Point cloud data correction method and related equipment

Abstract

The present application discloses a point cloud data correction method and related equipment, relating to the field of point cloud data technology. The method comprises: acquiring RTK data; filtering the RTK data to obtain filtered RTK data; time synchronizing POS data based on the filtered RTK data to generate aligned POS data; correcting the aligned POS data to generate corrected POS data; and performing point cloud solution on the corrected POS data to generate target point cloud data.

Description

Technical Field

[0001] The present application relates to the field of point cloud data technology, and in particular to a point cloud data correction method and related equipment. Background Art

[0002] RTK technology has been widely used in positioning systems for processing high-precision point cloud data. However, RTK data is often subject to errors and noise due to environmental factors such as signal obstruction, multipath effects, and atmospheric delay, resulting in reduced data accuracy.

[0003] Existing point cloud data correction methods typically rely on simple time synchronization and linear error correction, but these methods carry a significant risk of error accumulation, especially in dynamic and nonlinear environments. To address this issue, traditional methods often struggle to accurately capture complex error characteristics, which impacts the quality and accuracy of point cloud data. Therefore, a point cloud data correction method that can adapt to complex environmental changes is urgently needed. Summary of the Invention

[0004] The Summary of the Invention introduces a series of simplified concepts that will be further described in the Detailed Description of the Invention. The Summary of the Invention of this application is not intended to limit the key features and essential technical features of the claimed technical solution, nor is it intended to determine the scope of protection of the claimed technical solution.

[0005] In a first aspect, the present application provides a point cloud data correction method, comprising:

[0006] Get RTK data;

[0007] Filtering the above RTK data to obtain filtered RTK data;

[0008] Based on the above filtered RTK data, the POS data is time synchronized to generate aligned POS data;

[0009] Correcting the aligned POS data to generate corrected POS data;

[0010] Perform point cloud calculation on the above-mentioned corrected POS data to generate target point cloud data.

[0011] In some embodiments, filtering the RTK data to obtain filtered RTK data includes:

[0012] For each time point, calculate the elevation error and horizontal error of the above RTK data;

[0013] Based on the elevation error threshold and the plane error threshold, it is determined whether the elevation error and the horizontal error meet the preset conditions, and the RTK data that meets the preset conditions is screened out to obtain filtered RTK data.

[0014] In some embodiments, the above-mentioned time synchronization of POS data based on the above-mentioned filtered RTK data to generate aligned POS data includes:

[0015] Define RTK data sequence and POS data sequence;

[0016] Calculating the Euclidean distance based on the RTK data sequence and the POS data sequence;

[0017] Based on the above Euclidean distance, a cumulative distance matrix is ​​constructed and the optimal alignment path is determined by backtracking;

[0018] Based on the above optimal alignment path, the POS data is aligned to the above filtered RTK data to generate aligned POS data.

[0019] In some embodiments, correcting the aligned POS data to generate corrected POS data includes:

[0020] Using a dynamic spatiotemporal transformation network to correct the aligned POS data and determine the correction parameters;

[0021] Based on the above correction parameters, preliminary POS data is obtained;

[0022] Based on the above preliminary POS data, an error vector is constructed;

[0023] Define an adaptive weight matrix, optimize the above error vector, and generate an optimized error vector;

[0024] Construct a Bayesian network model to predict the above-mentioned optimized error vector to obtain a predicted error vector;

[0025] Based on the above-mentioned prediction error vector, corrected POS data is generated.

[0026] In some embodiments, performing point cloud calculation on the corrected POS data to generate target point cloud data includes:

[0027] Collecting raw point cloud data, and converting the raw point cloud data from the sensor coordinate system to the global coordinate system;

[0028] Splice and fuse continuously collected point cloud frames to form a continuous 3D point cloud model;

[0029] The voxel grid filtering method and Poisson surface reconstruction algorithm are applied to reconstruct the continuous surface of the point cloud and generate the target point cloud data.

[0030] In some embodiments, the above-mentioned correction of the aligned POS data using a dynamic spatiotemporal transformation network to determine the correction parameters includes:

[0031] Extracting the spatiotemporal features of the filtered RTK data and the aligned POS data through multi-layer convolution and recurrent neural networks;

[0032] Based on the above-mentioned spatiotemporal characteristics, correction parameters are determined.

[0033] In some embodiments, further comprising:

[0034] After generating the target point cloud data, setting point cloud data quality assessment indicators, wherein the above-mentioned quality assessment indicators include point density and surface smoothness;

[0035] Based on the above quality assessment indicators, the overall quality of the above target point cloud data is evaluated;

[0036] Based on the point cloud quality assessment results, the RTK data filtering threshold, correction parameters and filtering weights are dynamically adjusted to optimize the point cloud data generation process.

[0037] In a second aspect, the present application proposes a point cloud data correction device, comprising:

[0038] RTK data acquisition unit, used to acquire RTK data;

[0039] An RTK data filtering unit, configured to filter the RTK data to obtain filtered RTK data;

[0040] a POS data alignment unit, configured to time synchronize the POS data based on the filtered RTK data to generate aligned POS data;

[0041] a POS data correction unit, configured to correct the aligned POS data to generate corrected POS data;

[0042] The point cloud data generating unit is used to perform point cloud calculation on the above-mentioned corrected POS data to generate target point cloud data.

[0043] In a third aspect, an electronic device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor is configured to implement the steps of the point cloud data correction method of any one of the first aspects when executing the computer program stored in the memory.

[0044] In a fourth aspect, the present application further proposes a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the point cloud data correction method of any one of the first aspects is implemented.

[0045] In summary, the point cloud data correction method proposed in this application achieves high-precision RTK data filtering, POS data alignment, and correction by combining a dynamic spatiotemporal transformation network with a Bayesian network. This method not only effectively removes low-quality data but also accurately calculates and optimizes error vectors, significantly improving the overall quality and accuracy of point cloud data. By modeling and predicting the error vector field, combined with voxel grid filtering and Poisson surface reconstruction algorithms, the generated target point cloud data is smoother and noise-free, making it suitable for applications such as high-precision 3D modeling and environmental perception, and possessing greater practicality and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present description. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0047] Figure 1 A schematic diagram of a point cloud data correction method provided in an embodiment of the present application;

[0048] Figure 2 A schematic structural diagram of a point cloud data correction device provided in an embodiment of the present application;

[0049] Figure 3 A schematic diagram of the structure of an electronic device for point cloud data correction provided in an embodiment of the present application. DETAILED DESCRIPTION

[0050] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or that are inherent to these processes, methods, products or devices. The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the embodiments described are only part of the embodiments of the present application, not all of the embodiments.

[0051] See also Figure 1 , which is a flow chart of a point cloud data correction method provided in an embodiment of the present application, which may specifically include:

[0052] S110, acquiring RTK data;

[0053] Exemplarily, step S110 involves the acquisition of RTK (Real-Time Kinematic) data. RTK is a high-precision positioning technology that achieves centimeter-level positioning accuracy through carrier phase differential between a base station and a mobile station. The base station is located at a known high-precision position, continuously receives signals from the Global Navigation Satellite System (GNSS), and transmits the carrier phase information of these signals to the mobile station in real time. The mobile station receives GNSS signals from the base station and its own GNSS signals, and uses carrier phase differential technology to calculate its precise position coordinates in three-dimensional space in real time. This high-precision positioning capability makes RTK data a key reference information source in the point cloud data processing process.

[0054] The acquired RTK data not only includes the current three-dimensional coordinate information, but also includes VDOP (Vertical Dilution of Precision) and HDOP (Vertical Dilution of Precision), which are related to positioning accuracy. VDOP reflects the positioning system's geometric accuracy in the vertical direction; smaller VDOP values ​​indicate higher vertical positioning accuracy; HDOP reflects the geometric accuracy in the horizontal direction; smaller HDOP values ​​indicate higher horizontal positioning accuracy. In addition, each piece of RTK data is timestamped to record the specific time of data acquisition, ensuring the accuracy of subsequent data synchronization and time alignment. Real-time acquisition and recording of this high-precision RTK data provides a reliable positioning baseline for subsequent point cloud data filtering, time synchronization, and correction steps, significantly improving the overall accuracy and reliability of point cloud data processing.

[0055] S120, filtering the RTK data to obtain filtered RTK data;

[0056] For example, the RTK data filtering step first calculates the elevation error and horizontal error of the RTK data at each time point. The elevation error reflects the vertical positioning deviation of the RTK data, while the horizontal error represents the horizontal positioning deviation. Next, the calculated errors are evaluated using an error assessment model based on preset elevation error and horizontal error thresholds. These elevation error and horizontal error thresholds are set using a dynamic adjustment formula that combines the vertical and horizontal dilutions of precision to ensure flexible error tolerance under different environments and positioning conditions.

[0057] During the screening process, if the errors in both the elevation and horizontal directions of the RTK data at a certain point in time are lower than or equal to the corresponding dynamic thresholds, the data point is judged to be high-quality data and retained for subsequent point cloud correction. Conversely, if the error in any direction exceeds the set dynamic threshold, the data point will be eliminated to prevent low-quality data from negatively impacting subsequent processing steps. Through this dynamic threshold-based error screening method, the present application can effectively filter out unreliable RTK data, ensuring that subsequent time synchronization and correction steps are based on high-precision data, thereby significantly improving the overall accuracy and reliability of point cloud data processing.

[0058] First, real-time differential positioning data is obtained from the RTK device, including a timestamp that records the specific time of data acquisition to ensure time alignment of subsequent data; X, Y, and Z coordinates that represent the three-dimensional coordinate information of the current position; and VDOP and HDOP values, which are used to evaluate the geometric factors of positioning accuracy.

[0059] Based on the preset elevation error threshold and plane error threshold, the system judges and filters the RTK data at each time point. The elevation error threshold T z (t) and plane error threshold T xy The dynamic adjustment formula of (t) is as follows:

[0060] T z (t) = α·VDOP(t) β

[0061] T xy (t) = γ·HDOP(t) δ

[0062] Among them, α, β, γ and δ are custom parameters used to adjust the sensitivity of the error threshold and control the response of the error threshold to changes in VDOP and HDOP; T z (t) is the elevation error threshold at time t; T xy (t) is the horizontal error threshold at time t.

[0063] During the screening process, for each time point t, the following screening conditions are executed,

[0064] |Z(t)-Z ref (t)|≤T z (t)

[0065] (X(t)-X ref (t)) 2 +(Y(t)-Y ref (t)) 2 ≤T xy (t)

[0066] Where Z(t), X(t) and Y(t) are the elevation and horizontal coordinates of the RTK device at time t, respectively; Z ref (t), X ref (t) and Y ref (t) is the corresponding elevation and horizontal coordinate of the reference point (from a reference station or a known high-precision position).

[0067] If the RTK data's errors in both elevation and horizontal directions meet the above screening criteria, the data point is retained for subsequent point cloud correction. If the RTK data's error in any direction exceeds the corresponding dynamic threshold, the data point is discarded to prevent low-quality data from negatively impacting subsequent processing.

[0068] S130, based on the filtered RTK data, time synchronize the POS data to generate aligned POS data;

[0069] For example, the filtered RTK data is time-synchronized with the POS (Position and Orientation System) data to generate aligned POS data. Because the sampling frequencies and timestamps of RTK and POS data may differ slightly, direct data alignment can lead to error accumulation and inaccurate synchronization. Therefore, using the DTW (Dynamic Time Warping) algorithm to accurately align the two time series is crucial for ensuring the accuracy of subsequent calibration steps.

[0070] Filtered RTK data M i =(X m,i , Y m，i , Z m,i ) and POS data N j =(X n，j , Y n，j , Z n，j ) are aligned in time. Because there may be slight differences in the sampling frequencies and timestamps between the two, a dynamic time warping algorithm is required for time synchronization. The specific steps include:

[0071] Define two time series and Calculate the Euclidean distance of each pair of moments (i, j), expressed as:

[0072]

[0073] Where D(i, j) is the RTK data point M i and POS data point N j The Euclidean distance between m,iis the coordinate of the RTK data in the X direction at the i-th time step; Y m，i is the coordinate of the RTK data in the Y direction at the i-th time step; Z m,i is the coordinate of the RTK data in the Z direction at the i-th time step; X n,j is the coordinate of the POS data in the X direction at the jth time step; Y n，j is the coordinate of the POS data in the Y direction at the jth time step; Z n,j is the coordinate of the POS data in the Z direction at the jth time step.

[0074] Based on the calculated Euclidean distance, the cumulative distance matrix C(i, j) is defined, and its calculation formula is:

[0075] C(i,j)=D(i,j)+min{C(i-1,j),C(i,j-1),C(i-1,j-1)}

[0076] stC(0,0)=0∩C(i,0)=C(0,j)=∞for i,j>0

[0077] Among them, C(0,0)=0 is the starting point, indicating that both sequences have not started; C(i,0)=∞for i>o means that it is impossible to align the first i elements of sequence 1 with the 0th element of sequence 2; C(0,j)=∞for j>0 means that it is impossible to align the 0th element of sequence 1 with the first j elements of sequence 2.

[0078] Starting from C(N, N), at each step, the minimum value among C(i-1, j), C(i, j-1), and C(i-1, j-1) is selected, moved to the corresponding position, and backtracked to C(1, 1) to determine the optimal time alignment path, thereby achieving time synchronization between the two sequences. C(N, N), located in the lower right corner of the cumulative distance matrix, represents the minimum cumulative distance required to align the first N elements of sequence 1 with the first N elements of sequence 2. It represents the minimum total cost required for the entire DTW alignment process. Using C(N, N), the optimal alignment path for the two sequences can be obtained. This path represents the minimum cumulative distance required to align the POS data with the filtered RTK data, ensuring optimal temporal matching between the two sequences.

[0079] According to the optimal alignment path, POS data N j Align to RTK data M i , generate aligned POS data N′ i =(X′ n，i , Y′ n，i , Z′ n，i ).

[0080] It should be noted that, in the embodiment of the present application, i represents the index of the aligned matching point, that is, each i corresponds to a time step or sampling point. In the alignment process, the RTK data M i and POS data N′ i are matched as a pair of corresponding points; once the alignment is completed, each time step i corresponds to a matching point i, so in the subsequent error calculation and analysis, i naturally becomes the index of the matching point.

[0081] S140, correcting the aligned POS data to generate corrected POS data;

[0082] For example, a dynamic spatiotemporal transformation network is first used to correct the aligned POS data. This network, composed of a multi-layer convolutional neural network and a recurrent neural network, can extract the spatiotemporal features of the filtered RTK data and the aligned POS data and generate parameters for correction. These correction parameters are applied to the POS data via a nonlinear mapping function to obtain preliminary corrected POS data. On this basis, the system constructs an error vector representing the three-dimensional error components between the corrected POS data and the RTK data. To further optimize the error vector, step S140 defines an adaptive weight matrix to weight the error vector and generate an optimized error vector. By introducing a Bayesian network model, high-order statistical feature modeling and prediction are performed on the optimized error vector. The network parameters are dynamically updated using the variational Bayesian inference (VBI) method to achieve accurate prediction and correction of future error vectors. Finally, the predicted error vector is applied to the preliminary corrected POS data to generate the final corrected POS data vector, ensuring high accuracy and consistency of the POS data in terms of position and attitude.

[0083] Generate position and translation correction parameters Θ using a dynamic spatiotemporal transformer network i , the specific steps include:

[0084] Θ i =f DTM (STE(M i , N′ i ))

[0085] Among them, STE is a spatiotemporal encoder, which is composed of multi-layer convolution and recurrent neural networks and is used to extract the spatiotemporal features of data; DTM is the nonlinear mapping function of the dynamic transformation network.

[0086] Based on the correction parameter Θ i , POS data X after preliminary correction c,i =(X c,i , Y c,i ,Z c，i) is expressed as:

[0087] X c，i =X′ n，i +θ i，x ·cos(φ i )-θ i,y ·sin(ψ i )

[0088] Among them, X c，i is the X coordinate of the corrected POS data at the i-th time step; X′ n，i is the X coordinate of the aligned POS data at the i-th time step; θ i，x ·cos(φ i ) is the position correction amount θ i，x After the attitude angle φ i The cosine adjustment represents the effective correction amount in the X direction; θ i，y ·sin(ψ i ) is the position correction amount θ i,y Through the attitude angle ψ i The sinusoidal adjustment represents the effect in the X direction due to the Y direction correction.

[0089] Y c，i =Y′ n，i +θ i,y ·cos(ψ i )+θ i，z ·sin(ω i )

[0090] Among them, Y c，i is the Y coordinate of the corrected POS data at the i-th time step; Y′ n，i is the Y coordinate of the aligned POS data at the i-th time step; θ i，y ·cos(ψ i ) is the position correction amount θ i,y Through the attitude angle ψ i The cosine adjustment represents the effective correction amount in the Y direction; θ i，z ·sin(ω i ) is the position correction amount θ i，z After the attitude angle ω i The sinusoidal adjustment represents the effect in the Y direction due to the Z direction correction.

[0091] Z c，i =Z′ n，i +θ i，z ·cos(ω i )

[0092] Among them, Z c，iis the Z coordinate of the corrected POS data at the i-th time step; Z′ n，i is the Z coordinate of the aligned POS data at the i-th time step; θ i，z ·cos(ω i ) is the position correction amount θ i，z After the attitude angle ω i The cosine adjustment represents the effective correction in the Z direction.

[0093] Define the error vector, expressed as:

[0094] E i =X c,i -M i =(X c，i -X m,i , Y e,i -Y m，i , Z c，i -Z m，i )=(E x，i , E y,i ,E z，i )

[0095] Among them, E i is the overall error vector of the i-th matching point; X c,i =(X c,i , Y c,i ,Z c，i ) is the three-dimensional position of the corrected POS data vector at the i-th time step; M i =(X m,i , Y m，i , Z m,i ) is the three-dimensional position of the RTK data at the i-th time step; (E x，i , E y,i , E z,i ) are the error components of the i-th matching point in the X, Y and Z directions.

[0096] Define the adaptive weight matrix W i =diag(w x，i , w y,i , w z，i ), where the adaptive weight matrix of the i-th matching point is a diagonal matrix; w x,i , w y,i , w z,i are the weights in the X, Y, and Z directions respectively.

[0097] The weight calculation formula is expressed as:

[0098]

[0099] Where, κ dTo adjust the parameters, control the steepness of the weight adjustment, and affect the sensitivity of the weight function; σ d is the error variance in direction d; θ d is the error threshold in the dth direction.

[0100] The error vector field is weighted using the adaptive weight matrix to calculate the corrected error vector E′ i , expressed as:

[0101]

[0102] Among them, E′ i is the weighted error vector of the i-th matching point;

[0103] Optimize the weighted error vector field E′ i To improve the accuracy of error prediction and correction, a Bayesian network that can capture the complex dependencies between POS and RTK data is established to model the high-order statistical characteristics of the error vector field. This can more effectively capture the complex dependencies and uncertainties in the error vector field and achieve accurate prediction and correction of the error vector. The specific steps include:

[0104] Building a Bayesian network Where V is a node set, and each node represents the error vector field E′ i A component E′ in d，i , d∈{x,y,z}, represents the errors in the X, Y, and Z directions; E is the dependency between nodes. For example, the error in the X direction may be related to the error in the Y direction, and this relationship is represented by edges.

[0105] Dynamically update the parameters of the Bayesian network to adapt to changes in the error vector field, including:

[0106] Set the prior distribution p(E′) for each node v∈V d，i ), usually using Gaussian distribution, expressed as:

[0107]

[0108] Among them, μ d is the mean; is the variance.

[0109] Based on the VBI method, the posterior distribution q(E′ d，i ), expressed as:

[0110]

[0111] Among them, μ′ d and It is iteratively updated by VBI to maximize the variational lower bound.

[0112] Using the optimized Bayesian network model, the future error vector Make a prediction, expressed as:

[0113]

[0114] in, For the Bayesian network Next, the error vector E′ i expected value.

[0115] The predicted error vector E′ i Applied to the POS data after preliminary correction, it is expressed as:

[0116]

[0117] Generate the corrected POS data vector, expressed as:

[0118] X 校正,i =(X 校正，i , Y 校正,i , Z 校正，i )

[0119] S150 , performing point cloud calculation on the corrected POS data to generate target point cloud data.

[0120] Exemplarily, step S150 involves resolving the point cloud data using the corrected POS data to generate high-precision target point cloud data. Specifically, this step includes converting the raw point cloud data collected by the sensor from the sensor coordinate system to the global coordinate system, splicing and fusing continuously collected point cloud frames, and optimizing the point cloud data using a voxel grid filtering method and a Poisson surface reconstruction algorithm.

[0121] First, the collected raw point cloud data is converted from the sensor coordinate system to the global coordinate system. This conversion is based on the corrected POS data and attitude angles (including roll, pitch, and yaw), ensuring the accurate positioning of the point cloud data in three-dimensional space. A rotational transformation is implemented using a rotation matrix to reflect changes in the sensor's orientation; a translational transformation adjusts the point cloud's position based on the corrected POS data. This process ensures the consistency and accuracy of the point cloud data in the global coordinate system, laying the foundation for subsequent point cloud stitching and fusion.

[0122] Subsequently, the continuously collected point cloud frames are spliced ​​and fused to form a continuous three-dimensional point cloud model. In this process, the iterative closest point algorithm (ICP) or feature matching method is used to accurately align the point clouds of different time frames to ensure the geometric consistency between the frames. In order to improve the quality and processing efficiency of point cloud data, the voxel grid filtering method is applied to downsample the point cloud, remove outliers and reduce the point cloud density, retaining key geometric features. Finally, the Poisson surface reconstruction algorithm is used to perform continuous surface reconstruction on the filtered point cloud to enhance the geometric details of the point cloud and generate high-precision target point cloud data. Through the above steps, the present invention can effectively generate a three-dimensional point cloud model with high precision and high quality, which is widely applicable to various high-precision positioning and three-dimensional modeling application scenarios.

[0123] In some instances, this also includes:

[0124] After generating the target point cloud data, setting point cloud data quality assessment indicators, wherein the above-mentioned quality assessment indicators include point density and surface smoothness;

[0125] Based on the above quality assessment indicators, the overall quality of the above target point cloud data is evaluated;

[0126] Based on the point cloud quality assessment results, the RTK data filtering threshold, correction parameters and filtering weights are dynamically adjusted to optimize the point cloud data generation process.

[0127] For example, this step first sets quality assessment metrics for the point cloud data, primarily including point density and surface smoothness. Point density measures the number of points per unit volume, ensuring the uniformity and integrity of the spatial distribution of the point cloud data, thereby avoiding information loss or redundancy caused by sparse or overly dense data. Surface smoothness assesses the continuity and detail of the point cloud surface, ensuring the accuracy and realism of the geometric features of the point cloud data.

[0128] The system then conducts an overall quality assessment of the generated target point cloud data based on the set quality assessment indicators. By comparing the point density and surface smoothness of the actual point cloud data with the preset quality standards, the system can objectively determine whether the point cloud data meets the expected accuracy requirements. If the assessment results show that the point cloud data fails to meet the standards in certain indicators, the system will automatically trigger a dynamic adjustment mechanism to optimize the RTK data filtering threshold, correction parameters, and filter weights accordingly. This adjustment process responds to changes in point cloud quality in real time through feedback control, ensuring that subsequently generated point cloud data can be continuously optimized and gradually achieve higher quality standards.

[0129] Finally, based on the results of the point cloud quality assessment, the system dynamically adjusts relevant parameters to optimize the point cloud data generation process. Specifically, when the point density of the point cloud data is too low or the surface smoothness is insufficient, the system will appropriately relax the RTK data filtering threshold to increase the number of valid data points; conversely, it will tighten the threshold to reduce the interference of noisy data. At the same time, the correction parameters and filter weights will also be adjusted based on the feedback from the quality assessment to further improve the accuracy and stability of the point cloud data. This dynamic optimization mechanism not only improves the adaptability and robustness of point cloud data generation, but also ensures that the point cloud data always maintains high quality and high reliability under different environmental conditions, thereby significantly enhancing the technical advantages and practical value of the present invention in practical applications.

[0130] See also Figure 2 , is a schematic structural diagram of a point cloud data correction device provided in an embodiment of the present application, comprising:

[0131] RTK data acquisition unit 21, used to acquire RTK data;

[0132] The RTK data filtering unit 22 is used to filter the RTK data to obtain filtered RTK data;

[0133] a POS data alignment unit 23 for time synchronizing the POS data based on the filtered RTK data to generate aligned POS data;

[0134] a POS data correction unit 24 for correcting the aligned POS data to generate corrected POS data;

[0135] The point cloud data generating unit 25 is used to perform point cloud calculation on the corrected POS data to generate target point cloud data.

[0136] See also Figure 3 An embodiment of the present application also provides an electronic device 300, including a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and executable on the processor. When the processor 320 executes the computer program 311, it implements the steps of any method of point cloud data correction.

[0137] Since the electronic device introduced in this embodiment is a device used to implement a point cloud data correction device in the embodiment of the present application, based on the method introduced in the embodiment of the present application, technical personnel in this field can understand the specific implementation of the electronic device of this embodiment and its various variations. Therefore, how the electronic device implements the method in the embodiment of the present application is no longer introduced in detail here. As long as the equipment used by technical personnel in this field to implement the method in the embodiment of the present application falls within the scope of protection of this application.

[0138] During the specific implementation process, when the computer program 311 is executed by the processor, any implementation method of the embodiments corresponding to the first aspect can be implemented.

[0139] It should be noted that, in the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0140] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-readable storage media containing computer-readable program code.

[0141] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0142] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0143] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0144] The present application also provides a computer program product, which includes computer software instructions. When the computer software instructions are executed on a processing device, the processing device executes Figure 1 A process of a point cloud data correction method in the corresponding embodiment.

[0145] A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function according to the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via wired or wireless means. The computer-readable storage medium can be any available medium that a computer can store or a data storage device such as a server or data center that includes one or more available media integrated therein. The available medium can be a magnetic medium, an optical medium, or a semiconductor medium, etc.

[0146] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0147] In the several embodiments provided in this application, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0148] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0149] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0150] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory, a random access memory, a magnetic disk, or an optical disk.

[0151] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

[0152] Although the preferred embodiments of this specification have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of this specification.

[0153] Obviously, those skilled in the art may make various changes and modifications to this specification without departing from the spirit and scope of this specification. Thus, if such changes and modifications fall within the scope of the claims of this specification and their equivalents, this specification is intended to include such changes and modifications.

Claims

1. A point cloud data correction method, characterized in that: The method comprises: Get RTK data; filtering the RTK data to obtain filtered RTK data; Based on the filtered RTK data, time synchronizing the POS data to generate aligned POS data; Correcting the aligned POS data to generate corrected POS data; Correcting the aligned POS data to generate corrected POS data includes: Correcting the aligned POS data using a dynamic spatiotemporal transformation network to determine correction parameters; Based on the correction parameters, preliminary POS data is obtained; constructing an error vector based on the preliminary POS data; defining an adaptive weight matrix, optimizing the error vector, and generating an optimized error vector; Constructing a Bayesian network model to predict the optimized error vector to obtain a predicted error vector; generating corrected POS data based on the prediction error vector; Perform point cloud calculation on the corrected POS data to generate target point cloud data.

2. The point cloud data correction method according to claim 1, characterized in that: The filtering of the RTK data to obtain filtered RTK data includes: For each time point, calculating the elevation error and horizontal error of the RTK data; Based on the elevation error threshold and the plane error threshold, it is determined whether the elevation error and the horizontal error meet preset conditions, and the RTK data that meets the preset conditions is screened out to obtain filtered RTK data.

3. The point cloud data correction method according to claim 1, characterized in that: The step of time synchronizing the POS data based on the filtered RTK data to generate aligned POS data includes: Define RTK data sequence and POS data sequence; Calculating a Euclidean distance based on the RTK data sequence and the POS data sequence; Based on the Euclidean distance, a cumulative distance matrix is ​​constructed and an optimal alignment path is determined by backtracking; Based on the optimal alignment path, the POS data is aligned to the filtered RTK data to generate aligned POS data.

4. The point cloud data correction method according to claim 1, characterized in that: The performing point cloud calculation on the corrected POS data to generate target point cloud data includes: Collecting raw point cloud data, and converting the raw point cloud data from a sensor coordinate system to a global coordinate system; Splice and fuse continuously collected point cloud frames to form a continuous 3D point cloud model; The voxel grid filtering method and Poisson surface reconstruction algorithm are applied to reconstruct the continuous surface of the point cloud and generate the target point cloud data.

5. The point cloud data correction method according to claim 1, characterized in that: The method of correcting the aligned POS data using a dynamic spatiotemporal transformation network and determining correction parameters includes: Extracting spatiotemporal features of the filtered RTK data and the aligned POS data through multi-layer convolution and recurrent neural networks; Based on the spatiotemporal characteristics, correction parameters are determined.

6. The point cloud data correction method according to claim 1, characterized in that: Also includes: After generating the target point cloud data, setting point cloud data quality assessment indicators, wherein the quality assessment indicators include point density and surface smoothness; Based on the quality assessment index, evaluating the overall quality of the target point cloud data; Based on the point cloud quality assessment results, the RTK data filtering threshold, correction parameters and filtering weights are dynamically adjusted to optimize the point cloud data generation process.

7. A point cloud data correction device, characterized in that: include: RTK data acquisition unit, used to acquire RTK data; an RTK data filtering unit, configured to filter the RTK data to obtain filtered RTK data; a POS data alignment unit, configured to time synchronize the POS data based on the filtered RTK data to generate aligned POS data; a POS data correction unit, configured to correct the aligned POS data to generate corrected POS data; Correcting the aligned POS data to generate corrected POS data includes: Correcting the aligned POS data using a dynamic spatiotemporal transformation network to determine correction parameters; Based on the correction parameters, preliminary POS data is obtained; constructing an error vector based on the preliminary POS data; defining an adaptive weight matrix, optimizing the error vector, and generating an optimized error vector; Constructing a Bayesian network model to predict the optimized error vector to obtain a predicted error vector; generating corrected POS data based on the prediction error vector; The point cloud data generating unit is used to perform point cloud calculation on the corrected POS data to generate target point cloud data.

8. An electronic device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor is configured to implement the steps of the point cloud data correction method according to any one of claims 1 to 6 when executing the computer program stored in the memory.

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 point cloud data correction method according to any one of claims 1 to 6 is implemented.

Images