configured to generate at least one dynamic point cloud cluster and determine the target point cloud cluster from each of the dynamic point cloud clusters by using a preset clustering algorithm and the initial point cloud data set;
[0097] The initial point cloud strategy determining unit is used to determine the cluster type corresponding to the target point cloud cluster, and set the initial point cloud strategy of the target point cloud cluster according to the cluster type.
[0098] Furthermore, the target point cloud cluster determination unit includes:
[0099] A dynamic point cloud cluster generating subunit, configured to perform dynamic clustering processing on each initial point cloud data in the initial point cloud data set by using the preset clustering algorithm to generate at least one dynamic point cloud cluster;
[0100] The target point cloud cluster determination subunit is used to determine the point cloud cluster with the most point cloud data in each of the dynamic point cloud clusters as the target point cloud cluster.
[0101] Furthermore, the initial point cloud strategy determination unit includes:
[0102] A first strategy determination subunit is used to determine the initial point cloud strategy as deleting the human point cloud cluster when the cluster type is a human point cloud cluster;
[0103] The second strategy determination subunit is used to determine the initial point cloud strategy according to the motion state of the vehicle point cloud cluster when the cluster type is a vehicle point cloud cluster.
[0104] Furthermore, the initial point cloud data set determination module 410 includes:
[0105] A detection frame range determination unit, configured to identify position information of the preset object in the field of view of the target camera through the camera target detection algorithm, and determine the detection frame range based on the position information;
[0106] The initial point cloud data set determining unit is used to determine all point cloud data within the detection frame as the initial point cloud data set.
[0107] It should be noted that those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described device and each module can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.
[0108] The above-mentioned device can be implemented in the form of a computer program. Figure 4 Runs on the computer device shown.
[0109] See also Figure 4 , Figure 4 1 is a schematic block diagram of a computer device provided in an embodiment of the present application. The computer device may be a server.
[0110] See also Figure 4 The computer device includes a processor, a memory and a network interface connected through a system bus, wherein the memory may include a non-volatile storage medium and an internal memory.
[0111] The non-volatile storage medium can store an operating system and a computer program. The computer program includes program instructions, and when the program instructions are executed, the processor can execute any mapping method based on dynamic point clouds.
[0112] The processor is used to provide computing and control capabilities and support the operation of the entire computer equipment.
[0113] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor, the processor can execute any mapping method based on dynamic point clouds.
[0114] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0115] It should be understood that the processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.
[0116] In one embodiment, the processor is used to run a computer program stored in the memory to implement the following steps:
[0117] Determine the detection frame range of the target camera through the camera target detection algorithm and the preset object, and determine the initial point cloud data set within the detection frame range;
[0118] Determining a target point cloud cluster and a cluster type of the target point cloud cluster from the initial point cloud data set, and determining an initial point cloud strategy according to the cluster type;
[0119] The target point cloud cluster is dynamically tracked based on a preset filtering model to determine the current position of the target point cloud cluster, and the target point cloud mapping strategy is determined according to the current position and the initial point cloud strategy.
[0120] In one embodiment, the target point cloud cluster is dynamically tracked based on a preset filtering model to determine the current position of the target point cloud cluster, and the target point cloud mapping strategy is determined according to the current position and the initial point cloud strategy to achieve:
[0121] Dynamically tracking the target point cloud cluster through the Kalman filter model to determine the current position of the target point cloud cluster;
[0122] Predicting a predicted position of the target point cloud cluster;
[0123] Determining a built-in tracker state of the Kalman filter model according to the current position and the predicted position;
[0124] The target point cloud mapping strategy is determined according to the built-in tracker state.
[0125] In one embodiment, the built-in tracker state of the Kalman filter model is determined according to the current position and the predicted position to achieve:
[0126] Calculating a position error between the current position and the predicted position;
[0127] If the position error is within a preset error range, updating the built-in tracker state according to the predicted position;
[0128] If the position error exceeds the preset error range, the built-in tracker state is initialized.
[0129] In one embodiment, a target point cloud cluster and a cluster type of the target point cloud cluster are determined from the initial point cloud data set, and an initial point cloud strategy is determined according to the cluster type, for achieving:
[0130] Generate at least one dynamic point cloud cluster by using a preset clustering algorithm and the initial point cloud data set and determine the target point cloud cluster from each of the dynamic point cloud clusters;
[0131] A cluster type corresponding to the target point cloud cluster is determined, and the initial point cloud strategy of the target point cloud cluster is set according to the cluster type.
[0132] In one embodiment, at least one dynamic point cloud cluster is generated by using a preset clustering algorithm and the initial point cloud data set, and the target point cloud cluster is determined from each of the dynamic point cloud clusters, so as to achieve:
[0133] Performing dynamic clustering processing on each initial point cloud data in the initial point cloud data set by using the preset clustering algorithm to generate at least one dynamic point cloud cluster;
[0134] The point cloud cluster with the most point cloud data among the dynamic point cloud clusters is determined as the target point cloud cluster.
[0135] In one embodiment, the initial point cloud strategy of the target point cloud cluster is set according to the cluster type to achieve:
[0136] In a case where the cluster type is a human point cloud cluster, determining the initial point cloud strategy to delete the human point cloud cluster;
[0137] In a case where the cluster type is a vehicle point cloud cluster, the initial point cloud strategy is determined according to a motion state of the vehicle point cloud cluster.
[0138] In one embodiment, the detection frame range of the target camera is determined by a camera target detection algorithm and a preset object, and an initial point cloud data set within the detection frame range is determined to achieve:
[0139] Identifying the position information of the preset object in the field of view of the target camera by the camera target detection algorithm, and determining the detection frame range based on the position information;
[0140] All point cloud data within the detection frame are determined as the initial point cloud data set.
[0141] A computer-readable storage medium is also provided in an embodiment of the present application, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and the processor executes the program instructions to implement any one of the dynamic point cloud-based mapping methods provided in the embodiments of the present application.
[0142] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiment, such as a hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a smart memory card (Smart Medi a Card, SMC), a secure digital (Secure Digital, SD) card, a flash memory card (Flash Card), etc., equipped on the computer device.
[0143] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.
Claims
1. A mapping method based on dynamic point cloud, characterized in that: include: Determine the detection frame range of the target camera through the camera target detection algorithm and the preset object, and determine the initial point cloud data set within the detection frame range; Determining a target point cloud cluster and a cluster type of the target point cloud cluster from the initial point cloud data set, and determining an initial point cloud strategy according to the cluster type; The target point cloud cluster is dynamically tracked based on a preset filtering model to determine the current position of the target point cloud cluster, and the target point cloud mapping strategy is determined according to the current position and the initial point cloud strategy.
2. The mapping method based on dynamic point cloud according to claim 1, characterized in that: The preset filtering model includes a Kalman filtering model, and the target point cloud cluster is dynamically tracked based on the preset filtering model to determine the current position of the target point cloud cluster, and the target point cloud mapping strategy is determined according to the current position and the initial point cloud strategy, including: Dynamically tracking the target point cloud cluster through the Kalman filter model to determine the current position of the target point cloud cluster; Predicting a predicted position of the target point cloud cluster; Determining a built-in tracker state of the Kalman filter model according to the current position and the predicted position; The target point cloud mapping strategy is determined according to the built-in tracker state.
3. The mapping method based on dynamic point cloud according to claim 2, characterized in that: The step of determining the built-in tracker state of the Kalman filter model according to the current position and the predicted position includes: Calculating a position error between the current position and the predicted position; If the position error is within a preset error range, updating the built-in tracker state according to the predicted position; If the position error exceeds the preset error range, the built-in tracker state is initialized.
4. The mapping method based on dynamic point cloud according to claim 1, characterized in that: The step of determining a target point cloud cluster and a cluster type of the target point cloud cluster from the initial point cloud data set, and determining an initial point cloud strategy according to the cluster type, includes: Generate at least one dynamic point cloud cluster by using a preset clustering algorithm and the initial point cloud data set and determine the target point cloud cluster from each of the dynamic point cloud clusters; A cluster type corresponding to the target point cloud cluster is determined, and the initial point cloud strategy of the target point cloud cluster is set according to the cluster type.
5. The mapping method based on dynamic point cloud according to claim 4, characterized in that: The generating at least one dynamic point cloud cluster by using a preset clustering algorithm and the initial point cloud data set and determining the target point cloud cluster from each of the dynamic point cloud clusters comprises: Performing dynamic clustering processing on each initial point cloud data in the initial point cloud data set by using the preset clustering algorithm to generate at least one dynamic point cloud cluster; The point cloud cluster with the most point cloud data among the dynamic point cloud clusters is determined as the target point cloud cluster.
6. The mapping method based on dynamic point cloud according to claim 5, characterized in that: The step of setting the initial point cloud strategy of the target point cloud cluster according to the cluster type includes: In a case where the cluster type is a human point cloud cluster, determining the initial point cloud strategy to delete the human point cloud cluster; In a case where the cluster type is a vehicle point cloud cluster, the initial point cloud strategy is determined according to a motion state of the vehicle point cloud cluster.
7. The mapping method based on dynamic point cloud according to any one of claims 1 to 6, characterized in that: The step of determining the detection frame range of the target camera by using a camera target detection algorithm and a preset object, and determining an initial point cloud data set within the detection frame range, includes: Identifying the position information of the preset object in the field of view of the target camera by the camera target detection algorithm, and determining the detection frame range based on the position information; All point cloud data within the detection frame are determined as the initial point cloud data set.
8. A mapping device based on dynamic point cloud, characterized in that: include: An initial point cloud data set determination module is used to determine the detection frame range of the target camera through a camera target detection algorithm and a preset object, and to determine an initial point cloud data set within the detection frame range; An initial point cloud strategy determination module, used to determine a target point cloud cluster and a cluster type of the target point cloud cluster from the initial point cloud data set, and determine an initial point cloud strategy according to the cluster type; The target point cloud mapping strategy determination module is used to dynamically track the target point cloud cluster based on a preset filtering model, determine the current position of the target point cloud cluster, and determine the target point cloud mapping strategy according to the current position and the initial point cloud strategy.
9. A computer device, characterized in that: The computer device includes a memory and a processor; The memory is used to store computer programs; The processor is used to execute the computer program and implement the dynamic point cloud-based mapping method according to any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor implements the dynamic point cloud-based mapping method according to any one of claims 1 to 7.