A Registration Method for InSAR Point Clouds and a Terminal Device

Abstract

This application is applicable to the field of point cloud registration technology and provides an InSAR point cloud registration method and terminal device. The method uses MT-InSAR technology to estimate the height values of InSAR monitoring points, and combines the coordinates of the InSAR monitoring points in an external DEM to obtain InSAR point clouds under different orbits. InSAR point cloud data are clustered using the DBSCAN clustering algorithm with different clustering radii to obtain multiple point cloud classes. For each point cloud class, the facade points and roof points of all buildings are extracted, and each point is projected into a north-east coordinate system to obtain a two-dimensional projection of the InSAR point cloud. Corresponding points are obtained based on the corresponding point search method and two-dimensional projection corresponding to the point cloud class. Corresponding points are matched to obtain corresponding point pairs. Based on the ICP algorithm and corresponding point pairs, InSAR point cloud data are registered. This application can improve the universality and accuracy of InSAR point cloud registration.

Description

Technical Field

[0001] The present application belongs to the field of point cloud registration technology, and in particular relates to an InSAR point cloud registration method and terminal device. Background Art

[0002] Realistic 3D, which provides a realistic, three-dimensional, and time-sequential reflection of the spatiotemporal information of human production, life, and ecological space, is the development direction and fundamental model for surveying and mapping geographic information services. With the continued maturity of radar remote sensing satellite technology, 3D laser scanning technology, and optical stereo mapping technology, the 3D reconstruction of buildings, combining radar remote sensing, laser scanning, and optical mapping, has become a new option for large-scale digital urban mapping.

[0003] Synthetic Aperture Radar (InSAR) interferometry, with its advantages of large range, high precision, and high resolution, has become an effective tool for urban deformation monitoring and building modeling. Advances in Synthetic Aperture Radar (SAR) satellite observations and Multi-Temporal Interferometry SAR (MT-InSAR) technologies, such as Persistent Scatterer Interferometric Synthetic Aperture Radar (PS-InSAR), Distributed Synthetic Aperture Radar (DS-InSAR), and Tomographic Synthetic Aperture Radar (TomoSAR), have laid the foundation for high-resolution surface monitoring. High-resolution SAR provides monitoring results in the form of "point clouds," providing the opportunity to study high-dimensional imaging of the urban built environment, enabling the acquisition of the three-dimensional structure and deformation of individual buildings.

[0004] However, because synthetic aperture radar (SAR) uses side-view imaging, the InSAR point cloud generated from a single track only shows information on one side of a building. At least two or more intersecting SAR tracks are required to generate point clouds to capture the complete structure of the building within the target area. Previous studies have shown that factors such as reference point errors, track errors, atmospheric delay, and satellite timing errors can easily cause 3D coordinate discrepancies between SAR data products from multiple tracks. Therefore, InSAR point cloud registration is a necessary step before applying multi-track SAR data products.

[0005] In the field of InSAR, Gernhardt et al. pioneered a method for registering PS-InSAR point clouds of the urban area of Berlin, Germany, in Universal Transverse Mercator Grid (UTM) coordinates. The method's effectiveness was verified by comparison with lidar data, but its accuracy was limited by inaccurate nearest-neighbor correspondences. Wang et al. automatically aligned ascending and descending TomoSAR point clouds by extracting L-shaped endpoints from building facade point clouds. However, the proposed algorithm relied on accurate point cloud segmentation, which required assistance from an external Geographic Information System (GIS) data source. Furthermore, the final fusion accuracy depended on the number and distribution of quadrilateral buildings in the scene. Zhu and Montazeri et al. used ground control points extracted from geodetic stereo SAR methods to register TomoSAR point clouds from different platforms, effectively fusing multi-track urban TomoSAR point clouds. However, this method required the use of complex external data to correct for signal propagation errors. Wang et al. fused TomoSAR point clouds and optical point clouds using the Anisotropic Iterative Closest Point (ICP) method. They then combined precise deformation estimation from InSAR with semantic classification labels from optical images through semantic texture analysis, proposing a semantic-level deformation monitoring framework for buildings. More recently, Tong et al. proposed a robust registration method for InSAR point clouds from urban area arrays, using the concave and convex facades of buildings for rotation correction and fine-grained displacement. However, this method is limited to buildings with concave and convex facades.

[0006] In summary, the universality and accuracy of traditional InSAR point cloud registration methods are low. Summary of the Invention

[0007] This application provides an InSAR point cloud registration method and terminal device, which can solve the problems of low universality and accuracy of traditional InSAR point cloud registration methods.

[0008] In a first aspect, the present application provides an InSAR point cloud registration method, comprising:

[0009] The height values of InSAR monitoring points were estimated using MT-InSAR technology, and the InSAR point clouds under different orbits were obtained by combining the coordinates of the InSAR monitoring points in the external DEM.

[0010] The InSAR point cloud data of the study area are clustered using the DBSCAN clustering algorithm with different clustering radii to obtain multiple point cloud classes; the InSAR point cloud data includes InSAR point clouds corresponding to multiple tracks, and the point cloud class includes point sets of buildings of the same building type;

[0011] For each of the multiple point cloud classes, extract the facade points and roof points of all buildings in the point cloud class, project the facade points and roof points into a north-east coordinate system, and obtain the two-dimensional projection of the InSAR point cloud corresponding to the buildings in the point cloud class. Then, based on the corresponding point search method and two-dimensional projection corresponding to the point cloud class, obtain the corresponding points of the buildings in the InSAR point cloud of different orbits in the point cloud class; the corresponding point search methods for different point cloud classes are different.

[0012] Match the corresponding points to obtain the corresponding point pairs of the building in different tracks;

[0013] The InSAR point cloud data are registered based on the ICP algorithm and the corresponding point pairs of buildings in different orbits.

[0014] Optionally, the height values of the InSAR monitoring points are estimated using the MT-InSAR technology, and the coordinates of the InSAR monitoring points in the external DEM are combined to obtain the InSAR point clouds under different orbits, including:

[0015] Perform the following operations on the SAR images of each track in different tracks:

[0016] Perform image registration and interference processing on the multi-view SAR images collected from the orbit to obtain amplitude sequence and differential interferogram;

[0017] Perform high coherence scatterer detection on the amplitude sequence to obtain the PS point;

[0018] Perform Delaunay triangulation on PS points to obtain a spatial network;

[0019] Estimate the differential height error and differential deformation parameters on the arcs of the spatial network, and integrate the differential parameter estimates relative to a known reference point to obtain the absolute height error parameters and deformation parameters;

[0020] The height error estimates of the PS points are added to the external DEM coordinates, and the radar timing and updated height of each PS are geocoded using the satellite orbit parameters and the reference ellipsoid to obtain the InSAR point cloud under track.

[0021] Optionally, the different clustering radii include two clustering radii of different sizes.

[0022] Optionally, the InSAR point cloud data can be clustered using the DBSCAN clustering algorithm with different clustering radii to obtain multiple point cloud classes, including:

[0023] For each InSAR point cloud under each track, perform the following operations:

[0024] For each point in the InSAR point cloud, if the number of neighboring points within the preset range of the point is less than the preset neighboring point threshold, the point will be deleted;

[0025] In three-dimensional space, the points in the InSAR point cloud are clustered using the DBSCAN clustering algorithm corresponding to a large clustering radius, and the first point cloud clustering result corresponding to the InSAR point cloud is obtained; the first point cloud clustering result includes the point cloud of large buildings;

[0026] In two-dimensional space, the points in the InSAR point cloud are clustered using the DBSCAN clustering algorithm corresponding to a small clustering radius to obtain the second point cloud clustering result corresponding to the InSAR point cloud; the second point cloud clustering result includes the point cloud of small buildings;

[0027] Merge the first point cloud clustering result and the second point cloud clustering result to obtain multiple initial point cloud classes corresponding to the InSAR point cloud;

[0028] For each of the multiple initial point cloud classes, if the number of points in the initial point cloud class is less than a preset point number threshold, the initial point cloud class is deleted to obtain multiple point cloud classes corresponding to the track.

[0029] Optionally, the point cloud class includes a first type of building point cloud class whose building area is greater than a preset area threshold, a second type of building point cloud class whose building area is less than a preset area threshold, and a third type of building point cloud class with roof points.

[0030] Optionally, when the point cloud class is a building point cloud class, corresponding points of the buildings in the point cloud class in the InSAR point clouds of different orbits are obtained according to the corresponding point search method and two-dimensional projection corresponding to the point cloud class, including:

[0031] For each InSAR point cloud of a building, perform the following steps:

[0032] The vertical points and their two-dimensional coordinates of the InSAR point cloud are obtained according to the density of the two-dimensional projection;

[0033] Initialize the parameter matrix M(ρ,θ) in the Hough space; the initialized parameter matrix M(ρ,θ) is a zero matrix, the row ρ of the parameter matrix M(ρ,θ) represents the distance from the origin to the line in the Hough coordinates, and the column θ of the parameter matrix M(ρ,θ) represents the angle between the line and the axis in the Hough coordinates;

[0034] For each vertical point in the InSAR point cloud, the two-dimensional projection (x i ,y i ), transform the point from rectangular coordinates to Hough space, and obtain the corresponding parameter (ρ i ,θ i ), and replace (ρ i ,θ i )The value of the position corresponding to ) is increased by 1 to obtain the final parameter matrix M'(ρ,θ);

[0035] The straight lines corresponding to the parameters at positions where the values in the final parameter matrix M'(ρ,θ) are greater than the preset threshold are regarded as identified straight lines, multiple identified straight lines are obtained, and the horizontal coordinates of the intersection points of the multiple identified straight lines are obtained;

[0036] The intersection point in the InSAR point cloud is taken as the target point, the average height value of the neighboring points near the target point is calculated, and the average value is used as the vertical coordinate of the target point. The horizontal coordinate of the target point is used as the horizontal coordinate of the corresponding point to obtain the corresponding point.

[0037] Optionally, when the point cloud class is a Class II building point cloud class, corresponding points of the buildings in the point cloud class in the InSAR point clouds of different orbits are obtained according to the corresponding point search method and two-dimensional projection corresponding to the point cloud class, including:

[0038] For each InSAR point cloud corresponding to the second-class building, perform the following operations:

[0039] According to the density of the two-dimensional projection of the InSAR point cloud, the vertical points of the InSAR point cloud and their two-dimensional coordinates are obtained;

[0040] Calculate the distance between each point in the InSAR point cloud based on the two-dimensional coordinates, and use the straight line corresponding to the two points with the farthest distance as the first reference line;

[0041] Calculate the distance between each point in the point cloud class and the first reference line, use the point with the farthest distance as the reference point, and use the line passing through the reference point and perpendicular to the first reference line as the second reference line. Divide the point cloud class according to the second reference line to obtain a first part point set and a second part point set;

[0042] The first part of the point set and the second part of the point set are fitted using the RANSAC algorithm to obtain a third reference straight line and a fourth reference straight line;

[0043] If the third reference line and the fourth reference line are in a vertical relationship, the point cloud class is regarded as an L-shape target point cloud, and the distance between each point in the L-shape target point cloud and the third reference line and the fourth reference line is calculated respectively. The projection point of the point corresponding to the maximum distance on the third reference line and the fourth reference line is obtained, and the projection point is used as the target point. The average value of the height values of the neighboring points near the target point is calculated, and the average value is used as the vertical coordinate of the target point. The horizontal coordinate of the target point is used as the horizontal coordinate of the corresponding point to obtain the corresponding point.

[0044] Optionally, when the point cloud class is a three-type building point cloud class, the corresponding points of the buildings in the point cloud class in the InSAR point clouds of different orbits are obtained according to the corresponding point search method and two-dimensional projection corresponding to the point cloud class, including:

[0045] For each of the three types of buildings, perform the following steps:

[0046] The roof points and their two-dimensional coordinates of the InSAR point cloud are obtained according to the density of the two-dimensional projection of the InSAR point cloud;

[0047] The calculated two-dimensional center coordinates of the roof point and the width of the roof monitoring point on each coordinate axis in the two-dimensional coordinate system; the width includes the width corresponding to the horizontal coordinate axis and the width corresponding to the vertical coordinate axis;

[0048] If the difference is less than or equal to the preset difference threshold, the point corresponding to the coordinates of the two-dimensional center point in the InSAR point cloud is taken as the target point, and the average height value of the neighboring points near the target point is calculated. The average value is used as the vertical coordinate of the corresponding point, and the horizontal coordinate of the target point is used as the horizontal coordinate of the corresponding point to obtain the corresponding point.

[0049] Optionally, matching the corresponding points is performed to obtain corresponding point pairs of the building in different tracks, including:

[0050] For the coordinates of each corresponding point, calculate the distance between corresponding points of different tracks;

[0051] Two corresponding points whose distance is less than a preset distance threshold are regarded as corresponding point pairs.

[0052] Optionally, InSAR point cloud data can be registered based on the ICP algorithm and corresponding point pairs of buildings in different orbits, including:

[0053] Step i: Select the InSAR point cloud of any track from the InSAR point cloud data as the reference point cloud, and use the InSAR point clouds of other tracks except the track as the point clouds to be registered. Use the RANSAC algorithm to obtain the corresponding point pairs between the reference point cloud and each point cloud to be registered.

[0054] Step ii: Based on the ICP algorithm and the corresponding point pairs, the translation parameter R and rotation parameter T for point cloud registration are obtained;

[0055] Step iii, by calculating the formula

[0056]

[0057] Get the registration loss value ∑(R,T); where p n represents the reference point cloud, Q n represents the point cloud to be registered, N represents the number of matching point pairs between the reference point cloud and the point cloud to be registered, ||·|| 2 Indicates calculation of 2 norm;

[0058] In step iv, if the registration loss value is less than the preset registration loss threshold, a rigid body transformation is performed on the point cloud to be registered according to the translation parameter R and rotation parameter T in step ii to obtain the registered InSAR point cloud data; otherwise, return to step i.

[0059] In a second aspect, the present application provides a terminal device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned InSAR point cloud registration method when executing the computer program.

[0060] In a third aspect, the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned InSAR point cloud registration method is implemented.

[0061] The above solution of the present application has the following beneficial effects:

[0062] The InSAR point cloud registration method provided in this application utilizes the DBSCAN clustering algorithm with different clustering radii to cluster the InSAR point cloud data of the study area, and can accurately extract point clouds corresponding to different buildings and obtain accurate point cloud classes, thereby improving the accuracy of InSAR point cloud registration; according to the corresponding point search method corresponding to different point cloud classes, the corresponding points of buildings in each point cloud class in the InSAR point clouds of different tracks are obtained, and the corresponding points of different types of buildings can be extracted, thereby improving the universality of the method. At the same time, the corresponding point pairs extracted in this way are more accurate, thereby improving the accuracy of InSAR point cloud registration.

[0063] Other beneficial effects of the present application will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments or descriptions of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0065] Figure 1 A flowchart of an InSAR point cloud registration method provided in one embodiment of the present application;

[0066] Figure 2a A schematic diagram of the detection principle of a method for detecting intersections of straight lines provided in one embodiment of the present application;

[0067] Figure 2b A schematic diagram of the detection principle of an L-shaped corner point detection method provided in one embodiment of the present application;

[0068] Figure 2c A schematic diagram of the detection principle of a roof center point detection method provided in one embodiment of the present application;

[0069] Figure 3 A schematic diagram of the structure of a terminal device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0070] In the following description, specific details such as specific system structures and techniques are provided for purposes of illustration rather than limitation to facilitate a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obscuring the description of the present application with unnecessary detail.

[0071] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections thereof.

[0072] It will also be understood that the term "and / or" used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0073] As used in this specification and the appended claims, the term "if" can be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]," depending on the context.

[0074] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0075] References to "one embodiment" or "some embodiments" in this specification mean that a particular feature, structure, or characteristic described in conjunction with that embodiment is included in one or more embodiments of the present application. Thus, phrases such as "in one embodiment," "in some embodiments," "in other embodiments," and "in other embodiments" appearing in various places in this specification do not necessarily refer to the same embodiment, but rather mean "one or more but not all embodiments," unless otherwise specifically emphasized. The terms "including," "comprising," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0076] In response to the problems of low universality and accuracy of traditional InSAR point cloud registration methods, the present application provides an InSAR point cloud registration method and terminal device. The registration method uses the DBSCAN clustering algorithm with different clustering radii to cluster the InSAR point cloud data of the study area, which can accurately extract point clouds corresponding to different buildings and obtain accurate point cloud classes, thereby improving the accuracy of InSAR point cloud registration; according to the corresponding point search method corresponding to different point cloud classes, the corresponding points of buildings in each point cloud class in InSAR point clouds of different tracks are obtained, and the corresponding points of different types of buildings can be extracted, which improves the universality of the method. At the same time, the corresponding point pairs extracted in this way are more accurate, thereby improving the accuracy of InSAR point cloud registration.

[0077] like Figure 1 As shown, the InSAR point cloud registration method provided in this application includes the following steps:

[0078] In step 11, the height values of the InSAR monitoring points are estimated using the MT-InSAR technology, and the InSAR point clouds under different orbits are obtained by combining the coordinates of the InSAR monitoring points in the external DEM.

[0079] The above-mentioned InSAR point cloud includes PSI point cloud and TomoSAR point cloud.

[0080] In the embodiments of the present application, the multi-orbit SAR data can be acquired via the Gaofen-3 (GF-3) satellite, RadarSat-2 (a high-resolution commercial radar satellite equipped with a C-band sensor), ALOS-2, TerraSAR-X (TerraSAR-X add-on for Digital Elevation Measurement) / TanDEM-X, COSMO-SkyMed (the second satellite in the COSMO-SkyMed high-resolution radar satellite constellation, jointly developed by the Italian Space Agency and the Italian Ministry of Defense), Hisea-1, and ICEYE. The number of images is greater than 20.

[0081] The following is an exemplary description of the process of obtaining InSAR point cloud data in step 11.

[0082] In step i, a master image is selected from the N+1 image stacks for image registration, and then interferometry is performed to obtain N interferograms, while their topographic phase components are removed using a reference digital elevation model (DEM).

[0083] Exemplarily, the i-th interferometric phase consisting of the master and slave images is It can be expressed as: in, represents the deformation phase, represents the terrain residual phase, represents the atmospheric delay phase, represents the noise phase.

[0084] In step ii, coherent scatterers (PS) are detected using various methods, such as thresholding the amplitude dispersion index (ADI) or signal-to-noise ratio (SCR). These candidate PSs are connected to form a reference network. The differential height error and differential deformation parameters are estimated over the arc using double-difference PS phase measurements, i.e., differences in time and space.

[0085] Specifically, the time-series SAR imagery is preprocessed to generate a tangled differential interferogram stack. High-quality PS candidate points are identified based on discrete amplitude thresholds. These candidate points are then linked using a Delaunay triangulation network to generate a first-layer spatial network, which serves as a reference for the second-layer network. Building on the first-layer network, the second-layer network identifies more high-quality PS points and constructs a star-shaped network around these newly identified PS points, resulting in dense PS arcs and increasing the density of monitoring points.

[0086] When the constructed PS arc meets the conditions The height error Δε and deformation rate Δv can be estimated from the M winding differential interferograms. Specifically, the height error Δε and deformation rate Δv can be solved by using the maximum optimized value of the following function:

[0087]

[0088] Among them, γ represents the temporal coherence coefficient of the arc segment, Δω i represents the difference between the observed value and the fitted value corresponding to the i-th winding differential interferogram, λ represents the radar wavelength, R represents the slant range, θ represents the satellite incident angle, ΔT i represents the time baseline of the ith winding differential interferogram, represents the spatial baseline of the i-th entangled differential interferogram.

[0089] Then, by searching in two-dimensional space and using the periodic spectrum method to estimate each arc segment, the elevation error and deformation rate of the corresponding arc segment can be obtained.

[0090] In step iii, the differential parameter estimates are integrated relative to the selected reference point to obtain the absolute height error parameter and the deformation parameter, and the two phase components (deformation component and height component) are removed from the interferometric phase. The remaining phase contributions include nonlinear deformation, atmosphere, and noise. The atmospheric component is then extracted using low-pass filtering in the spatial domain and high-pass filtering in the temporal domain. This atmospheric component is interpolated over the entire scene and subtracted from the differential interferogram.

[0091] In step iv, the remaining PSs are detected based on the corrected differential interferograms, connected to the nearest points in the reference network to increase the monitoring point density, and their modeling parameters are estimated.

[0092] Specifically, due to the potential for phase discontinuities in built environments, it is necessary to optimally globally integrate the differential increments (height error Δε and deformation rate Δv) to obtain the terrain residual and deformation values for each monitoring point. In the embodiments of this application, low-quality arcs are first removed through outlier detection, and then a subset network is constructed to increase the spatial coverage of the original spatial network. On this basis, ridge estimation is used to adjust and reduce the ill-conditioning of the normal equation matrix, resulting in a globally optimal estimate of the parameter values.

[0093] The expression of step iv is as follows:

[0094] B*X=L+R

[0095] X=(B T PB+σI) -1 BT PL

[0096] Where B represents the coefficient matrix, L represents the observation value, R represents the residual, X represents the estimated deformation rate and height error of all monitoring points, σ represents the regularization factor, I represents the identity matrix, and P represents the weight matrix with a diagonal form, which can be determined by the time series coherence coefficient.

[0097] In step v, the DEM heights of all PSs are added to their height error estimates, and the radar timing and updated height of each PS are geocoded using the satellite orbit parameters and the reference ellipsoid to produce PS monitoring points in the geodetic coordinate system to obtain the InSAR point cloud.

[0098] In the embodiment of the present application, before clustering the InSAR point cloud data, it is necessary to filter the InSAR point cloud data of the study area using a filter, as follows:

[0099] The presence of SAR image speckle noise and estimation errors causes InSAR point clouds to contain a large number of noise points (isolated points), which hinders the point cloud registration process and the integrated application of different types of point clouds. Commonly used point cloud filters can effectively filter out InSAR noise points. For example, statistical filters can eliminate most noise and abnormal points based on the mean and standard deviation relationship of the neighborhood point distance by performing statistical analysis on the neighborhood of each point. The radius filter counts the number of points in a fixed range and excludes noise points based on a number threshold. In InSAR point clouds, urban buildings are mostly manifested as incomplete facades and roofs. The facade and roof points have different densities on the horizontal projection plane. The building facade points and roof points can be extracted by thresholding the point density. The above two filters are only for illustrative purposes. In actual applications, other common filters can be used to filter InSAR point cloud data. Their specific forms and steps are not limited here.

[0100] In step 12, the InSAR point cloud data of the study area are clustered using the DBSCAN clustering algorithm with different clustering radii to obtain multiple point cloud classes.

[0101] The InSAR point cloud data includes InSAR point clouds corresponding to multiple tracks, and the point cloud class includes point sets of buildings of the same building type.

[0102] In an embodiment of the present application, the minimum number of included points of the density-based clustering algorithm (DBSCAN, Density-Based Spatial Clustering of Applications with Noise) is set to 5. This can filter out noise points to a certain extent and produce a more complete clustering result. The above-mentioned different clustering radii include two clustering radii of different sizes.

[0103] In an embodiment of the present application, the point cloud class includes a first type of building point cloud class whose building area is greater than a preset area threshold, a second type of building point cloud class whose building area is less than a preset area threshold, and a third type of building point cloud class with roof points.

[0104] Step 13: For each point cloud class in the multiple point cloud classes, extract the facade points and roof points of all buildings in the point cloud class, project the facade points and roof points into the north-east coordinate system, obtain the two-dimensional projection of the InSAR point cloud corresponding to the building in the point cloud class, and obtain the corresponding points of the building in the point cloud class in the InSAR point cloud of different tracks according to the corresponding point search method and two-dimensional projection corresponding to the point cloud class.

[0105] Among them, the corresponding point search methods corresponding to different point cloud classes are different.

[0106] Step 14: Match the corresponding points to obtain corresponding point pairs of the building in different tracks.

[0107] Step 15: Based on the ICP algorithm and the corresponding point pairs of buildings in different orbits, the InSAR point cloud data is registered.

[0108] The ICP algorithm is essentially an iterative registration method that integrates random sampling consensus (RANSAC) and least squares estimation. The algorithm repeats the steps of selecting matching point pairs and estimating the rigid body transformation matrix until the convergence accuracy requirements for correct registration are met. The ICP algorithm will eventually find the optimal rotation parameter R and translation parameter T between the point cloud data to be registered and the reference point cloud data, achieving precise alignment of the coordinate systems of the two point cloud data. The advantage of RANSAC is that it can robustly estimate model parameters. It can estimate high-precision parameters from a data set containing a large number of outliers, which can be used to select the correct matching point pairs.

[0109] The process of step 12 (clustering the InSAR point cloud data of the study area using the DBSCAN clustering algorithm with different clustering radii to obtain multiple point cloud classes) is exemplified below.

[0110] For each track in the InSAR point cloud data, perform the following operations:

[0111] For each point in the InSAR point cloud, if the number of neighboring points within the preset range of the point is less than the preset neighboring point threshold, the point is deleted.

[0112] In three-dimensional space, the points in the InSAR point cloud are clustered using the DBSCAN clustering algorithm corresponding to a large clustering radius to obtain the first point cloud clustering result corresponding to the InSAR point cloud.

[0113] Exemplarily, the first point cloud clustering result includes a large building point cloud and a dense building cluster point cloud.

[0114] In two-dimensional space, the points in the InSAR point cloud are clustered using the DBSCAN clustering algorithm corresponding to a small clustering radius to obtain the second point cloud clustering result corresponding to the InSAR point cloud.

[0115] Exemplarily, the second point cloud clustering result includes small building point clouds, high-rise building point clouds, and small and medium-sized building point clouds.

[0116] The first point cloud clustering results and the second point cloud clustering results are merged to obtain multiple initial point cloud classes corresponding to the InSAR point cloud.

[0117] For each of the multiple initial point cloud classes, if the number of points in the initial point cloud class is less than a preset point number threshold, the initial point cloud class is deleted to obtain multiple point cloud classes corresponding to the track.

[0118] Next, step 13 (for each point cloud class in the multiple point cloud classes, extracting the facade points and roof points of all buildings in the point cloud class, projecting the facade points and roof points into the north-east coordinate system, obtaining the two-dimensional projection of the InSAR point cloud corresponding to the buildings in the point cloud class, and obtaining the corresponding points of the buildings in the point cloud class in the InSAR point clouds of different tracks according to the corresponding point search method and two-dimensional projection corresponding to the point cloud class) is performed. The corresponding points of the buildings in the point cloud class in the InSAR point clouds of different tracks are obtained according to the corresponding point search method and two-dimensional projection corresponding to the point cloud class; the corresponding point search methods corresponding to different point cloud classes are different.

[0119] Case 1: When the point cloud class is a building point cloud class.

[0120] For each InSAR point cloud of a building, perform the following steps:

[0121] The vertical points and their two-dimensional coordinates of the InSAR point cloud are obtained according to the density of the two-dimensional projection;

[0122] Initialize the parameter matrix M(ρ,θ) in the Hough space; the initialized parameter matrix M(ρ,θ) is a zero matrix, the row ρ of the parameter matrix M(ρ,θ) represents the distance from the origin to the line in the Hough coordinates, and the column θ of the parameter matrix M(ρ,θ) represents the angle between the line and the axis in the Hough coordinates;

[0123] For each vertical point in the InSAR point cloud, the two-dimensional projection (x i ,y i ), transform the point from rectangular coordinates to Hough space, and obtain the corresponding parameter (ρ i ,θ i ), and replace (ρ i ,θ i )The value of the position corresponding to ) is increased by 1 to obtain the final parameter matrix M'(ρ,θ);

[0124] The straight lines corresponding to the parameters at the positions where the values in the final parameter matrix M'(ρ,θ) are greater than the preset threshold are regarded as identified straight lines, and multiple identified straight lines are obtained (such as Figure 2a and obtaining the horizontal coordinates of the intersection points of the multiple identified straight lines;

[0125] The intersection point in the InSAR point cloud is taken as the target point, and the average value of the height values of the neighboring points near the target point is calculated. The average value is used as the vertical coordinate of the target point, and the horizontal coordinate of the target point is used as the horizontal coordinate of the corresponding point to obtain the corresponding point (such as Figure 2a mid-corner).

[0126] Case 2: When the point cloud class is the second-class building point cloud class.

[0127] For each InSAR point cloud corresponding to the second-class building, perform the following steps:

[0128] For the second type of buildings with smaller areas and shorter side lengths, it is difficult to use Hough transform to detect point cloud straight lines. The following steps should be used for this point cloud class:

[0129] According to the density of the two-dimensional projection of the InSAR point cloud, the vertical points of the InSAR point cloud and their two-dimensional coordinates are obtained;

[0130] Calculate the distance between each point in the InSAR point cloud based on the two-dimensional coordinates, and use the straight line corresponding to the two points with the farthest distance as the first reference line;

[0131] Calculate the distance between each point in the point cloud class and the first reference line, take the point with the farthest distance as the reference point, and take the line passing through the reference point and perpendicular to the first reference line as the second reference line. Divide the point cloud class according to the second reference line to obtain the first part of the point set (such as Figure 2b Part 1) and the second part of the point set (as shown in Figure 2b (as shown in part 2);

[0132] The RANSAC algorithm is used to fit the first part of the point set and the second part of the point set respectively to obtain the third reference straight line and the fourth reference straight line.

[0133] If the third reference line and the fourth reference line are in a vertical relationship, the point cloud class is regarded as an L-shape target point cloud, and the distances between each point in the L-shape target point cloud and the third reference line and the fourth reference line are calculated respectively. The projection points of the points corresponding to the maximum distance on the third reference line and the fourth reference line are obtained, and the projection points are used as the target points. The average value of the height values of the neighboring points near the target point is calculated, and the average value is used as the vertical coordinate of the target point. The horizontal coordinate of the target point is used as the horizontal coordinate of the corresponding point, and the corresponding point is obtained (such as Figure 2b mid-corner).

[0134] Case three, when the point cloud class is the third type of building point cloud class.

[0135] For each of the three types of buildings, perform the following steps:

[0136] The roof points and their two-dimensional coordinates of the InSAR point cloud are obtained according to the density of the two-dimensional projection of the InSAR point cloud;

[0137] The calculated two-dimensional center coordinates of the roof point and the corresponding width of the roof monitoring point on each coordinate axis in the two-dimensional coordinate system;

[0138] The above width includes the width ΔX corresponding to the horizontal coordinate axis and the width ΔY corresponding to the vertical coordinate axis (such as Figure 2c (as shown in the point cloud of the middle and lower orbits).

[0139]

[0140] Among them, P center and Q center Represents the roof center point of different tracks, R and T represent the rotation matrix and bias matrix, ΔX A and ΔX D Represents the width of the roof point of different tracks on the horizontal coordinate axis, ΔY A and ΔY D Represents the width of the roof point of different tracks on the vertical coordinate axis.

[0141] Calculate the difference between the widths ΔX and ΔY of the roof monitoring points of different tracks. If the difference is less than or equal to the preset difference threshold, the point corresponding to the coordinates of the two-dimensional center point in the InSAR point cloud is taken as the target point. Calculate the average value of the height values of the neighboring points near the target point, and use the average value as the vertical coordinate of the corresponding point. Use the horizontal coordinate of the target point as the horizontal coordinate of the corresponding point to obtain the corresponding point (such as Figure 2c (as indicated by the center point of the middle lift rail roof).

[0142] The specific process of step 14 (matching corresponding points to obtain corresponding point pairs of buildings in different tracks) is exemplified below.

[0143] Step 14.1: For each coordinate of a corresponding point, calculate the distance between corresponding points of different tracks.

[0144] Step 14.2: Two corresponding points whose distance is less than a preset distance threshold are regarded as corresponding point pairs.

[0145] The process of step 15 (registering the InSAR point cloud data based on the ICP algorithm and the corresponding point pairs of buildings in different orbits) is exemplarily described below.

[0146] In step i, the InSAR point cloud of any track is selected from the InSAR point cloud data as the reference point cloud, and the InSAR point clouds of other tracks except the track are used as the point clouds to be registered. The RANSAC algorithm is used to obtain the corresponding point pairs between the reference point cloud and each point cloud to be registered.

[0147] It should be noted that, in one embodiment of the present application, when the InSAR point cloud data is ascending or descending InSAR point cloud data, the ascending InSAR point cloud is used as the reference point cloud, and the descending InSAR point cloud is used as the point cloud to be registered.

[0148] It should be noted that in the registration process, the weight of the corresponding point pairs of type 1 and type 2 buildings is set to 1, and the weight of the corresponding point pairs of type 3 buildings is set to Where ΔX A and ΔX D They represent the width of the roof point of the lifting rail on the horizontal coordinate axis, ΔY A and ΔY D They represent the width of the roof point of the lifting rail on the vertical coordinate axis.

[0149] In step ii, based on the ICP algorithm and the corresponding point pairs, the translation parameter R and rotation parameter T for point cloud registration are obtained.

[0150] Step iii, by calculating the formula

[0151]

[0152] Get the registration loss value ∑(R,T); where p n represents the reference point cloud, Q n represents the point cloud to be registered, N represents the number of matching point pairs between the reference point cloud and the point cloud to be registered, ||·|| 2 Indicates calculation of the 2-norm.

[0153] In step iv, if the registration loss value is less than the preset registration loss threshold, a rigid body transformation is performed on the point cloud to be registered according to the translation parameter R and rotation parameter T in step ii to obtain the registered InSAR point cloud data; otherwise, return to step i.

[0154] For example, the registration loss threshold may be set to 10^-5, and the number of iterations may be set to 20.

[0155] like Figure 3 As shown, an embodiment of the present application provides a terminal device, as shown in Figure 2, the terminal device D10 of this embodiment includes: at least one processor D100 (only one processor is shown in Figure 2), a memory D101, and a computer program D102 stored in the memory D101 and executable on the at least one processor D100, and when the processor D100 executes the computer program D102, the steps in any of the above-mentioned method embodiments are implemented.

[0156] Specifically, when the processor D100 executes the computer program D102, the InSAR point cloud data of the study area are clustered using the DBSCAN clustering algorithm with different clustering radii to obtain multiple point cloud classes, and then for each point cloud class in the multiple point cloud classes, the facade points and roof points of all buildings in the point cloud class are extracted, and the facade points and roof points are projected into the northeast coordinate system to obtain the two-dimensional projection of the InSAR point cloud corresponding to the building in the point cloud class, and according to the corresponding point search method and two-dimensional projection corresponding to the point cloud class, the corresponding points of the building in the point cloud class in the InSAR point cloud in different orbits are obtained, and then the corresponding points are matched to obtain corresponding point pairs of the building in different orbits. Finally, based on the ICP algorithm and the corresponding point pairs of the building in different orbits, the InSAR point cloud data of the study area are aligned. Among them, the InSAR point cloud data of the study area are clustered using the DBSCAN clustering algorithm with different clustering radii, which can accurately extract the point clouds corresponding to different buildings and obtain accurate point cloud classes, thereby improving the accuracy of InSAR point cloud registration; according to the corresponding point search method corresponding to different point cloud classes, the corresponding points of buildings in each point cloud class in the InSAR point clouds of different tracks are obtained, and the corresponding points of different types of buildings can be extracted, which improves the universality of the method. At the same time, the corresponding point pairs extracted in this way are more accurate, thereby improving the accuracy of InSAR point cloud registration.

[0157] The processor D100 may be a central processing unit (CPU), or may be another general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. A general-purpose processor may be a microprocessor or any conventional processor.

[0158] In some embodiments, the memory D101 may be an internal storage unit of the terminal device D10, such as a hard disk or memory of the terminal device D10. In other embodiments, the memory D101 may also be an external storage device of the terminal device D10, such as a plug-in hard disk, a smart memory card (SMC, SmartMedia Card), a secure digital (SD, Secure Digital) card, a flash card, etc. equipped on the terminal device D10. Furthermore, the memory D101 may also include both an internal storage unit of the terminal device D10 and an external storage device. The memory D101 is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program. The memory D101 may also be used to temporarily store data that has been output or is to be output.

[0159] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned various method embodiments can be implemented.

[0160] An embodiment of the present application provides a computer program product. When the computer program product is run on a terminal device, the terminal device can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0161] 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 present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by the processor, it can implement the steps of the above-mentioned various method embodiments. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include at least: any entity or device capable of carrying the computer program code to the terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electric carrier signal, a telecommunication signal and a software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electric carrier signals and telecommunication signals.

[0162] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0163] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0164] In the embodiments provided in this application, it should be understood that the disclosed devices / network equipment and methods can be implemented in other ways. For example, the device / network equipment embodiments described above are merely illustrative. For example, the division of the modules or units is merely 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 through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0165] The units described as separate components may or may not be physically separate, and the 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.

[0166] The above is a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles described in the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A registration method for InSAR point clouds, characterized in that: include: The height values of InSAR monitoring points were estimated using MT-InSAR technology, and the InSAR point clouds under different orbits were obtained by combining the coordinates of the InSAR monitoring points in the external DEM. Clustering the InSAR point cloud data using the DBSCAN clustering algorithm with different clustering radii to obtain a plurality of point cloud classes; the InSAR point cloud data includes InSAR point clouds corresponding to a plurality of tracks, and the point cloud class includes a point set of buildings of the same building type; For each point cloud class in the plurality of point cloud classes, extracting facade points and roof points of all buildings in the point cloud class, projecting the facade points and roof points into a north-east coordinate system to obtain a two-dimensional projection of an InSAR point cloud corresponding to the building in the point cloud class, and obtaining corresponding points of the building in the point cloud class in InSAR point clouds of different orbits based on a corresponding point search method corresponding to the point cloud class and the two-dimensional projection; the corresponding point search methods corresponding to different point cloud classes are different; Matching the corresponding points to obtain corresponding point pairs of the building on different tracks; The InSAR point cloud data is registered based on the ICP algorithm and the corresponding point pairs of the buildings in different orbits; The method of estimating the height values of InSAR monitoring points using the MT-InSAR technology and obtaining InSAR point clouds under different orbits by combining the coordinates of the InSAR monitoring points in the external DEM includes: Perform the following operations on the SAR images of each track in different tracks: Performing image registration and interference processing on the multi-scene SAR images collected on the track to obtain an amplitude sequence and a differential interferogram; Performing high coherence scatterer detection on the amplitude sequence to obtain PS points; Performing Delaunay triangulation on the PS points to obtain a spatial network; estimating differential height errors and differential deformation parameters on arcs of the spatial network, and integrating the differential parameter estimates relative to a known reference point to obtain absolute height error parameters and deformation parameters; The altitude error estimates of the PS points are added to the external DEM coordinates, and the radar timing and updated altitude of each PS are geocoded using the satellite orbit parameters and the reference ellipsoid to obtain the InSAR point cloud under the orbit.

2. The registration method according to claim 1, characterized in that The different DBSCAN clustering radii include two clustering radii of different sizes; The InSAR point cloud data are clustered using the DBSCAN clustering algorithm with different clustering radii to obtain multiple point cloud classes, including: For each InSAR point cloud under each track, perform the following operations: For each point in the InSAR point cloud, if the number of neighboring points within a preset range of the point is less than a preset neighboring point threshold, the point is deleted; In three-dimensional space, clustering the points in the InSAR point cloud using a DBSCAN clustering algorithm corresponding to a large clustering radius to obtain a first point cloud clustering result corresponding to the InSAR point cloud; the first point cloud clustering result includes a point cloud of a large building; In a two-dimensional space, clustering the points in the InSAR point cloud using a DBSCAN clustering algorithm corresponding to a small clustering radius to obtain a second point cloud clustering result corresponding to the InSAR point cloud; the second point cloud clustering result includes a point cloud of a small building; Merging the first point cloud clustering result and the second point cloud clustering result to obtain a plurality of initial point cloud classes corresponding to the InSAR point cloud; For each of the multiple initial point cloud classes, if the number of points in the initial point cloud class is less than a preset point number threshold, the initial point cloud class is deleted to obtain multiple point cloud classes corresponding to the track.

3. The registration method according to claim 2, characterized in that: The point cloud classes include a first type of building point cloud class whose building area is greater than a preset area threshold, a second type of building point cloud class whose building area is less than a preset area threshold, and a third type of building point cloud class with roof points.

4. The registration method according to claim 3, wherein: When the point cloud class is a building point cloud class, obtaining corresponding points of the building in the point cloud class in the InSAR point clouds of different orbits according to the corresponding point search method corresponding to the point cloud class and the two-dimensional projection includes: For each InSAR point cloud of a building, perform the following steps: Obtaining the vertical points and the two-dimensional coordinates of the InSAR point cloud according to the density of the two-dimensional projection; Initialize the parameter matrix M(ρ,θ) in the Hough space; wherein the initialized parameter matrix M(ρ,θ) is a zero matrix, the row ρ of the parameter matrix M(ρ,θ) represents the distance from the origin to the straight line in the Hough coordinates, and the column θ of the parameter matrix M(ρ,θ) represents the angle between the straight line and the axis in the Hough coordinates; For each elevation point in the InSAR point cloud, the two-dimensional projection (x i ,y i ), transform the point from rectangular coordinates to Hough space, and obtain the corresponding parameter (ρ i ,θ i ), and replace (ρ i ,θ i )The value of the position corresponding to ) is increased by 1 to obtain the final parameter matrix M'(ρ,θ); Taking the straight lines corresponding to the parameters at positions where the values of the parameters in the final parameter matrix M'(ρ,θ) are greater than a preset threshold as the identified straight lines, obtaining multiple identified straight lines, and obtaining the horizontal coordinates of the intersection points of the multiple identified straight lines; The intersection point in the InSAR point cloud is used as the target point, the average height value of the neighboring points near the target point is calculated, and the average value is used as the vertical coordinate of the target point, and the horizontal coordinate of the target point is used as the horizontal coordinate of the corresponding point to obtain the corresponding point.

5. The registration method according to claim 3, characterized in that: When the point cloud class is the second type of building point cloud class, obtaining corresponding points of the building in the point cloud class in the InSAR point clouds of different tracks according to the corresponding point search method corresponding to the point cloud class and the two-dimensional projection includes: For each InSAR point cloud corresponding to the second-class building, perform the following steps: Obtaining the elevation points of the InSAR point cloud and their two-dimensional coordinates according to the density of the two-dimensional projection of the InSAR point cloud; Calculating the distance between points in the InSAR point cloud according to the two-dimensional coordinates, and using the straight line corresponding to the two points with the greatest distance as a first reference straight line; Calculating the distance between each point in the point cloud class and the first reference line, taking the point with the farthest distance as a reference point, taking a line passing through the reference point and perpendicular to the first reference line as a second reference line, and dividing the point cloud class according to the second reference line to obtain a first partial point set and a second partial point set; Using the RANSAC algorithm to fit the first part of the point set and the second part of the point set respectively, to obtain a third reference straight line and a fourth reference straight line; If the third reference straight line and the fourth reference straight line are in a vertical relationship, the point cloud class is taken as an L-shape target point cloud, and the distance between each point in the L-shape target point cloud and the third reference straight line and the fourth reference straight line is calculated respectively, and the projection point of the point corresponding to the maximum distance on the third reference straight line and the fourth reference straight line is obtained, and the projection point is used as the target point, and the average value of the height values of the neighboring points near the target point is calculated, and the average value is used as the vertical coordinate of the target point, and the horizontal coordinate of the target point is used as the horizontal coordinate of the corresponding point to obtain the corresponding point.

6. The registration method according to claim 3, characterized in that: When the point cloud class is the three types of building point cloud classes, obtaining corresponding points of the buildings in the point cloud class in the InSAR point clouds of different tracks according to the corresponding point search method corresponding to the point cloud class and the two-dimensional projection includes: For each of the three types of buildings, perform the following steps: Obtaining the roof points of the InSAR point cloud and their two-dimensional coordinates according to the density of the two-dimensional projection of the InSAR point cloud; Calculate the two-dimensional center point coordinates of the roof point and the width corresponding to each coordinate axis in the two-dimensional coordinate system of the roof monitoring point; the width includes the width corresponding to the horizontal coordinate axis and the width corresponding to the vertical coordinate axis; Calculate the difference between the widths of the roof monitoring points of the different tracks. If the difference is less than or equal to a preset difference threshold, use the point corresponding to the coordinates of the two-dimensional center point in the InSAR point cloud as the target point, calculate the average value of the height values of the neighboring points near the target point, and use the average value as the vertical coordinate of the corresponding point. Use the horizontal coordinate of the target point as the horizontal coordinate of the corresponding point to obtain the corresponding point.

7. The registration method according to claim 1, characterized in that: Matching the corresponding points to obtain corresponding point pairs of buildings on different tracks includes: For the coordinates of each corresponding point, calculate the distance between corresponding points of different tracks; The two corresponding points whose distance is less than a preset distance threshold are regarded as the corresponding point pair.

8. The registration method according to claim 1, wherein: The InSAR point cloud data is registered based on the ICP algorithm and corresponding point pairs of the building in different orbits, including: Step i, selecting an InSAR point cloud of any track from the InSAR point cloud data as a reference point cloud, using the InSAR point clouds of other tracks except the track as the point clouds to be registered, and using the RANSAC algorithm to obtain corresponding point pairs between the reference point cloud and each point cloud to be registered; Step ii, based on the ICP algorithm and the corresponding point pairs, obtaining the translation parameter R and the rotation parameter T for point cloud registration; Step iii, by calculating the formula Get the registration loss value ∑(R,T); where p n represents the reference point cloud, Q n represents the point cloud to be registered, N represents the number of matching point pairs between the reference point cloud and the point cloud to be registered, ||·|| 2 Indicates calculation of 2 norm; In step iv, if the registration loss value is less than the preset registration loss threshold, a rigid body transformation is performed on the point cloud to be registered according to the translation parameter R and the rotation parameter T in step ii to obtain the registered InSAR point cloud data; otherwise, the process returns to step i.

9. A terminal device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the InSAR point cloud registration method according to any one of claims 1 to 8 is implemented.

Images