Point cloud data thinning processing and change detection method and device, storage medium and computer program product

Abstract

Embodiments of the present application provide a point cloud data thinning processing and change detection method, device, storage medium and computer program product. In the embodiments of the present application, the importance of each point cloud point in the point cloud data is divided according to the application scene requirement, and one or more levels of thinning processing are performed on the point cloud data according to the application scene requirement, thereby realizing intelligent hierarchical processing of the point cloud data. The thinning processing greatly reduces the data amount of the point cloud data, so that the point cloud data better adapts to the application scene, and improves the use efficiency and application effect of the point cloud data in the application scene.

Description

Technical Field

[0001] This application relates to the field of point cloud data processing technology, and in particular to a method, device, storage medium, and computer program product for thinning point cloud data and detecting changes. Background Technology

[0002] In today's rapidly developing digital age, point cloud data, as an important type of three-dimensional spatial data, is widely used in various fields such as urban planning, environmental monitoring, architectural design, virtual reality, intelligent transportation, and autonomous driving. However, the massive volume and complexity of point cloud data pose challenges to real-time processing and analysis, resulting in low efficiency in downstream applications. Summary of the Invention

[0003] This application provides a method, device, storage medium, and computer program product for thinning point cloud data and detecting changes, in order to improve the efficiency of point cloud data in downstream application scenarios.

[0004] This application provides a point cloud data thinning method, comprising: acquiring point cloud data, the point cloud data including data of multiple point cloud points; determining the importance metric value corresponding to each of the multiple point cloud points according to the application scenario requirements corresponding to the point cloud data; determining the numerical range of the importance metric value corresponding to at least one level according to the application scenario requirements; performing at least one level of thinning on the point cloud data according to the importance metric value corresponding to each of the multiple point cloud points and the numerical range of the importance metric value corresponding to at least one level, to obtain point cloud data thinned at at least one level, wherein the importance metric value of the point cloud points in the point cloud data thinned at different levels is different, and the point cloud data thinned at each level includes point cloud points whose importance metric value falls within the numerical range of the importance metric value corresponding to that level.

[0005] This application embodiment also provides a method for detecting changes in point cloud data, including: acquiring first point cloud data at a first time point and second point cloud data at a second time point, wherein the first time point is earlier than the second time point; performing at least one level of thinning processing on the first point cloud data and the second point cloud data respectively to obtain at least one level of third point cloud data corresponding to the first point cloud data and at least one level of fourth point cloud data corresponding to the second point cloud data; for the target third point cloud data and the target fourth point cloud data at the target level, dividing the target third point cloud data into multiple first voxel grids and dividing the target fourth point cloud data into multiple second voxel grids; the target level is any one of at least one level, the target third point cloud data is any one of the third point cloud data at at least one level, and the target fourth point cloud data is any one of the fourth point cloud data at at least one level; for the target first voxel grid and the target second voxel grid representing the same spatial range, determining whether the target first voxel grid has changed relative to the target second voxel grid based on the existence of point cloud points in the target first voxel grid and the existence of point cloud points in the target second voxel grid.

[0006] This application also provides an electronic device, including: a memory and a processor; the memory for storing a computer program; the processor coupled to the memory for executing the computer program to perform steps in a point cloud data thinning or change detection method.

[0007] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to implement steps in point cloud data thinning or change detection methods.

[0008] This application also provides a computer program product, including a computer program / instruction, which, when executed by a processor, enables the processor to implement the steps in the point cloud data thinning process or change detection method.

[0009] In this embodiment, based on the application scenario requirements corresponding to the point cloud data, the importance metric values ​​corresponding to multiple point cloud points are determined; the numerical range of the importance metric values ​​corresponding to at least one level is determined according to the application scenario requirements; based on the importance metric values ​​corresponding to the multiple point cloud points and the numerical range of the importance metric values ​​corresponding to at least one level, the point cloud data is subjected to at least one level of thinning processing to obtain point cloud data thinned at at least one level. The importance metric values ​​of the point cloud points in the thinned point cloud data at different levels are different, and the thinned point cloud data at each level includes point cloud points whose importance metric values ​​fall within the numerical range of the importance metric values ​​corresponding to that level. This achieves the goal of classifying the importance of each point cloud point in the point cloud data according to the application scenario requirements, and performing one or more levels of thinning processing on the point cloud data according to the application scenario requirements. This realizes intelligent hierarchical processing of point cloud data, and the thinning processing greatly reduces the data volume of the point cloud data, making the point cloud data better adaptable to the application scenario and improving the efficiency and application effect of the point cloud data in the application scenario. Attached Figure Description

[0010] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0011] Figure 1 This is an example application scenario diagram;

[0012] Figure 2 A flowchart illustrating a point cloud data thinning process provided in this application embodiment;

[0013] Figure 3 A flowchart illustrating a point cloud data change detection method provided in this application embodiment;

[0014] Figure 4 This is an example of full point cloud data;

[0015] Figure 5 This is an example of thinned point cloud data;

[0016] Figure 6 The origin cloud is used to illustrate the actual area of ​​change.

[0017] Figure 7 The origin cloud of the detected change region is marked as an example;

[0018] Figure 8 A schematic diagram of a point cloud data thinning processing device provided in an embodiment of this application;

[0019] Figure 9A schematic diagram of a point cloud data change detection device provided in an embodiment of this application;

[0020] Figure 10 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] In the embodiments of this application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the access relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone, where A and B can be singular or plural. In the textual description of this application, the character " / " generally indicates that the preceding and following associated objects have an "or" relationship. Furthermore, in the embodiments of this application, "first," "second," "third," etc., are only used to distinguish the content of different objects and have no other special meaning.

[0023] The technical solutions of this application and how they solve the aforementioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The technical solutions provided by each embodiment of this application are described in detail below with reference to the accompanying drawings.

[0024] Figure 1 This is an example application scenario diagram. In the field of autonomous driving, LiDAR is an indispensable part of autonomous vehicles. By providing high-precision environmental perception data, it greatly improves the safety and intelligence level of vehicles. Specifically, LiDAR generates high-precision point cloud data by emitting laser pulses and receiving the reflected signals. Point cloud data can depict the environment around the vehicle in detail, including objects along the road, road boundaries, obstacle locations, pedestrians, and other vehicles. Objects along the road include, but are not limited to, power equipment, base stations, or buildings. High-precision environmental perception data is crucial for autonomous vehicles, helping them accurately locate themselves and understand their surroundings. This enables autonomous vehicles to drive safely and efficiently in complex and changing road environments, effectively avoiding potential risks and improving the overall driving experience.

[0025] For specific examples, see Figure 1 As shown in ①, the lidar collects point cloud data of the environment surrounding the vehicle and sends it to the server; see also Figure 1 As shown in ②, the server processes point cloud data to generate a 3D (three-dimensional) map of the vehicle's surroundings; see also Figure 1 As shown in ③, the server sends a 3D map of the vehicle's surroundings to the autonomous vehicle. This allows the autonomous vehicle to make more accurate and safer driving decisions based on this high-precision map information, ensuring that the autonomous vehicle can understand its surrounding environment in real time and accurately, thereby effectively improving driving safety and efficiency.

[0026] In practical applications, the point cloud data collected by LiDAR can be quite large and redundant, leading to high storage costs and low utilization efficiency. Therefore, point cloud data collected by LiDAR can be thinned on demand to reduce its volume, lower storage costs, and improve utilization efficiency. For example, the point cloud data collected by LiDAR can be thinned and stored locally in the autonomous vehicle. The server then loads this locally stored thinned point cloud data. Because the thinned point cloud data is smaller, it consumes fewer storage resources and loads faster, thus effectively improving the server's 3D map creation efficiency.

[0027] It should be noted that, Figure 1 The application scenario shown is merely an exemplary scenario, and the embodiments of this application do not limit the application scenarios. The embodiments of this application do not... Figure 1 The included equipment is not limited, nor is it restricted. Figure 1 The positional relationships between the devices are defined.

[0028] Figure 2 A flowchart illustrating a point cloud data thinning method provided in an embodiment of this application. See also... Figure 2 The method may include the following steps:

[0029] 201. Obtain point cloud data, which includes data from multiple point cloud points.

[0030] In practical applications, point cloud data collected in real time by a sensing system can be directly acquired. Sensing systems include, but are not limited to, LiDAR, binocular stereo vision systems, or binocular structured light depth cameras. Of course, the point cloud data collected by the sensing system can be stored in a storage device, and point cloud data can be retrieved from the storage device; there are no restrictions on this.

[0031] Specifically, point cloud data is a collection of data consisting of a large number of points. Here, the points in the point cloud data are referred to as point cloud points, and each point cloud point typically represents a specific spatial location in the real world. Point cloud data includes data from multiple point cloud points, and the data of a point cloud point includes, but is not limited to: the three-dimensional spatial coordinates and attribute information of the point cloud point; the attribute information of the point cloud point includes, but is not limited to: the color information, reflection intensity, and normal vector of the point cloud point.

[0032] 202. Based on the application scenario requirements corresponding to the point cloud data, determine the importance metric value corresponding to each of the multiple point cloud points.

[0033] Specifically, the application scenario requirements for point cloud data refer to the needs of application scenarios that use point cloud data for related processing. In practice, different application scenarios have diverse requirements for point cloud data. For example, the requirements for data processing speed, data volume, or data type vary depending on the application scenario.

[0034] In practical applications, the importance metric value of each point cloud point can be determined according to the application scenario requirements corresponding to the point cloud data, thereby enabling the point cloud data to better adapt to the application scenario and improve the application effect of the point cloud data in the application scenario.

[0035] In this embodiment, the importance metric of point cloud points reflects their importance and can be understood as a quantification of their importance. Based on the importance metric, decisions can be made regarding which point cloud points can be retained and which can be removed. In practical applications, a higher importance metric generally indicates higher point cloud point importance; conversely, a lower importance metric generally indicates higher point cloud point importance, and there is no restriction on this.

[0036] In practical applications, the requirements of the application scenario may indicate that the relevant information of the point cloud points is related to the importance metric of the point cloud points. The relevant information of the point cloud points includes, but is not limited to: the 3D coordinates of the point cloud points, the distance information of the point cloud points, the color information of the point cloud points, the reflection information of the point cloud points, etc. For example, in autonomous driving scenarios, the distance information of the point cloud points can be determined based on the 3D coordinates of the point cloud points and the 3D coordinates of the vehicle.

[0037] Points with smaller distance information are closer to the vehicle, and these closer points are more important because nearby obstacles have a greater impact on vehicle safety. For example, in 3D modeling, point cloud data is divided into multiple regions, and the color change rate of the point clouds in each region is determined based on the color information of the points in that region. Point clouds in regions with higher color change rates are more important than those in regions with uniform color because these regions contain more visual details. Similarly, in autonomous driving scenarios, points with high reflectivity may represent hard objects (such as metal or glass), and these points are more important than points with low reflectivity (such as plants or soil).

[0038] In some optional embodiments, point cloud data can be processed more precisely hierarchically based on the category of point cloud points (referred to herein as point category). Hierarchical processing can be understood as dividing point cloud data into different levels to meet the needs of flexible and diverse application scenarios. Through hierarchical processing, a large amount of point cloud data can be divided into multiple small subsets, each containing point cloud points with similar characteristics. This can reduce unnecessary computational and storage overhead and improve data processing efficiency.

[0039] Based on the above, and according to the application scenario requirements corresponding to the point cloud data, the implementation method for determining the importance metric values ​​corresponding to multiple point cloud points is as follows: if the application scenario requirements indicate that the point category of the point cloud point is related to the importance metric value of the point cloud point, then the correspondence between multiple different point categories and multiple different importance metrics is determined according to the application scenario requirements; for any target point cloud point among multiple point cloud points, the importance metric value of the target point cloud point is determined according to the point category and correspondence of the target point cloud point.

[0040] In practical applications, the correlation between the point cloud point category and its importance metric can be flexibly determined based on the application scenario requirements, without any restrictions. For example, if the application scenario requires high-precision point cloud data, then the application scenario requirement indicates a correlation between the point cloud point category and its importance metric. Specifically, in urban planning scenarios, high-precision point cloud data is needed to reconstruct detailed models of buildings, roads, and other infrastructure. Alternatively, if the application scenario requires focus on specific categories of point cloud data, then the application scenario requirement indicates a correlation between the point cloud point category and its importance metric. For example, in autonomous driving scenarios, focus is needed on specific categories of point cloud points such as obstacles, pedestrians, and traffic signs on roads, while neglecting point cloud points in background areas (such as the sky and distant buildings). Another example is in industrial inspection scenarios where customers explicitly require high-precision inspection of specific components. Based on these requirements, the focus is on retaining the point cloud points of these specific components while reducing point cloud points in other areas; in this case, the application scenario requirement indicates a correlation between the point cloud point category and its importance metric.

[0041] In this embodiment, the correspondence between multiple different point categories and multiple different importance metrics is determined according to the application scenario requirements. Specifically, different application scenarios focus on different point categories. The correspondence between multiple different point categories and multiple different importance metrics is flexibly determined according to the requirements of different application scenarios, thereby enabling point cloud data to better adapt to the application scenario and improve the application effect of point cloud data in the application scenario. For example, in an urban planning scenario, the point categories of point cloud points include, but are not limited to: buildings, roads, gas stations, trees, etc. They are sorted from highest to lowest importance as follows: buildings, roads, gas stations, and trees. The importance metric value corresponding to buildings is 1, that of roads is 2, that of gas stations is 3, and that of trees is 4. The smaller the importance metric value, the higher the importance.

[0042] In this embodiment, after determining the correspondence between multiple different point categories and multiple different importance metrics, for any target point cloud point among multiple point cloud points, the importance metric of the target point cloud point can be determined according to the point category and correspondence of the target point cloud point.

[0043] In some optional embodiments, the current point category of multiple point clouds in the point cloud data may not meet the application scenario requirements, necessitating fine-grained classification of the multiple point clouds in the point cloud data. For example, the current point category of the point cloud is "building," but the application scenario requires a finer-grained point category that can distinguish between more specific categories such as the roof, walls, and windows of a building.

[0044] In practical applications, there are no restrictions on how the current point category of a point cloud is determined. For example, the point cloud data can be targeted using various information such as the three-dimensional coordinates, color information, or reflection intensity information of the point cloud points to obtain the current point category of each point cloud point.

[0045] Based on the above, in order to better adapt point cloud data to application scenarios and improve its application effectiveness, the importance metric of the target point cloud point is determined according to its point category and correspondence as follows: if the current point category of the target point cloud point does not meet the application scenario requirements, a machine learning model is invoked to classify the point cloud data, obtaining updated point categories for multiple point cloud points; the importance metric of the target point cloud point is then determined based on the updated point category and correspondence of the target point cloud point.

[0046] It is worth noting that the machine learning model is trained using a large amount of training data. This machine learning model has classification capabilities and can meet the classification requirements of the application scenario. Optionally, to improve the generalization performance and accuracy of the machine learning model, firstly, the model is trained using training data from different application scenarios to obtain a pre-trained machine learning model with classification capabilities. Then, for different application scenarios, the pre-trained machine learning model is fine-tuned using the training data of that application scenario to obtain a machine learning model suitable for that application scenario, thereby improving the effectiveness of the machine learning model in downstream application scenarios. The training data includes collected point cloud data, where each point in the point cloud data is labeled with its category. During model training, the point cloud data is input into the machine learning model to be trained, and the classification results output by the machine learning model are obtained. The loss function is determined based on the category of each point in the classification results and the labeled category of each point. The model parameters of the machine learning model are adjusted according to the loss function, and the model training operation is repeated until a converged machine learning model is obtained.

[0047] In some optional embodiments, to improve the accuracy and reliability of the importance measurement of point cloud points, the importance measurement value determined based on the point cloud point category can be corrected by combining auxiliary information of the point cloud points. Therefore, the implementation method for determining the importance measurement value of the target point cloud point according to the point category and correspondence is as follows: determine the initial importance measurement value of the target point cloud point according to the point category and correspondence; determine the auxiliary information of the target point cloud point participating in the importance measurement according to the application scenario requirements; correct the initial importance measurement value of the target point cloud point according to the auxiliary information to obtain the importance measurement value of the target point cloud point.

[0048] Specifically, the auxiliary information used in importance measurement is dynamically determined based on the application scenario requirements, enabling flexible responses to the needs of different application scenarios. For example, in autonomous driving, information such as the reflection intensity, distance, or color information of point cloud points can be identified as auxiliary information for importance measurement. Similarly, in urban planning, information such as the height, distance, texture, and color information of point cloud points can be identified as auxiliary information for importance measurement. And in virtual reality scenarios, information such as the texture and color information of point cloud points can be identified as auxiliary information for importance measurement.

[0049] In practical applications, when revising the initial importance metric of a target point cloud point based on auxiliary information to obtain its overall importance metric, the auxiliary information can be quantified to obtain its auxiliary importance metric. For example, professionals can quantify the auxiliary information to obtain the auxiliary importance metric. Another example is determining the auxiliary importance metric by querying pre-established correspondences between multiple different auxiliary information sets and multiple importance metrics based on the auxiliary information of the target point cloud point. Yet another example is training a multi-dimensional scoring model, inputting the auxiliary information into the model, and obtaining the auxiliary importance metric of the target point cloud point.

[0050] Optionally, to improve the accuracy and reliability of the importance measurement of point cloud points, the initial importance measurement value of the target point cloud point is corrected based on auxiliary information. The implementation method for obtaining the importance measurement value of the target point cloud point is as follows: determine the auxiliary importance measurement value of the target point cloud point based on the auxiliary information; perform a weighted summation of the initial importance measurement value and the auxiliary importance measurement value of the target point cloud point to obtain the importance measurement value of the target point cloud point, wherein the weight corresponding to the initial importance measurement value of the target point cloud point is greater than the weight corresponding to the auxiliary importance measurement value of the target point cloud point.

[0051] It is worth noting that the initial importance metric of the target point cloud is weighted greater than the auxiliary importance metric of the target point cloud, thereby ensuring that the point cloud category is the primary basis for evaluating the importance of the point cloud, while the auxiliary information of the point cloud is secondary.

[0052] Optionally, to improve the accuracy and reliability of the importance measurement of point cloud points, the initial importance measurement value and auxiliary importance measurement value of the target point cloud point are weighted and summed to obtain the importance measurement value of the target point cloud point. The implementation method is as follows: multiple point cloud points are classified into multiple levels according to the initial importance measurement values ​​of multiple point cloud points to obtain multiple point cloud points of different levels; the range of importance measurement values ​​of point cloud points of the same level is determined according to the initial importance measurement values ​​of multiple point cloud points of the same level; with the goal of controlling the importance measurement value of the target point cloud point to fall within the importance measurement value range of the level to which the target point cloud point belongs, the initial importance measurement value and auxiliary importance measurement value of the target point cloud point are weighted and summed to obtain the importance measurement value of the target point cloud point.

[0053] Specifically, multiple point cloud points in the point cloud data are classified into different levels, and the importance metric value range for each level is determined to facilitate reasonable adjustment of the importance metric values ​​later. In practical applications, the importance metric value range for point cloud points of the same level can be defined by the minimum and maximum importance metric values ​​of the point cloud points of the same level.

[0054] When revising the importance metric value of a target point cloud, it is necessary to ensure that the revised importance metric value falls within the range of importance metric values ​​for the target point cloud's corresponding level, thus preventing any change in the target point cloud's level. Specifically, this is achieved by adjusting the weights of the initial and auxiliary importance metrics of the target point cloud, thereby controlling the revised importance metric value to remain within the range of importance metric values ​​for the target point cloud's corresponding level.

[0055] It is worth noting that the process of correcting the importance metric of point cloud points not only considers the auxiliary information of individual point cloud points, but also the overall distribution of point cloud data. Through reasonable classification and weight setting, it can more accurately reflect the relative importance of each point cloud point in the point cloud data.

[0056] In some optional embodiments, the implementation method for determining the importance metric value corresponding to each of the multiple point cloud points according to the application scenario requirements corresponding to the point cloud data is as follows: if the application scenario requirements indicate that the importance metric value of the point cloud point is randomly generated, then multiple random numbers are randomly generated, and the multiple random numbers are randomly assigned to the multiple point cloud points as multiple importance metric values.

[0057] In practical applications, the requirement of the application scenario to randomly generate the importance metrics of point cloud points can be flexibly determined without restriction. For example, some application scenarios have high requirements for data processing speed, focusing primarily on reducing the amount of point cloud data and not on the importance metrics of the point cloud points. In this case, the application scenario requirement can be determined to instruct the random generation of the importance metrics of the point cloud points. Another example is when the application scenario explicitly requires the random generation of the importance metrics of the point cloud points. Yet another example is when the application scenario belongs to a predefined scenario that requires the random generation of the importance metrics of the point cloud points.

[0058] It is worth noting that randomly setting multiple importance metrics for point cloud points can ensure that every point cloud point in the point cloud data has a chance to be selected, increasing the diversity of point cloud points in the thinned point cloud data.

[0059] Alternatively, the random number generation can be achieved by generating multiple random numbers that conform to a uniform distribution. It is understood that since the random numbers are generated from a uniform distribution, each point cloud point has an equal chance of obtaining a high importance metric value, which helps ensure the diversity and representativeness of point cloud points in the thinned point cloud data.

[0060] 203. Determine the numerical range of importance metrics corresponding to at least one level according to the application scenario requirements; based on the numerical range of importance metrics corresponding to each of the multiple point cloud points and the importance metrics corresponding to at least one level, perform at least one level of thinning on the point cloud data to obtain point cloud data thinned at at least one level.

[0061] The importance metric values ​​of point cloud points differ at different levels of the thinned point cloud data. Each level of the thinned point cloud data includes point cloud points whose importance metric values ​​fall within the numerical range of the corresponding importance metric values ​​for that level.

[0062] In practical applications, point cloud data can be thinned at one or more levels. The range of importance metrics corresponding to the point cloud data after each level of thinning can be flexibly determined according to the application scenario requirements, without any restrictions.

[0063] In practical applications, for each point in a multi-point cloud dataset, the decision to assign the point to a specific level is made based on whether its corresponding importance metric value falls within the range of the importance metric values ​​for each level. This allows for at least one level of thinning of the point cloud data, resulting in thinned point cloud data at at least one level. Thinning involves selecting a subset of point cloud data points to form the thinned point cloud dataset. For example, if the importance metric values ​​of multiple point cloud points fall within the range [0,10], points with importance metric values ​​in [0,1] are grouped into one level of point cloud data, points with importance metric values ​​in (0,2) are grouped into another level, and points with importance metric values ​​in (2,10) are grouped into yet another level.

[0064] It is worth noting that point cloud data undergoes at least one level of thinning based on the importance metric values ​​of point cloud points and the numerical range of the importance metric values ​​corresponding to at least one level. The importance metric values ​​of point cloud points differ after thinning at different levels, achieving intelligent hierarchical processing of the point cloud data. Furthermore, during the thinning process, the importance metric values ​​of point cloud points at each level can be smoothly and continuously controlled, enabling hierarchical processing of point cloud data based on CLOD (Continuous Level of Detail). Understandably, by determining the importance of each point cloud point, it becomes possible to control the smooth and continuous changes in the importance of point cloud points between adjacent levels during multi-level thinning.

[0065] In practical applications, different scenarios have different requirements. For example, some scenarios have high requirements for data processing speed, thus necessitating a reduction in the amount of point cloud data. Other scenarios require high-precision point cloud data, requiring a sufficient amount of data. Still others require focusing on specific categories of point cloud data. Hierarchical processing of point cloud data can meet the needs of different application scenarios. Thinning the point cloud data reduces its volume, optimizes transmission efficiency, lowers storage costs, and enables real-time analysis and processing of point cloud data.

[0066] The technical solution provided in this application determines the importance metric values ​​corresponding to multiple point cloud points based on the application scenario requirements of the point cloud data; it determines the numerical range of the importance metric values ​​corresponding to at least one level based on the application scenario requirements; and it performs at least one level of thinning processing on the point cloud data based on the importance metric values ​​corresponding to the multiple point cloud points and the numerical range of the importance metric values ​​corresponding to at least one level, resulting in point cloud data thinned at at least one level. The importance metric values ​​of the point cloud points in the thinned point cloud data at different levels are different, and the thinned point cloud data at each level includes point cloud points whose importance metric values ​​fall within the numerical range of the importance metric values ​​corresponding to that level. This achieves the goal of classifying the importance of each point cloud point in the point cloud data according to the application scenario requirements, and performing one or more levels of thinning processing on the point cloud data according to the application scenario requirements. This realizes intelligent hierarchical processing of point cloud data, and the thinning processing greatly reduces the data volume of the point cloud data, making the point cloud data better adaptable to the application scenario and improving the efficiency and application effect of the point cloud data in the application scenario.

[0067] Figure 3 A flowchart illustrating a method for detecting changes in point cloud data provided in an embodiment of this application. See also... Figure 3 The method may include the following steps:

[0068] 301. Obtain the first point cloud data at the first time point and the second point cloud data at the second time point, where the first time point is earlier than the second time point.

[0069] In practical applications, change detection of point cloud data can identify differences in point cloud data within the same spatial range at different points in time. Based on the change detection results, the changed areas (also called changed regions) in a specific scene can be located. For ease of understanding and distinction, one of the two different time points is called the first time point, and the other is called the second time point. The first time point is earlier than the second time point. Point cloud data collected at the first time point is called the first point cloud data, and point cloud data collected at the second time point is called the second point cloud data.

[0070] 302. Perform at least one level of thinning processing on the first point cloud data and the second point cloud data respectively to obtain at least one level of third point cloud data corresponding to the first point cloud data and at least one level of fourth point cloud data corresponding to the second point cloud data.

[0071] In practical applications, the point cloud data thinning method described in the aforementioned embodiments can be used to thin the first point cloud data, obtaining at least one level of third point cloud data corresponding to the first point cloud data. It can be understood that the third point cloud data is the thinned first point cloud data.

[0072] In practical applications, the point cloud data thinning method described in the aforementioned embodiments can also be used to thin the second point cloud data, obtaining at least one fourth point cloud data at a corresponding level. It can be understood that the fourth point cloud data is the thinned second point cloud data.

[0073] 303. For the target third point cloud data and the target fourth point cloud data at the target level, divide the target third point cloud data into multiple first voxel grids and divide the target fourth point cloud data into multiple second voxel grids.

[0074] Among them, the target level is any one of at least one level, the target third point cloud data is any one of the third point cloud data under at least one level, and the target fourth point cloud data is any one of the fourth point cloud data under at least one level.

[0075] In this embodiment, voxelization is performed on the point cloud data, which divides the 3D point cloud data into multiple voxel grids. Voxelization transforms irregular point cloud data into a structured voxel representation, facilitating subsequent processing and analysis. For ease of understanding and differentiation, the voxel grid obtained by dividing the target third point cloud data is referred to as the first voxel grid, and the voxel grid obtained by dividing the target fourth point cloud data is referred to as the second voxel grid.

[0076] 304. For a target first voxel mesh and a target second voxel mesh representing the same spatial range, determine whether the target first voxel mesh has changed relative to the target second voxel mesh based on the existence of point cloud points in the target first voxel mesh and the existence of point cloud points in the target second voxel mesh.

[0077] In practical applications, the size and number of the first voxel mesh are the same as those of the second voxel mesh. The target first voxel mesh is any one of the multiple first voxel meshes, and the target second voxel mesh is any one of the multiple second voxel meshes. For target first voxel meshes and target second voxel meshes representing the same spatial range, the existence of point cloud points in the target first voxel mesh and the target second voxel mesh is used to determine whether the target first voxel mesh has changed relative to the target second voxel mesh.

[0078] It is understandable that if the target first voxel mesh changes relative to the target second voxel mesh, then the spatial range corresponding to the target first voxel mesh is the changed region.

[0079] For example, the method for determining whether the target first voxel grid has changed relative to the target second voxel grid based on the existence of point cloud points in the target first voxel grid and the existence of point cloud points in the target second voxel grid is as follows: if point cloud points exist in the target first voxel grid and point cloud points do not exist in the target second voxel grid, it is determined that the target first voxel grid has changed relative to the target second voxel grid; or, if point cloud points do not exist in the target first voxel grid and point cloud points exist in the target second voxel grid, it is determined that the target first voxel grid has changed relative to the target second voxel grid.

[0080] It is understandable that if both the first and second target voxel grids contain point cloud points, it is determined that the first target voxel grid has not changed relative to the second target voxel grid; if neither the first nor the second target voxel grid contains point cloud points, it is determined that the first target voxel grid has not changed relative to the second target voxel grid.

[0081] The point cloud data change detection method provided in this application performs at least one level of thinning processing on point cloud data at different time points, and performs voxelization processing on the thinned point cloud data at different time points under the same level. Change detection is performed based on the existence of point cloud points in the voxel grid representing the same spatial range at different time points. This method efficiently detects changes in point cloud data at different levels, accurately locates the changed area, and does not require processing the entire point cloud data, thus speeding up the change detection process and reducing the consumption of computing resources.

[0082] In some optional embodiments, after determining that the target first voxel grid has changed relative to the target second voxel grid, the above method may further include: marking the point cloud points in the target first voxel grid with change labels; and performing visualization processing on the target third point cloud data with the goal of making the display color of the point cloud points marked with change labels represent the change of the corresponding point cloud points, to obtain visualized target third point cloud data.

[0083] Specifically, change labels indicate that the corresponding point cloud points have changed. In practical applications, the display color representing changes in point cloud points can be flexibly set. By assigning specific display colors to point cloud points that have changed, users can intuitively identify which areas have changed, improving the visualization effect of change detection and making the changed areas immediately apparent.

[0084] In some optional embodiments, in order to ensure the effectiveness and reliability of change detection results, it is also possible to: determine the change detection performance at the target level based on the point cloud points in the target third point cloud data that are tagged with changes and the point cloud points in the target third point cloud data that have actually changed.

[0085] In practical applications, one or more evaluation metrics can be used to assess the performance of change detection. These metrics include, but are not limited to, True Positive Rate (TPR), False Positive Rate (FPR), Precision, Recall, and F1 Score. There are no restrictions on the specific metrics used. The F1 score is the harmonic mean of precision and recall, balancing the two. A high F1 score indicates good performance in both precision and recall.

[0086] To facilitate understanding, the following example of an urban scene will be used to illustrate the process of detecting changes in point cloud data.

[0087] First, point cloud data of the same city at two different points in time are obtained. The point cloud data from the earlier time is called the original point cloud, and the point cloud data from the later time is called the new point cloud.

[0088] Secondly, based on the category information of the point cloud data, the original point cloud and the new point cloud are thinned respectively.

[0089] Figure 4 The point cloud data shown is the full point cloud data. Figure 5 The point cloud data shown is from Figure 4 The local point cloud data shown is obtained by thinning the full point cloud data. Figure 4 and Figure 5 In this context, different colors of point cloud points correspond to different category information.

[0090] In practical applications, changes in buildings are of greater concern in urban scenarios. Therefore, when assigning importance metrics to point cloud points, building-related point cloud points can be assigned lower importance metrics to ensure they are prioritized during the thinning process. Other point cloud types, considered less important, can be assigned higher importance metrics. The lower the importance metric of a point cloud point, the higher its importance.

[0091] Next, the original point cloud and the new point cloud after thinning are voxelized respectively, and the voxel grids representing the same spatial range in the original point cloud and the new point cloud after thinning are compared to detect changes.

[0092] Finally, the change detection results are visualized.

[0093] Figure 6 The yellow portion represents the point cloud points that actually changed in the original point cloud. Figure 7 The red part in the image represents the point cloud points detected from the origin cloud and labeled with changes. Visually, most of the building changes have been accurately identified.

[0094] The F1 score is calculated based on the actual points in the original point cloud that have changed and the points in the original point cloud that have been detected and labeled with changes, as follows:

[0095] 1. Change detection was performed on the entire original point cloud and the new point cloud, with an F1 score of 0.816. The entire original point cloud includes data from more than 7 million point cloud points.

[0096] 2. Change detection was performed on the original point cloud and the new point cloud with importance metric values ​​falling within the range of (0, 0.5). The F1 score was 0.783. The original point cloud with importance metric values ​​falling within the range of (0, 0.5) includes data of approximately 120,000 point cloud points.

[0097] 3. Change detection was performed on the original point cloud and the new point cloud whose importance metric values ​​fall within the range of (0, 1). The F1 score was 0.798. The original point cloud with importance metric values ​​falling within the range of (0, 1) includes data of approximately 700,000 point cloud points.

[0098] 4. Change detection was performed on the original point cloud and the new point cloud whose importance metric values ​​fall within the range of (0, 2). The F1 score was 0.804. The original point cloud with importance metric values ​​falling within the range of (0, 2) includes data of about 1.8 million point cloud points.

[0099] 5. Change detection was performed on the original point cloud and the new point cloud whose importance metric values ​​fall within the range of (0, 3). The F1 score was 0.811. The original point cloud with importance metric values ​​falling within the range of (0, 3) includes data of approximately 2.4 million point cloud points.

[0100] The data above shows that the accuracy and reliability of change detection are both good for point cloud data of any level.

[0101] Figure 8 This is a schematic diagram of a point cloud data thinning processing device provided in an embodiment of this application. See also... Figure 8 The device may include:

[0102] The acquisition module 81 is used to acquire point cloud data, which includes data from multiple point cloud points.

[0103] The metric module 82 is used to determine the importance metric values ​​of multiple point cloud points according to the application scenario requirements corresponding to the point cloud data.

[0104] The thinning module 83 is used to perform at least one level of thinning on the point cloud data based on the importance metric values ​​corresponding to each of the multiple point cloud points, so as to obtain point cloud data thinned at at least one level. The importance metric values ​​of the point cloud points in the point cloud data thinned at different levels are different.

[0105] Optionally, the measurement module 82 is specifically used to: if the application scenario requires that the point category of the point cloud point is related to the importance measurement value of the point cloud point, then determine the correspondence between multiple different point categories and multiple different importance measurement values ​​according to the application scenario requirements; for any target point cloud point among multiple point cloud points, determine the importance measurement value of the target point cloud point according to the point category and correspondence of the target point cloud point.

[0106] Optionally, when determining the importance metric value of the target point cloud points, the metric module 82 is specifically used for:

[0107] If the current point category of the point cloud does not meet the application scenario requirements, a machine learning model is invoked to classify the point cloud data, resulting in updated point categories for multiple point cloud points. Based on the updated point categories and corresponding relationships of the target point cloud, the importance metric of the target point cloud is determined.

[0108] Optionally, when determining the importance metric value of the target point cloud, the metric module 82 is specifically used to: determine the initial importance metric value of the target point cloud based on the point category and correspondence of the target point cloud; determine the auxiliary information of the target point cloud participating in the importance metric based on the application scenario requirements; and correct the initial importance metric value of the target point cloud based on the auxiliary information to obtain the importance metric value of the target point cloud.

[0109] Optionally, when the measurement module 82 corrects the initial importance measurement value of the target point cloud point based on the auxiliary information, it is specifically used to: determine the auxiliary importance measurement value of the target point cloud point based on the auxiliary information; and perform a weighted summation of the initial importance measurement value and the auxiliary importance measurement value of the target point cloud point to obtain the importance measurement value of the target point cloud point, wherein the weight corresponding to the initial importance measurement value of the target point cloud point is greater than the weight corresponding to the auxiliary importance measurement value of the target point cloud point.

[0110] Optionally, when the measurement module 82 performs weighted summation, it is specifically used to: classify multiple point cloud points into different levels based on their initial importance measurement values; determine the range of importance measurement values ​​for point cloud points of the same level based on their initial importance measurement values; and, with the goal of controlling the importance measurement value of the target point cloud point to fall within the range of importance measurement values ​​of the target point cloud point's level, perform weighted summation of the initial importance measurement value and the auxiliary importance measurement value of the target point cloud point to obtain the importance measurement value of the target point cloud point.

[0111] Optionally, the measurement module 82 is specifically used to: if the application scenario requires that the importance measurement value of the point cloud points be generated randomly, then generate multiple random numbers and assign the multiple random numbers as multiple importance measurement values ​​to the multiple point cloud points.

[0112] Optionally, when the measurement module 82 randomly generates multiple random numbers, it is specifically used to: randomly generate multiple random numbers that conform to a uniform distribution.

[0113] Figure 8 The device shown can perform Figure 2 The implementation principle and technical effects of the method shown in the embodiments will not be elaborated further. Regarding the above embodiments... Figure 8 The specific ways in which each module and unit of the device performs operations have been described in detail in the embodiments of the method, and will not be elaborated here.

[0114] Figure 9 This is a schematic diagram of a point cloud data change detection device provided in an embodiment of this application. See also... Figure 9 The device may include:

[0115] The acquisition module 91 is used to acquire the first point cloud data at a first time point and the second point cloud data at a second time point, wherein the first time point is earlier than the second time point;

[0116] The thinning module 92 is used to perform at least one level of thinning processing on the first point cloud data and the second point cloud data respectively, to obtain at least one level of third point cloud data corresponding to the first point cloud data and at least one level of fourth point cloud data corresponding to the second point cloud data.

[0117] The voxelization module 93 is used to divide the target third point cloud data into multiple first voxel grids and the target fourth point cloud data into multiple second voxel grids for the target third point cloud data and the target fourth point cloud data at the target level; the target level is any one of at least one level, the target third point cloud data is any one of the third point cloud data at at least one level, and the target fourth point cloud data is any one of the fourth point cloud data at at least one level;

[0118] The change detection module 94 is used to determine whether the target first voxel grid has changed relative to the target second voxel grid, based on the existence of point cloud points in the target first voxel grid and the target second voxel grid, which represent the same spatial range.

[0119] Optionally, the change detection module 94 is specifically used to: determine that the target first voxel grid has changed relative to the target second voxel grid if there are point cloud points in the target first voxel grid and there are no point cloud points in the target second voxel grid; or, determine that the target first voxel grid has changed relative to the target second voxel grid if there are no point cloud points in the target first voxel grid and there are point cloud points in the target second voxel grid.

[0120] Optionally, the above apparatus further includes: a visualization module, used to mark change labels on point cloud points in the target first voxel grid; and to perform visualization processing on the target third point cloud data with the goal of making the display color of the marked point cloud points represent changes in the corresponding point cloud points, so as to obtain visualized target third point cloud data.

[0121] Optionally, the above device further includes: a performance evaluation module, used to determine the change detection performance at the target level based on the point cloud points in the target third point cloud data that are tagged with changes and the point cloud points in the target third point cloud data that have actually changed.

[0122] Figure 9 The device shown can perform Figure 3 The implementation principle and technical effects of the method shown in the embodiments will not be elaborated further. Regarding the above embodiments... Figure 9 The specific ways in which each module and unit of the device performs operations have been described in detail in the embodiments of the method, and will not be elaborated here.

[0123] It should be noted that the execution subject of each step of the method provided in the above embodiments can be the same device, or the method can be executed by different devices. For example, the execution subject of steps 201 to 203 can be device A; or the execution subject of steps 201 and 202 can be device A, and the execution subject of step 203 can be device B; and so on.

[0124] Furthermore, in some of the processes described in the above embodiments and accompanying drawings, multiple operations appear in a specific order. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or they may be executed in parallel. The operation numbers, such as 201, 202, etc., are merely used to distinguish different operations and do not represent any execution order. Additionally, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first" and "second" in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.

[0125] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation portals are provided for users to choose to authorize or refuse.

[0126] Figure 10 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 10 As shown, the electronic device includes: a memory 11 and a processor 12;

[0127] Memory 11 is used to store computer programs and can be configured to store various other data to support operation on the computing platform. Examples of this data include instructions for any application or method operating on the computing platform, contact data, phone book data, messages, pictures, videos, etc.

[0128] The memory 11 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random-access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0129] The processor 12, coupled to the memory 11, is used to execute a computer program in the memory 11 for: performing steps in a point cloud data thinning process or change detection method.

[0130] Optional, such as Figure 10 As shown, the electronic device also includes other components such as a communication component 13, a display 14, a power supply component 15, and an audio component 16. Figure 10 The diagram only shows some components and does not mean that the electronic device includes only these components. Figure 10 The components shown. Additionally... Figure 10 The components within the dashed box are optional, not mandatory, and their specific requirements depend on the product form of the electronic device. The electronic device in this embodiment can be a desktop computer, laptop computer, smartphone, or IoT (Internet of Things) device, or a server-side device such as a conventional server, cloud server, or server array. If the electronic device in this embodiment is a desktop computer, laptop computer, or smartphone, it may include... Figure 10 The components within the dashed box; if the electronic device in this embodiment is implemented as a conventional server, cloud server, or server array, etc., it may be omitted. Figure 10 The component within the dashed box.

[0131] For a detailed description of the implementation process of each action by the processor, please refer to the relevant descriptions in the foregoing method embodiments or device embodiments, which will not be repeated here.

[0132] Accordingly, embodiments of this application also provide a computer-readable storage medium storing a computer program, which, when executed, can implement the steps that can be performed by an electronic device in the above method embodiments.

[0133] Accordingly, this application also provides a computer program product, including a computer program / instructions, which, when executed by a processor, enables the processor to perform the steps that can be executed by an electronic device in the above method embodiments.

[0134] The aforementioned communication components are configured to facilitate wired or wireless communication between the device containing the communication components and other devices. The device containing the communication components can access wireless networks based on communication standards, such as WiFi (Wireless Fidelity), 2G (2nd Generation), 3G (3rd Generation), 4G (4th Generation) / LTE (long Term Evolution), 5G (5th Generation), or combinations thereof. In one exemplary embodiment, the communication components receive broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, the communication components also include a Near Field Communication (NFC) module to facilitate short-range communication. For example, the NFC module may be based on Radio Frequency Identification (RFID), Infrared Data Association (IrDA), Ultra Wide Band (UWB), Bluetooth, and other technologies.

[0135] The aforementioned display includes a screen, which may include a Liquid Crystal Display (LCD) and a Touch Panel (TP). If the screen includes a Touch Panel, the screen can be implemented as a touchscreen to receive input signals from the user. The Touch Panel includes one or more touch sensors to sense touches, swipes, and gestures on the Touch Panel. The touch sensors can sense not only the boundaries of touch or swipe actions but also the duration and pressure associated with the touch or swipe operation.

[0136] The aforementioned power supply components provide power to various components within the device in which they reside. These power supply components may include a power management system, one or more power sources, and other components associated with generating, managing, and distributing power to the device in which they reside.

[0137] The aforementioned audio component can be configured to output and / or input audio signals. For example, the audio component includes a microphone (MIC) configured to receive external audio signals when the device containing the audio component is in an operating mode, such as call mode, recording mode, or voice recognition mode. The received audio signals can be further stored in memory or transmitted via a communication component. In some embodiments, the audio component also includes a speaker for outputting audio signals.

[0138] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0139] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0140] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0141] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0142] In a typical configuration, a computing device includes one or more processors (Central Processing Unit, CPU), input / output interfaces, network interfaces, and memory.

[0143] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0144] Computer-readable media, including both permanent and non-permanent, removable and non-removable media, can store information using any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change RAM (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, CD-ROM, digital versatile disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device. As defined in this article, computer-readable media do not include transient media, such as modulated data signals and carrier waves.

[0145] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0146] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for thinning point cloud data, characterized in that, include: Acquire point cloud data, wherein the point cloud data includes data of multiple point cloud points; Based on the application scenario requirements corresponding to the point cloud data, determine the importance metric value corresponding to each of the multiple point cloud points; Determine the numerical range of importance metrics for at least one level based on the application scenario requirements; Based on the importance metric values ​​corresponding to each of the multiple point cloud points and the numerical range of the importance metric values ​​corresponding to at least one level, the point cloud data is subjected to at least one level of thinning processing to obtain point cloud data thinned at at least one level. The importance metric values ​​of the point cloud points in the point cloud data thinned at different levels are different. The point cloud data thinned at each level includes point cloud points whose importance metric values ​​fall within the numerical range of the importance metric values ​​corresponding to that level.

2. The method according to claim 1, characterized in that, Based on the application scenario requirements corresponding to the point cloud data, determine the importance metric value corresponding to each of the multiple point cloud points, including: If the application scenario requirement indicates that the point category of the point cloud points is related to the importance metric value of the point cloud points, then the correspondence between multiple different point categories and multiple different importance metrics is determined according to the application scenario requirement. For any target point cloud point among the plurality of point cloud points, the importance metric value of the target point cloud point is determined according to the point category of the target point cloud point and the correspondence.

3. The method according to claim 2, characterized in that, Based on the point categories of the target point cloud points and the corresponding relationships, the importance metric value of the target point cloud points is determined, including: If the current point category of the point cloud does not meet the requirements of the application scenario, a machine learning model is invoked to classify the point cloud data to obtain the updated point categories of the multiple point cloud points. The importance metric of the target point cloud is determined based on the updated point category of the target point cloud and the corresponding relationship.

4. The method according to claim 2, characterized in that, Based on the point categories of the target point cloud points and the corresponding relationships, the importance metric value of the target point cloud points is determined, including: Based on the point categories of the target point cloud points and the corresponding relationships, determine the initial importance metric value of the target point cloud points; Based on the requirements of the application scenario, determine the auxiliary information of the target point cloud points that participate in the importance measurement; The initial importance metric of the target point cloud is corrected based on the auxiliary information to obtain the importance metric of the target point cloud.

5. The method according to claim 4, characterized in that, The initial importance metric of the target point cloud points is corrected based on the auxiliary information to obtain the importance metric of the target point cloud points, including: The auxiliary importance metric value of the target point cloud points is determined based on the auxiliary information; The initial importance metric and the auxiliary importance metric of the target point cloud are weighted and summed to obtain the importance metric of the target point cloud, wherein the weight corresponding to the initial importance metric of the target point cloud is greater than the weight corresponding to the auxiliary importance metric of the target point cloud.

6. The method according to claim 5, characterized in that, The initial importance metric and the auxiliary importance metric of the target point cloud are weighted and summed to obtain the importance metric of the target point cloud, including: The multiple point cloud points are classified into different levels based on their initial importance metric values, resulting in multiple point cloud points of different levels. Based on the initial importance metric values ​​of multiple point cloud points of the same level, determine the range of importance metric values ​​for point cloud points of the same level. With the goal of controlling the importance metric value of the target point cloud point to fall within the importance metric value range of the level to which the target point cloud point belongs, the initial importance metric value and the auxiliary importance metric value of the target point cloud point are weighted and summed to obtain the importance metric value of the target point cloud point.

7. The method according to claim 1, characterized in that, Based on the application scenario requirements corresponding to the point cloud data, determine the importance metric value corresponding to each of the multiple point cloud points, including: If the application scenario requires that the importance metric value of the point cloud points be generated randomly, then multiple random numbers are generated randomly, and these multiple random numbers are randomly assigned as multiple importance metric values ​​to the multiple point cloud points.

8. The method according to claim 7, characterized in that, Generate multiple random numbers, including: Generate multiple random numbers that conform to a uniform distribution.

9. A method for detecting changes in point cloud data, characterized in that, include: Acquire first-point cloud data at a first time point and second-point cloud data at a second time point, wherein the first time point is earlier than the second time point; According to any one of claims 1-8, the thinning process is performed on the first point cloud data and the second point cloud data at at least one level to obtain the third point cloud data at at least one level corresponding to the first point cloud data and the fourth point cloud data at at least one level corresponding to the second point cloud data. For the target third point cloud data and the target fourth point cloud data at the target level, the target third point cloud data is divided into multiple first voxel grids and the target fourth point cloud data is divided into multiple second voxel grids; the target level is any one of at least one level, the target third point cloud data is any one of the third point cloud data at at least one level, and the target fourth point cloud data is any one of the fourth point cloud data at at least one level. For a target first voxel grid and a target second voxel grid representing the same spatial range, it is determined whether the target first voxel grid has changed relative to the target second voxel grid based on the existence of point cloud points in the target first voxel grid and the existence of point cloud points in the target second voxel grid.

10. The method according to claim 9, characterized in that, Based on the existence of point cloud points in the first target voxel mesh and the existence of point cloud points in the second target voxel mesh, determine whether the first target voxel mesh has changed relative to the second target voxel mesh, including: If point cloud points exist in the first target voxel mesh, and point cloud points do not exist in the second target voxel mesh, it is determined that the first target voxel mesh has changed relative to the second target voxel mesh; or... If there are no point cloud points in the first target voxel grid, but there are point cloud points in the second target voxel grid, it is determined that the first target voxel grid has changed relative to the second target voxel grid.

11. The method according to claim 9, characterized in that, After determining that the target first voxel mesh has changed relative to the target second voxel mesh, the method further includes: Add change labels to the point cloud points in the first voxel mesh of the target; With the goal of making the display color of the point cloud points marked with change represent the change of the corresponding point cloud points, the target third point cloud data is visualized to obtain visualized target third point cloud data.

12. The method according to claim 11, characterized in that, Also includes: Based on the point cloud points in the target third point cloud data that are tagged with changes, and the point cloud points in the target third point cloud data that have actually changed, the change detection performance at the target level is determined.

13. An electronic device, characterized in that, include: Memory and processor; The memory is used to store computer programs; The processor is coupled to the memory for executing the computer program to perform the steps of the method according to any one of claims 1-12.

14. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it causes the processor to perform the steps of the method according to any one of claims 1-12.

15. A computer program product, characterized in that, Includes a computer program / instruction that, when executed by a processor, causes the processor to perform the steps of the method according to any one of claims 1-12.

Images