Laser dynamic point determination method and device and computer equipment

Abstract

The invention relates to the technical field of positioning, in particular to a laser dynamic point determination method and device and computer equipment. The method comprises the following steps: acquiring three-dimensional point cloud data of each laser point in a target area; determining a transformation matrix according to geometric information of the three-dimensional point cloud data in the three-dimensional space; wherein the transformation matrix comprises a rotation transformation parameter and a translation transformation parameter; performing alignment processing on the three-dimensional point cloud data in the translation direction and the rotation direction of the three-dimensional space by adopting the transformation matrix; and determining laser dynamic points in the laser points according to the aligned three-dimensional point cloud data.

Description

Technical Field

[0001] The present application relates to the field of positioning technology, and in particular to a laser dynamic point determination method, device and computer equipment. Background Art

[0002] Automated Guided Vehicle (AGV), as an advanced intelligent transport equipment, is widely used in factories, warehouses and other places to efficiently perform cargo handling tasks. AGV relies on advanced laser technology to achieve precise mapping and positioning, thereby ensuring stable and accurate execution of tasks in complex and changing environments.

[0003] However, pure laser mapping technology has accuracy limitations. Mapping and registration are based on the assumption that the environment is static, and the emergence of laser dynamic points will undoubtedly interfere with the accuracy of this process.

[0004] When the proportion of dynamic points in a frame of data is too high, the trajectory accuracy will be seriously affected, so it is urgently needed to be improved. Summary of the invention

[0005] Based on this, it is necessary to provide a laser dynamic point determination method, device and computer equipment that can improve the accuracy of mapping in response to the above technical problems.

[0006] In a first aspect, the present application provides a laser dynamic point determination method, the method comprising:

[0007] Obtain three-dimensional point cloud data of each laser point in the target area;

[0008] Determine a transformation matrix according to geometric information of each 3D point cloud data in the 3D space; wherein the transformation matrix includes rotation transformation parameters and translation transformation parameters;

[0009] Using the transformation matrix, the 3D point cloud data is aligned in the translation direction and rotation direction of the 3D space;

[0010] According to the three-dimensional point cloud data after the alignment process, the laser dynamic point in each laser point is determined.

[0011] In one embodiment, determining a transformation matrix according to geometric information of each 3D point cloud data in a 3D space includes:

[0012] Input each 3D point cloud data into the rotation network to obtain the rotation transformation parameters of the rotation direction output by the rotation network;

[0013] Analyze the geometric information of the three-dimensional point cloud data of each laser point in the translation direction to obtain the translation transformation parameters corresponding to the translation direction;

[0014] The rotation transformation parameters are combined with the translation transformation parameters corresponding to the translation direction to obtain a transformation matrix.

[0015] In one embodiment, the rotation transformation parameter is combined with the translation transformation parameter corresponding to the translation direction to obtain a transformation matrix, including:

[0016] The rotation transformation parameter is used as the first row and first column element in the two-dimensional square matrix, and the translation transformation parameter corresponding to the translation direction is used as the first row and second column element in the two-dimensional square matrix;

[0017] Use 0 as the element of the second row and first column in the two-dimensional square matrix, and use 1 as the element of the second row and second column in the two-dimensional square matrix to obtain the transformation matrix.

[0018] In one embodiment, determining the laser dynamic point in each laser point according to each three-dimensional point cloud data after alignment processing includes:

[0019] Obtain laser point description data of each laser point;

[0020] Determine the laser point features of each laser point according to the laser point description data of each laser point and the three-dimensional point cloud data after alignment processing;

[0021] According to the laser point characteristics of each laser point, the laser dynamic point in each laser point is determined.

[0022] In one embodiment,

[0023] According to the laser point characteristics of each laser point, the laser dynamic point in each laser point is determined, including:

[0024] Aggregate the laser point features of each laser point to obtain the global feature;

[0025] According to the global features and the laser point features of each laser point, the laser dynamic point in each laser point is determined.

[0026] In one embodiment, obtaining three-dimensional point cloud data of each laser point in the target area includes:

[0027] For each laser point in the target area, the three-dimensional point cloud data of the laser point is determined based on the acquisition configuration information of the laser acquisition device and the length value reflected back by the laser point; wherein the acquisition configuration information of the laser acquisition device includes horizontal resolution, vertical resolution, horizontal scanning starting angle and vertical scanning starting angle.

[0028] In a second aspect, the present application also provides a laser dynamic point determination device, comprising:

[0029] An acquisition module is used to acquire three-dimensional point cloud data of each laser point in the target area;

[0030] An offset parameter transformation module is used to determine a transformation matrix according to geometric information of each 3D point cloud data in a 3D space; wherein the transformation matrix includes rotation transformation parameters and translation transformation parameters;

[0031] An alignment processing module is used to align the three-dimensional point cloud data in the translation direction and rotation direction of the three-dimensional space by using a transformation matrix;

[0032] The laser dynamic point recognition module is used to determine the laser dynamic point in each laser point according to each three-dimensional point cloud data after alignment processing.

[0033] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0034] Obtain three-dimensional point cloud data of each laser point in the target area;

[0035] Determine a transformation matrix according to geometric information of each 3D point cloud data in the 3D space; wherein the transformation matrix includes rotation transformation parameters and translation transformation parameters;

[0036] Using the transformation matrix, the 3D point cloud data is aligned in the translation direction and rotation direction of the 3D space;

[0037] According to the three-dimensional point cloud data after the alignment process, the laser dynamic point in each laser point is determined.

[0038] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the following steps are implemented:

[0039] Obtain three-dimensional point cloud data of each laser point in the target area;

[0040] Determine a transformation matrix according to geometric information of each 3D point cloud data in the 3D space; wherein the transformation matrix includes rotation transformation parameters and translation transformation parameters;

[0041] Using the transformation matrix, the 3D point cloud data is aligned in the translation direction and rotation direction of the 3D space;

[0042] According to the three-dimensional point cloud data after the alignment process, the laser dynamic point in each laser point is determined.

[0043] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0044] Obtain three-dimensional point cloud data of each laser point in the target area;

[0045] Determine a transformation matrix according to geometric information of each 3D point cloud data in the 3D space; wherein the transformation matrix includes rotation transformation parameters and translation transformation parameters;

[0046] Using the transformation matrix, the 3D point cloud data is aligned in the translation direction and rotation direction of the 3D space;

[0047] According to the three-dimensional point cloud data after the alignment process, the laser dynamic point in each laser point is determined.

[0048] The above-mentioned laser dynamic point determination method, device and computer equipment align each three-dimensional point cloud data in the translation direction and rotation direction through the transformation matrix determined according to the geometric information of the three-dimensional point cloud data in the three-dimensional space. This alignment process can eliminate the misalignment of the point cloud data caused by factors such as the position and posture changes of the laser scanner and the irregularity of the environment itself, thereby improving the accuracy of mapping. By identifying and processing laser dynamic points, the present application enables laser mapping and positioning technology to better adapt to dynamic environments. In a dynamic environment, traditional methods may not work properly due to the interference of dynamic points, while the present application can maintain a high mapping accuracy and positioning stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0050] Figure 1 A diagram showing an application environment of a laser dynamic point determination method in one embodiment;

[0051] Figure 2 A schematic diagram of a process of a laser dynamic point determination method in one embodiment;

[0052] Figure 3 is a schematic diagram of a laser point cloud image in one embodiment;

[0053] Figure 4 A schematic diagram of a flow chart of a step of determining a laser dynamic point in each laser point in an embodiment;

[0054] Figure 5 A schematic diagram of a flow chart of a step of determining a transformation matrix in one embodiment;

[0055] Figure 6 is a schematic diagram of PointNet in one embodiment;

[0056] Figure 7 A schematic diagram of the first T-Net in PointNet in one embodiment;

[0057] Figure 8 is a structural block diagram of a laser dynamic point determination device in one embodiment;

[0058] Fig. 9 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0060] In an exemplary embodiment, the laser dynamic point determination method provided in the embodiment of the present application can be applied to Figure 1 The computer device shown is a laser acquisition device, which can be a 16-line navigation laser radar. A single-line 360° scanning range has 2,000 laser points, a resolution of 0.18°, and a frame of laser data has a total of 32,000 points.

[0061] In an exemplary embodiment, Figure 2 As shown, a laser dynamic point determination method is provided, comprising:

[0062] S201, obtaining three-dimensional point cloud data of each laser point in the target area.

[0063] The target area can be a room, a forest, a construction site, or any area where you want to obtain 3D data. The determination of the target area is usually based on actual application requirements and research purposes.

[0064] It can be understood that 3D point cloud data refers to a set of 3D coordinate data of a large number of points in the target area obtained by some means (such as lidar scanning, photogrammetry, 3D scanning, etc.). Each point contains its X, Y, and Z coordinate information in 3D space, and sometimes may also contain other attribute information such as color and intensity.

[0065] For example, Figure 3 It is a 3D laser scanning point cloud map.

[0066] S202, determining a transformation matrix according to geometric information of each 3D point cloud data in the 3D space.

[0067] It can be understood that the geometric information of 3D point cloud data, specifically, includes the point position (x, y, z coordinates), normal vector, curvature, etc. This information is used to describe the shape, structure and topological relationship of the point cloud data in 3D space.

[0068] The transformation matrix includes rotation transformation parameters and translation transformation parameters.

[0069] Specifically, the transformation matrix usually includes a translation vector, a transformation matrix (or quaternion), and a scaling factor, etc. These parameters are used to describe the position, direction, and size difference of the point cloud data to be aligned relative to the reference point cloud data in three-dimensional space.

[0070] It can be understood that the translation vector describes the translation position difference of the point cloud to be aligned in the three-dimensional space, which can be obtained by calculating the average position difference between corresponding points.

[0071] The transformation matrix (or quaternion) describes the difference in the rotation direction of the point cloud to be aligned in three-dimensional space. It can be obtained by calculating the covariance matrix between corresponding points and performing singular value decomposition (SVD) on the covariance matrix. In addition, quaternion is also a commonly used way to represent rotation, which can avoid the universal joint deadlock problem and is more stable in calculation.

[0072] The scaling factor describes the size difference of the point clouds to be aligned in 3D space. It can be obtained by calculating the average distance ratio between corresponding points. However, in point cloud alignment, the scaling factor is usually not necessary because most alignment algorithms assume that the point cloud data are similar in size.

[0073] Optionally, PointNet can predict the affine transformation matrix through the built-in T-Net structure to align the input point cloud data. This method enables PointNet to directly process unordered point cloud data and extract spatially consistent feature representations. When determining the transformation matrix, PointNet makes full use of the geometric information of the point cloud data (such as 3D coordinates) and the learning ability of the deep learning model to achieve effective alignment and feature extraction of point cloud data.

[0074] S203, using a transformation matrix, aligning each 3D point cloud data in the translation direction and rotation direction of the 3D space.

[0075] It is understandable that point cloud data collected from different sources or at different times may be in different coordinate systems. Using the affine transformation matrix, these data can be transformed into a unified coordinate system to facilitate subsequent data processing and analysis. Due to factors such as measurement equipment and environmental conditions, point cloud data may have certain errors. Through alignment processing, these errors can be eliminated and the accuracy and reliability of point cloud data can be improved. Alignment processing is an essential step when multiple point cloud data sets need to be fused into a complete model. Only after alignment processing can the seamless connection between different data sets be ensured.

[0076] Optionally, before the alignment process, point cloud data collected from different sources or at different times may be in different positions and orientations. After the alignment process, these data are transformed into a unified coordinate system, and the position and orientation differences are eliminated.

[0077] S204, determining the laser dynamic point in each laser point according to each three-dimensional point cloud data after the alignment processing.

[0078] Optionally, in this embodiment, the three-dimensional point cloud data after the alignment processing can be input into the multi-layer perception machine layer of PointNet for feature extraction, and the laser dynamic point in each laser point can be determined according to the feature extraction result.

[0079] Traditional laser mapping technology is usually based on the assumption that the environment is static, and the appearance of dynamic points will interfere with the mapping accuracy. By identifying laser dynamic points and excluding them from the mapping process or specially processing them, the negative impact of dynamic points on mapping accuracy is reduced.

[0080] The above-mentioned laser dynamic point determination method aligns each three-dimensional point cloud data in the translation direction and rotation direction through a transformation matrix determined according to the geometric information of the three-dimensional point cloud data in three-dimensional space. This alignment process can eliminate the misalignment of point cloud data caused by factors such as the position and posture changes of the laser scanner and the irregularity of the environment itself, thereby improving the accuracy of mapping. By identifying and processing laser dynamic points, the present application enables laser mapping and positioning technology to better adapt to dynamic environments. In a dynamic environment, traditional methods may not work properly due to interference from dynamic points, while the present application can maintain a high level of mapping accuracy and positioning stability.

[0081] In an exemplary embodiment, the three-dimensional point cloud data of each laser point in the target area is obtained, including: for each laser point in the target area, the three-dimensional point cloud data of the laser point is determined according to the acquisition configuration information of the laser acquisition device and the length value reflected back by the laser point.

[0082] Among them, the acquisition configuration information of the laser acquisition device includes horizontal resolution, vertical resolution, horizontal scanning starting angle and vertical scanning starting angle.

[0083] Optionally, each laser point is represented by its three-dimensional coordinates (x, y, z). First, the laser data is converted to a coordinate system. The laser point cloud data is represented by a one-dimensional array. According to the starting angle and resolution of the scan, the following formula is used to convert the current array into a rectangular coordinate system.

[0084] Exemplarily, the calculation process of the three-dimensional point cloud data of laser point i is as follows:

[0085] (1)

[0086] (2)

[0087] (3)

[0088] The laser point cloud array stores the length value reflected from each laser point, laser[i] is the length value reflected from laser point i; h_resolution represents the horizontal resolution, v_resolution represents the vertical resolution, h_start_ang represents the horizontal scan start angle, and v_start_ang represents the vertical scan start angle.

[0089] In an exemplary embodiment, Figure 4 As shown, according to the three-dimensional point cloud data after alignment, the laser dynamic point in each laser point is determined, including:

[0090] S401, obtaining laser point description data of each laser point.

[0091] It is understandable that in the PointNet framework, when extracting features from each point through a multilayer perceptron (MLP), in addition to the basic feature of the three-dimensional coordinates (x, y, z) of each point, other features may include the color, reflectivity, normal, curvature, etc. of the point. These features are all attributes related to point cloud data, and they can provide more information for subsequent tasks such as segmentation and classification, so they are used as laser point description data. Specifically, color can reflect the material and surface characteristics of an object; reflectivity is related to the surface gloss and lighting conditions of the object; normal and curvature can describe the local geometry of the point cloud data. In practical applications, appropriate features can be selected for extraction and analysis based on specific needs and the characteristics of point cloud data.

[0092] S402, determining the laser point features of each laser point according to the laser point description data of each laser point and the three-dimensional point cloud data after the alignment processing.

[0093] Specifically, the point cloud data after coordinate transformation is input into the PointNet framework. PointNet is a point cloud processing framework based on deep learning. It can directly accept unordered point cloud data as input and extract features for each point through MLP. Finally, the category label of each point is output. In this way, different objects or regions in the point cloud data can be segmented.

[0094] S403, determining the laser dynamic point in each laser point according to the laser point characteristics of each laser point.

[0095] Optionally, according to the laser point features of each laser point, the laser dynamic point in each laser point is determined, including: aggregating the laser point features of each laser point to obtain a global feature; and determining the laser dynamic point in each laser point according to the global feature and the laser point features of each laser point.

[0096] Optionally, in PointNet, each point is feature extracted through a weight-sharing MLP to obtain the local features of the point. Then, PointNet uses a maximum pooling operation to aggregate the features of all points to obtain a global feature vector. This global feature vector contains key information of the entire point cloud and can be used for subsequent segmentation tasks. In the segmentation task, PointNet copies and broadcasts the global feature vector to each point and splices it with the local features of each point. Then, these spliced ​​features are again feature extracted through MLP, and dynamic targets such as pedestrians and vehicles in the point cloud data are segmented according to the output results of PointNet. These dynamic targets require special attention in applications such as autonomous driving and robot navigation, because they may affect tasks such as path planning and obstacle avoidance.

[0097] In this embodiment, in some application scenarios (such as mapping, positioning, etc.), only static background information may be required, but not dynamic target information. The segmented point cloud data of the dynamic target is removed from the original point cloud data to obtain point cloud data containing only the static background. In this way, the environment map can be constructed more accurately, and functions such as precise positioning can be achieved.

[0098] In an exemplary embodiment, the transformation matrix includes rotation transformation parameters and translation transformation parameters; Figure 5 As shown, according to the geometric information of each 3D point cloud data in the 3D space, a transformation matrix is ​​determined; including:

[0099] S501, input each 3D point cloud data into the rotation network, and obtain the rotation transformation parameters of the rotation direction output by the rotation network.

[0100] It is understandable that since point cloud data is collected in three-dimensional space, each point contains its three-dimensional coordinates (x, y, z). However, due to the installation position and angle of the sensor and the influence of environmental factors (such as terrain, buildings, etc.), these coordinates may not fully meet the requirements of subsequent processing.

[0101] Therefore, it is necessary to perform coordinate transformation on the point cloud data. This usually includes operations such as translation, rotation, and scaling to align the point cloud data with the reference coordinate system or processing requirements. The Input Transform layer in the PointNet framework, including a T-Net, is used to predict an affine transformation matrix to align the input point cloud data.

[0102] It is understandable that the T-Net rotation network is an important component in PointNet, and the InputTransform in PointNet is transformed by T-Net. T-Net maps the input point cloud data or features to a high-dimensional space and learns a transformation matrix through structures such as MLP and fully connected layers.

[0103] Input Transform is used to learn the point cloud transformation matrix. The input is point cloud data, and the output is a 3x3 transformation matrix, which is used to correct the posture of the point cloud so that the model can better learn the global features of the point cloud data.

[0104] S502, analyzing the geometric information of the three-dimensional point cloud data of each laser point in the translation direction to obtain the translation transformation parameters corresponding to the translation direction.

[0105] It is understandable that the Input Transform in PointNet only extracts the rotation features without taking into account the translation features.

[0106] Optionally, the translation transformation parameter is a translation vector, the translation vector is t= .

[0107] Optionally, the rotation transformation parameter is a rotation transformation parameter in matrix form;

[0108] S503: Combine the rotation transformation parameter with the translation transformation parameter corresponding to the translation direction to obtain a transformation matrix.

[0109] In this embodiment, the 3x3 transformation matrix output by T-Net is replaced by a 4x4 transformation matrix including rotation transformation parameters and translation transformation parameters.

[0110] For example, the 3x3 transformation matrix R is composed of α (rotation around the x-axis), β (rotation around the y-axis), and γ (rotation around the z-axis) as follows:

[0111] R=

[0112] Optionally, the rotation transformation parameter is combined with the translation transformation parameter corresponding to the translation direction to obtain a transformation matrix, including:

[0113] The rotation transformation parameter is used as the first row and first column element in the two-dimensional square matrix, and the translation transformation parameter corresponding to the translation direction is used as the first row and second column element in the two-dimensional square matrix;

[0114] Use 0 as the element of the second row and first column in the two-dimensional square matrix, and use 1 as the element of the second row and second column in the two-dimensional square matrix to obtain the transformation matrix.

[0115] For example, the 3x3 transformation matrix R and the translation vector t are combined into a 4x4 transformation matrix T = .

[0116] For each point in the point cloud, it can be represented as a three-dimensional coordinate vector (x, y, z). In order to apply the transformation matrix, you may need to expand this three-dimensional vector into a four-dimensional vector (x, y, z, 1), depending on whether the transformation matrix is ​​3x3 or 4x4. If it is a 3x3 matrix, only the rotation part needs to be considered, while if it is a 4x4 matrix, both translation and rotation can be handled at the same time.

[0117] Then, multiply this four-dimensional vector by the transformation matrix. This multiplication operation transforms the coordinates of the point according to the parameters in the transformation matrix to obtain the new coordinate position. This process is the same for every point in the point cloud, so a new, aligned point cloud dataset can be obtained.

[0118] Finally, it is worth noting that this transformation process is usually completed in the InputTransform layer of PointNet. Through training, T-Net can learn how to align the input point cloud data into a standardized space for subsequent processing and analysis.

[0119] For example, the PointNet in this embodiment is as follows: Figure 6 As shown:

[0120] First, the model receives unordered point cloud data as input. In order to align and normalize this data, an input transformation network (T-Net) is introduced. This T-Net is a small multi-layer perceptron (MLP) responsible for learning a rigid transformation of the input point cloud to ensure that the model is insensitive to the input order.

[0121] like Figure 7As shown, this embodiment combines the rotation matrix R and the translation vector t into a 4x4 homogeneous transformation matrix T = .

[0122] Next, the data is passed through the first MLP layer for feature extraction. This layer maps the input point cloud from the initial dimension to the feature space, and the output dimension is 64. The MLP layer contains multiple fully connected layers, and nonlinear activation functions (such as ReLU) may be used between layers to enhance the expressiveness of the model.

[0123] The feature data then enters the second transformation network (Feature Transform T-Net). This T-Net is also a small MLP that is used to learn the spatial transformation of the features, further align and normalize the features, and prepare for subsequent processing.

[0124] After being processed by Feature Transform T-Net, the data is fed into the second MLP layer. This layer maps features from the current dimension to a higher dimension in order to more deeply explore the information in the data. Similarly, this layer also contains multiple fully connected layers and non-linear activation functions.

[0125] Next, the model uses the Max Pooling operation to aggregate the features. Since the point cloud data is unordered, Max Pooling can effectively extract global information and aggregate the features of each point into a global feature vector.

[0126] Finally, the data is processed through the third MLP layer. This layer maps the maximum pooled features from high dimensions to low dimensions, with an output dimension of 512×256×k, where k represents the output dimension of the classification or regression task. This layer also contains multiple fully connected layers to further process the features and generate the final output score vector.

[0127] In classification tasks, the output score vector represents the probability of each category; in regression tasks, it represents the specific regression value. Through such a structural design, PointNet can efficiently process unordered point cloud data and demonstrate strong expressiveness and flexibility.

[0128] It should be understood that, although the steps in the flowcharts involved in the above embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0129] Based on the same inventive concept, the embodiment of the present application also provides a laser dynamic point determination device for implementing the laser dynamic point determination method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more laser dynamic point determination device embodiments provided below can refer to the limitations of the laser dynamic point determination method above, and will not be repeated here.

[0130] In an exemplary embodiment, Figure 8 As shown, a laser dynamic point determination device is provided, comprising: an acquisition module 11, an offset parameter transformation module 12, an alignment processing module 13 and a laser dynamic point recognition module 14, wherein:

[0131] An acquisition module 11 is used to acquire three-dimensional point cloud data of each laser point in the target area;

[0132] The offset parameter transformation module 12 is used to determine a transformation matrix according to the geometric information of each three-dimensional point cloud data in the three-dimensional space; wherein the transformation matrix includes rotation transformation parameters and translation transformation parameters;

[0133] An alignment processing module 13 is used to align each 3D point cloud data in the translation direction and rotation direction of the 3D space by using a transformation matrix;

[0134] The laser dynamic point recognition module 14 is used to determine the laser dynamic point among the laser points according to the three-dimensional point cloud data after the alignment processing.

[0135] In one embodiment, the transformation matrix includes rotation transformation parameters and translation transformation parameters, and an offset parameter transformation module is used to: input each three-dimensional point cloud data into a rotation network to obtain a rotation transformation parameter of a rotation direction output by the rotation network;

[0136] The geometric information of the three-dimensional point cloud data of each laser point in the translation direction is analyzed to obtain the translation transformation parameters corresponding to the translation direction.

[0137] In one embodiment, the offset parameter transformation module 12 is used to: the rotation transformation parameter is a rotation transformation parameter in matrix form; using the transformation matrix, aligning each three-dimensional point cloud data in the translation direction and the rotation direction of the three-dimensional space, including:

[0138] The rotation transformation parameter is combined with the translation transformation parameter corresponding to the translation direction to obtain a transformation matrix;

[0139] The three-dimensional point cloud data of each laser point is multiplied by the transformation matrix to obtain the aligned three-dimensional point cloud data of each laser point.

[0140] In one embodiment, the offset parameter transformation module 12 is used to combine the rotation transformation parameter with the translation transformation parameter corresponding to the translation direction to obtain a transformation matrix, including:

[0141] The rotation transformation parameter is used as the first row and first column element in the two-dimensional square matrix, and the translation transformation parameter corresponding to the translation direction is used as the first row and second column element in the two-dimensional square matrix;

[0142] Use 0 as the element of the second row and first column in the two-dimensional square matrix, and use 1 as the element of the second row and second column in the two-dimensional square matrix to obtain the transformation matrix.

[0143] In one embodiment, the laser dynamic point recognition module 12 is used to determine the laser dynamic point in each laser point according to each three-dimensional point cloud data after alignment processing, including:

[0144] Obtain laser point description data of each laser point;

[0145] Determine the laser point features of each laser point according to the laser point description data of each laser point and the three-dimensional point cloud data after alignment processing;

[0146] According to the laser point characteristics of each laser point, the laser dynamic point in each laser point is determined.

[0147] In one embodiment, the laser dynamic point recognition module 13 is used to determine the laser dynamic point among the laser points according to the laser point features of the laser points, including:

[0148] Aggregate the laser point features of each laser point to obtain the global feature;

[0149] According to the global features and the laser point features of each laser point, the laser dynamic point in each laser point is determined.

[0150] In one embodiment, the acquisition module 11 is used to: for each laser point in the target area, acquire the three-dimensional point cloud data of the laser point, including:

[0151] The three-dimensional point cloud data of the laser point is determined based on the length value reflected by the laser point, the horizontal resolution, vertical resolution, horizontal scanning starting angle and vertical scanning starting angle of the laser acquisition device.

[0152] Each module in the above-mentioned laser dynamic point determination device can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0153] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Fig. 9 As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, near field communication (NFC) or other technologies. When the computer program is executed by the processor, a laser dynamic point determination method is implemented. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0154] Those skilled in the art will understand that Fig. 9 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.

[0155] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0156] Obtain three-dimensional point cloud data of each laser point in the target area;

[0157] Determine a transformation matrix according to geometric information of each 3D point cloud data in the 3D space; wherein the transformation matrix includes rotation transformation parameters and translation transformation parameters;

[0158] Using the transformation matrix, the 3D point cloud data is aligned in the translation direction and rotation direction of the 3D space;

[0159] According to the three-dimensional point cloud data after the alignment process, the laser dynamic point in each laser point is determined.

[0160] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0161] Obtain three-dimensional point cloud data of each laser point in the target area;

[0162] Determine a transformation matrix according to geometric information of each 3D point cloud data in the 3D space; wherein the transformation matrix includes rotation transformation parameters and translation transformation parameters;

[0163] Using the transformation matrix, the 3D point cloud data is aligned in the translation direction and rotation direction of the 3D space;

[0164] According to the three-dimensional point cloud data after the alignment process, the laser dynamic point in each laser point is determined.

[0165] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0166] Obtain three-dimensional point cloud data of each laser point in the target area;

[0167] Determine a transformation matrix according to geometric information of each 3D point cloud data in the 3D space; wherein the transformation matrix includes rotation transformation parameters and translation transformation parameters;

[0168] Using the transformation matrix, the 3D point cloud data is aligned in the translation direction and rotation direction of the 3D space;

[0169] According to the three-dimensional point cloud data after the alignment process, the laser dynamic point in each laser point is determined.

[0170] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0171] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0172] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A laser dynamic point determination method, characterized in that: The method comprises: Obtain three-dimensional point cloud data of each laser point in the target area; Determine a transformation matrix according to geometric information of each three-dimensional point cloud data in three-dimensional space; wherein the transformation matrix includes rotation transformation parameters and translation transformation parameters; Using the transformation matrix, aligning each 3D point cloud data in the translation direction and rotation direction of the 3D space; According to the three-dimensional point cloud data after the alignment process, the laser dynamic point in each laser point is determined.

2. The method according to claim 1, characterized in that Determining the transformation matrix according to the geometric information of each three-dimensional point cloud data in the three-dimensional space includes: Input each 3D point cloud data into the rotation network to obtain the rotation transformation parameters corresponding to the rotation direction; Analyzing the geometric information of the three-dimensional point cloud data of each laser point in the translation direction to obtain the translation transformation parameters corresponding to the translation direction; The rotation transformation parameters are combined with the translation transformation parameters to obtain a transformation matrix.

3. The method according to claim 2, characterized in that The step of combining the rotation transformation parameter with the translation transformation parameter to obtain a transformation matrix comprises: The rotation transformation parameter is used as the first row and first column element in a two-dimensional square matrix, and the translation transformation parameter corresponding to the translation direction is used as the first row and second column element in the two-dimensional square matrix; A transformation matrix is ​​obtained by using 0 as the element of the second row and the first column in the two-dimensional square matrix and using 1 as the element of the second row and the second column in the two-dimensional square matrix.

4. The method according to claim 1, characterized in that: Determining the laser dynamic point in each laser point according to each three-dimensional point cloud data after the alignment process includes: Obtain laser point description data of each laser point; Determine the laser point features of each laser point according to the laser point description data of each laser point and the three-dimensional point cloud data after alignment processing; According to the laser point characteristics of each laser point, the laser dynamic point in each laser point is determined.

5. The method according to claim 4, characterized in that The step of determining the laser dynamic point in each laser point according to the laser point characteristics of each laser point includes: Aggregate the laser point features of each laser point to obtain the global feature; According to the global feature and the laser point feature of each laser point, a laser dynamic point in each laser point is determined.

6. The method according to any one of claims 1 to 5, characterized in that: The step of obtaining the three-dimensional point cloud data of each laser point in the target area includes: For each laser point in the target area, the three-dimensional point cloud data of the laser point is determined according to the acquisition configuration information of the laser acquisition device and the length value reflected back by the laser point; The acquisition configuration information of the laser acquisition device includes horizontal resolution, vertical resolution, horizontal scanning starting angle and vertical scanning starting angle.

7. A laser dynamic point determination device, characterized in that: The device comprises: An acquisition module is used to acquire three-dimensional point cloud data of each laser point in the target area; An offset parameter transformation module, used to determine a transformation matrix according to geometric information of each three-dimensional point cloud data in three-dimensional space; wherein the transformation matrix includes rotation transformation parameters and translation transformation parameters; An alignment processing module, used to use the transformation matrix to align each 3D point cloud data in the translation direction and rotation direction of the 3D space; The laser dynamic point recognition module is used to determine the laser dynamic point in each laser point according to each three-dimensional point cloud data after alignment processing.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images