CN115456937A - Deformation detection method directly based on TLS high-density point cloud measured value - Google Patents
Deformation detection method directly based on TLS high-density point cloud measured value Download PDFInfo
- Publication number
- CN115456937A CN115456937A CN202210871352.9A CN202210871352A CN115456937A CN 115456937 A CN115456937 A CN 115456937A CN 202210871352 A CN202210871352 A CN 202210871352A CN 115456937 A CN115456937 A CN 115456937A
- Authority
- CN
- China
- Prior art keywords
- point
- voxel
- point cloud
- deformation
- octree
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/16—Measuring arrangements characterised by the use of optical techniques for measuring the deformation in a solid, e.g. optical strain gauge
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/88—Lidar systems specially adapted for specific applications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Networks & Wireless Communication (AREA)
- Electromagnetism (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Image Generation (AREA)
Abstract
The invention relates to the technical field of surveying and mapping science, and discloses a deformation detection method directly based on a TLS high-density point cloud measured value, which comprises the following steps: 1) Performing point cloud voxelization segmentation based on an octree structure; 2) Registering the point cloud data of the front stage and the back stage, and 3) comparing the deformation of the point cloud voxel of the back stage with the deformation of the point cloud voxel of the front stage; each point in the later point cloud is positioned to the space in the voxel of the earlier point cloud according to the octree index, and the voxel of the point or the voxel closest to the point is contained; and calculating the distance between each point and each corresponding voxel to obtain the deformation of the corresponding point. The invention can comprehensively position the deformation part and count the deformation. The problem that the existing application range is limited by the distribution of target points or feature points is solved, and the image feature points can be effectively extracted in a smooth area with insignificant texture change.
Description
Technical Field
The invention relates to the technical field of surveying and mapping science, in particular to a deformation detection method directly based on a TLS high-density point cloud measured value.
Background
At present, large military and civil equipment/facilities such as antennas, oil tanks, bridges, tunnels and the like can gradually generate certain deformation in long-time use, generally can reach the magnitude of mm, the deformation size determines whether the safety of the equipment reaches the standard, and the method has important research significance for researching the deformation detection method of the large equipment. The deformation detection is generally generated at a local part of the tested equipment, and the deformation positioning and deformation amount detection are required to be carried out on the tested equipment regularly. The invention aims at the problem of non-contact deformation detection of mm-level or above of large equipment/facilities.
The core of the distortion detection problem is the comparison problem, i.e. the change between the measured value and the design value, or the change between the measured value at the previous stage and the measured value at the next stage.
According to a curved surface buckling deformation detection method based on point cloud characteristic comparison [ J ]. Zhejiang university academic edition, 2021,55 (01): 81-88 ] ] in literature [ Chenyang wave, yi Guomao, zhang Tree has ], aiming at the problem of curved surface buckling deformation, a deformation detection method based on point cloud characteristic comparison is provided, the method adopts the characteristic point position comparison of actual measurement point cloud and template point cloud space position to position a deformation part, and the template point cloud needs to be generated based on a three-dimensional model of equipment to be detected; the method comprises the following steps that (1) a document [ Yan Tianhao ] vertical rescue well wall deformation detection and three-dimensional modeling [ D ] Changan university, 2020 ] based on machine vision carries out three-dimensional reconstruction on a target area by adopting an image feature point extraction and matching mode, a target area section is constructed, and then a deformation part is detected according to the difference between an actual measurement value and a design value; the method comprises the following steps of generating dense three-dimensional point cloud of a detected target based on a stereoscopic vision imaging method, and detecting deformation of a target area based on local flatness of the point cloud.
According to the literature [ Yangqiao, lasting tiger, laser sensor-based intelligent detection system [ J ] laser journal 2020,41 (08) ] aiming at the problem of structural deformation of large buildings, a cooperative optical fiber is deployed at the deformation part, the displacement conditions of each point along the optical fiber are obtained through a plurality of laser sensors, and the deformation is detected; the method comprises the following steps of measuring point position change of a target part in a way of a cooperative target and monitoring deformation condition of the target part in a document [ quality of pipe and quality of large-size object deformation detection [ D ] based on binocular stereo vision, university of Western Ann Mills, 2018 ]; the method comprises the steps of obtaining three-dimensional point cloud of a target area based on an image matching theory and further detecting deformation through two-stage data feature point matching in a document [ Zhaojiaxing, image-based bridge deformation detection method research [ D ]. Changan university, 2020 ].
To summarize the existing correlation methods, they can be mainly classified as: 1) Based on the detection of the measured value of the target region and the known model (or local geometric regularity, such as characteristics of plane or circle), the method needs to know the three-dimensional model or geometric characteristics of the target region in advance, and can only detect the relative deformation of the target region, so the application range is limited; 2) The method detects the point location movement of the target area based on the cooperation target point or the extracted image feature point, and the method measures the deformation amount of the surface point location of the partial area of the target in a targeted manner, is easily limited to the distribution of the target point or the feature point, and is difficult to effectively extract the image feature point in a smooth area with insignificant texture change.
A ground three-dimensional Laser Scanning measuring System (TLS); the laser imaging radar can also be called a three-dimensional laser imaging radar, and Light Detect and Ranging, liDAR adopts an active polar coordinate laser imaging principle, senses the distance from a target to a laser emission point according to the laser emission and receiving time or the Light wave phase difference, realizes the measurement of the three-dimensional coordinates of the target point according to laser Ranging and circle angle measurement, can obtain the high-density three-dimensional point coordinates of the surface of a detected scene object in a covering manner, has high measurement precision, and can reach the point position measurement precision of +/-2 mm within the range of 100 meters in typical TLS.
Disclosure of Invention
The invention provides a deformation detection method directly based on TLS high-density point cloud measured values, aiming at the problems that the existing deformation detection method is narrow in applicability or limited in effectiveness and the like.
In order to achieve the purpose, the invention adopts the following technical scheme:
a deformation detection method directly based on TLS high-density point cloud measured values comprises the following steps:
1) Point cloud voxelization segmentation based on an octree structure, firstly, segmenting a previous-stage point cloud into a voxel set, wherein pixels in an image are square blocks, one pixel corresponds to a block of area of a measured space, and the voxel is a cubic block of a three-dimensional space; a voxel space range comprises a plurality of three-dimensional coordinate points, the point cloud represents a three-dimensional space object by using the three-dimensional points, the voxel represents by using a cube block, information statistics is carried out according to a point set in the voxel in the process of generating the voxel, and information with more dimensions is given to the voxel: the method comprises the steps of normal direction, curvature, principal components, dimension characteristic descriptors and point density, wherein after the voxel structure is generated, the point distribution in each voxel is in a consistent coplane, namely, the superpixel;
2) The registration of the point cloud data of the two stages before and after, and the point cloud comparison premise is that the two groups of point clouds (the data of the two stages before and after) are registered to a unified coordinate system, namely the two groups of point clouds are aligned. To realize two-site cloud data P = { P = 1 ,p 2 ,...,p n And Q = { Q = } 1 ,q 2 ,...,q n Unification of coordinate systems, i.e. calculation formula:
rotation matrix R in (1) PQ And translation vector T PQ Center point p (x) p ,y p ,z p ) And q (x) q ,y q ,z q ) Is a pair of homologous points: solving the equation set by adopting 4 pairs of common points, wherein R in the equation set PQ Expressed by a Reed-Reed matrix and solved by a singular value decomposition method;
registration of the point cloud needs to go through a process from coarse to fine, i.e. coarse registration → fine registration; in the coarse registration stage, a simpler mode is adopted, namely more than 4 groups of public point pairs are manually selected, and coordinate system conversion parameters between two groups of point clouds are calculated based on the public point pairs: r PQ And T PQ ;
3) Performing deformation comparison on the later point cloud and the earlier point cloud voxels to realize the premise of deformation comparison of the two-period data, namely determining the corresponding relation between data points in the two-period data;
establishing voxels by adopting an octree structure for earlier-stage point clouds, and directly searching the corresponding relation between each point in later-stage point clouds and the earlier-stage voxels on the basis of octree indexes: for a point P = (x, y, z) in the later point cloud, the distribution space contains the voxel LN of that point P And leaf node positions are determined according to the point coordinates and the octree coding mode, the deepest layer number of the octree subdivision is set to be n, and the coding codes of the n layers of leaf nodes where the P points are located are calculated P Then LN is performed according to the code P Addressed by layer; if LN P At m (m)<n) layer, the node obtained when the m layer is addressed is voxel LN P ;
Provided with LN P In the brother nodes, the same upper node is divided into 8 child nodes, and the three-axis direction numbers in the nodes which are brother nodes are (i) nx ,i ny ,i nz ),i n ∈[1,2]Then LN P The total number in the three-axis direction is:
and is provided with
Let n layers of voxels have a side length w o The relationship between the total number in the three-axis direction and the P point is LN P
Wherein floor () is a rounding function;
LN P the three-axis total number of the neighborhood voxels of
According to the formula, the index of the neighborhood voxels can be realized;
each point in the later point cloud is positioned in the space in the voxel of the earlier point cloud according to the octree index, and the voxel containing the point or the voxel closest to the point is contained; and calculating the distance between each point and each corresponding voxel to obtain the deformation amount of the corresponding point position.
A deformation detection method based on TLS high density point cloud measured value directly, the point cloud voxel segmentation adopts the point cloud plane extraction of octree voxel growth, according to the point cloud density and thickness distribution, the approximate plane of local voxel internal flatness of the voxel subdivision is used as subdivision termination index: adopting a fixed voxel size to divide the point cloud into a series of uniform octree voxels, and adjusting the number of initial subdivision layers according to the voxel average point density so as to enable an octree structure to be suitable for the density distribution of the point cloud; then calculating and counting the information of each voxel to obtain a voxel set which is distributed most similar to a plane, and further counting to obtain a subdivision termination condition of an octree structure; and finally, obtaining an octree voxel set adaptive to the density and thickness distribution of the point cloud through recursive subdivision.
Due to the adoption of the technical scheme, the invention has the following advantages:
according to the method, high-density three-dimensional point cloud data of the surface of the target equipment to be measured are obtained by TLS, and then the point cloud data in the front period and the point cloud data in the back period are directly compared to comprehensively locate a deformation part and count the deformation. The point cloud data can be described as three-dimensional coordinate data points of the visual surface of the measured scene which are distributed in a scattered manner, and the point cloud can be simply understood as pixel point coordinates containing depth (distance) information relative to the image.
The deformation detection method provided by the invention is not limited by the shape and texture of the target object, and can directly and completely acquire deformation detection information such as a deformation area, a deformation amount and the like of the target global surface by adopting TLS contactless measurement two-stage point cloud data.
Drawings
FIG. 1 is a cloud point diagram of a large oil tank facility;
FIG. 2 is a color differentiation map of a voxel set;
FIG. 3 is a schematic diagram of two-phase point cloud coarse registration based on 4 sets of common points;
FIG. 4 is a diagram of two-stage point cloud, and a noise map is artificially added to the later-stage point cloud;
FIG. 5 is a post-registration image of a two-phase point cloud;
FIG. 6 is a diagram of deformation detection effect.
Detailed Description
As shown in fig. 1, 2, 3, 4, 5, and 6, a deformation detection method directly based on the measured value of the TLS high-density point cloud includes the following steps:
1) And point cloud voxelization segmentation based on an octree structure, wherein pixels in the image are squares, one pixel corresponds to a block of area of a measured space, and the voxel is a cube block of a three-dimensional space. A voxel space range comprises a plurality of three-dimensional coordinate points, the point cloud represents a three-dimensional space object by the three-dimensional points, and the voxel is represented by a cube block. In the process of generating the voxels, information statistics is carried out according to the point sets in the voxels, more dimensional information (including normal direction, curvature, principal components, dimension feature descriptors, point density and the like) can be given to the voxels, and after the voxel structure is generated, the point distribution in each voxel has consistency (such as coplanarity), namely the hyper-voxels (still called the voxels for short).
The invention firstly segments the previous point cloud into a voxel set, the reference document of the point cloud voxel segmentation [ extracting the point cloud plane by octree voxel growth, optical precision engineering, 2018,26 (1): 172-183 ], adopts octree structure, and takes the local (in-voxel) flatness (plane approximation) of the voxel subdivision as the subdivision termination index according to the density and thickness distribution of the point cloud: adopting a fixed voxel size to divide the point cloud into a series of uniform octree voxels, and adjusting the number of initial subdivision layers according to the voxel average point density so as to enable an octree structure to be suitable for the density distribution of the point cloud; then, calculating and counting the information of each voxel to obtain a voxel set which is distributed most like a plane, and further counting to obtain a subdivision termination condition of an octree structure; and finally, obtaining an octree voxel set adaptive to the density and thickness distribution of the point cloud through recursive subdivision. As shown in fig. 1 and 2.
2) The registration of the point cloud data of the two stages before and after, and the premise of point cloud comparison is that two groups of point clouds (the data of the two stages before and after) are registered to a unified coordinate system, namely the two groups of point clouds are aligned. To realize two-site cloud data P = { P = 1 ,p 2 ,...,p n And Q = { Q = } 1 ,q 2 ,...,q n Unification of the coordinate systems, i.e. the formula
Rotation matrix R in (1) PQ And translation vector T PQ Center point p (x) p ,y p ,z p ) And q (x) q ,y q ,z q ) Is a pair of homologous points. Solving the equation set by adopting 4 pairs of common points, wherein R in the equation set PQ Expressed by a Rodrigue matrix and solved by a singular value decomposition method [ Yanfan, liguangyun, wangli, king force ] three-dimensional coordinate transformation method research [ J]Mapping notification 2010,0 (6): 5-7.]。
Registration (registration) of point clouds typically needs to go through a coarse-to-fine process, i.e., coarse registration → fine registration. In the coarse registration stage, a simpler mode is adopted, namely more than 4 groups of public point pairs are manually selected, and coordinate system conversion parameters (R) between two groups of point clouds are calculated based on the public point pairs PQ And T PQ ). As shown in fig. 3.
The fine Registration is achieved using the Iterative Closed Point (ICP) correlation algorithm proposed in the literature [ Besl, P.J. and McKay, N.D.A. Method for Registration of 3-D Shapes [ J ]. Transactions on Pattern Analysis and Machine understanding, 1992,14 (2): 239-256 ]. The effects before and after point cloud registration are shown in fig. 4-5.
3) And comparing the deformation of the later point cloud with the deformation of the earlier point cloud voxel on the premise of realizing the deformation comparison of the two-period data, namely determining the corresponding relationship between the data points in the two-period data. Because the voxels are established by adopting an octree structure for the earlier-stage point cloud, the corresponding relation between each point in the later-stage point cloud and the earlier-stage voxels is directly found based on octree indexes: for a point P = (x, y, z) in the later point cloud, the distribution space contains the voxel (LN) of that point P Leaf node) position is determined according to the point coordinate and the octree coding mode, the deepest layer number of the octree subdivision is set as n, and the coding code of the n layers of leaf nodes where the P point is located is calculated P Then LN is performed according to the code P Is addressed by layer, if LN P At m (m)<n) layers, the node obtained when the m layer is addressed is the voxel LN P 。
Let LN P The three-axis direction numbers in the sibling nodes (the same upper node is divided into 8 child nodes which are sibling nodes with each other) are respectively (i) nx ,i ny ,i nz ),i n ∈[1,2]Then the total number of LNP in the three-axis direction is
And is
Let n layers of voxels have a side length of w o The relationship between the total number in the three-axis direction and the P point is LN P
Where floor () is a rounding function. LN P Has a total three-axis number of neighborhood voxels of
According to the formula, the index of the neighborhood voxel can be realized.
Each point in the later point cloud can be positioned to a voxel containing the point in the former point cloud voxel or a voxel closest to the point in space according to the octree index. And calculating the distance between each point and each corresponding voxel to obtain the deformation amount of the corresponding point position.
Claims (2)
1. A deformation detection method directly based on a TLS high-density point cloud measured value is characterized by comprising the following steps: the method comprises the following steps:
1) Performing point cloud voxelization segmentation based on an octree structure, firstly segmenting a previous-stage point cloud into a voxel set, wherein pixels in the image are square blocks, one pixel corresponds to an area of a measured space, and the voxel is a cubic block of a three-dimensional space; a voxel space range comprises a plurality of three-dimensional coordinate points, the point cloud represents a three-dimensional space object by using the three-dimensional points, the voxel represents by using a cube block, information statistics is carried out according to a point set in the voxel in the process of generating the voxel, and information with more dimensions is given to the voxel: the method comprises the steps of normal direction, curvature, principal component, dimension characteristic descriptor and point density, wherein after the generation of a voxel structure is finished, the distribution of points in each voxel has a consistent coplane, namely, the superpixel;
2) The registration of the point cloud data in the two stages before and after, and the point cloud comparison premise is that the two groups of point clouds (the point cloud data in the two stages before and after) are registered to a unified coordinate system, namely the two groups of point clouds are aligned, so as to realize the two-station point cloud data P = { P = 1 ,p 2 ,...,p n } and Q = { Q = 1 ,q 2 ,...,q n Unification of coordinate systems, i.e. calculation formula:
rotation matrix R in (1) pQ And translation vector T pQ Center point p (x) p ,y p ,z p ) And q (x) q ,y q ,z q ) Is a pair of homologous points: solving the equation set by adopting 4 pairs of common points, wherein R in the equation set pQ Using a Rodrigue matrix to represent and using a singular value decomposition method to solve;
registration of the point cloud needs to go through a process from coarse to fine, namely coarse registration → fine registration; in the coarse registration stage, a simpler mode is adopted, namely more than 4 groups of common point pairs are manually selected, and coordinate system conversion parameters between two groups of point clouds are calculated based on the common point pairs: r pQ And T pQ ;
3) Performing deformation comparison on the later point cloud and the earlier point cloud voxels to realize the premise of deformation comparison of the two-period data, namely determining the corresponding relation between data points in the two-period data;
establishing voxels by adopting an octree structure for earlier-stage point clouds, and directly searching the corresponding relation between each point in later-stage point clouds and the earlier-stage voxels on the basis of octree indexes: for a point P = (x, y, z) in the later point cloud, the distribution space contains the voxel LN of that point P And leaf node positions are determined according to the point coordinates and the octree coding mode, the deepest layer number of the octree subdivision is set to be n, and the coding codes of the n layers of leaf nodes where the P points are located are calculated P Then LN is performed according to the code P Addressed by layer; if LN P At m (m)<n) layer, the node obtained when the m layer is addressed is voxel LN P ;
Let LN P In the brother nodes, the same upper node is divided into 8 child nodes, and the three-axis direction numbers in the brother nodes are (i) nx ,i ny ,i nz ),i n ∈[1,2]Then LN P The total number in the three-axis direction is:
and is
Let n layers of voxels have a side length w o The relationship between the total number in the three-axis direction and the P point is LN P
Wherein floor () is a rounding function;
LN P has a total three-axis number of neighborhood voxels of
According to the formula, the index of the neighborhood voxels can be realized;
each point in the later point cloud is positioned in the space in the voxel of the earlier point cloud according to the octree index, and the voxel containing the point or the voxel closest to the point is contained; and calculating the distance between each point and each corresponding voxel to obtain the deformation of the corresponding point.
2. The method as claimed in claim 1, wherein the method comprises the following steps: the point cloud voxel segmentation adopts point cloud plane extraction of octree voxel growth, and an approximate plane of local voxel internal flatness of voxel subdivision is used as a subdivision termination index according to point cloud density and thickness distribution: the method comprises the following steps of initially dividing a point cloud into a series of uniform octree voxels by adopting a fixed voxel size, and adjusting the number of initial dividing layers according to the average point density of voxels so as to enable an octree structure to be suitable for density distribution of the point cloud; then calculating and counting the information of each voxel to obtain a voxel set which is distributed most similar to a plane, and further counting to obtain a subdivision termination condition of an octree structure; and finally, obtaining an octree voxel set adaptive to the density and thickness distribution of the point cloud through recursive subdivision.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210871352.9A CN115456937A (en) | 2022-07-22 | 2022-07-22 | Deformation detection method directly based on TLS high-density point cloud measured value |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210871352.9A CN115456937A (en) | 2022-07-22 | 2022-07-22 | Deformation detection method directly based on TLS high-density point cloud measured value |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115456937A true CN115456937A (en) | 2022-12-09 |
Family
ID=84297111
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210871352.9A Pending CN115456937A (en) | 2022-07-22 | 2022-07-22 | Deformation detection method directly based on TLS high-density point cloud measured value |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115456937A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117433440A (en) * | 2023-10-24 | 2024-01-23 | 中南大学 | Roadway deformation real-time detection method and device based on laser point cloud and storage medium |
-
2022
- 2022-07-22 CN CN202210871352.9A patent/CN115456937A/en active Pending
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117433440A (en) * | 2023-10-24 | 2024-01-23 | 中南大学 | Roadway deformation real-time detection method and device based on laser point cloud and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110058237B (en) | InSAR point cloud fusion and three-dimensional deformation monitoring method for high-resolution SAR image | |
CN109903327B (en) | Target size measurement method of sparse point cloud | |
CN110443836A (en) | A kind of point cloud data autoegistration method and device based on plane characteristic | |
Wu et al. | A triangulation-based hierarchical image matching method for wide-baseline images | |
CN110006408B (en) | LiDAR data cloud control aerial image photogrammetry method | |
CN109241978B (en) | Method for rapidly extracting plane piece in foundation three-dimensional laser point cloud | |
CN105300316A (en) | Light stripe center rapid extraction method based on gray centroid method | |
CN109900338B (en) | Method and device for measuring volume of pavement pit | |
CN111553292A (en) | Rock mass structural plane identification and occurrence classification method based on point cloud data | |
CN109472802A (en) | A kind of surface grid model construction method constrained certainly based on edge feature | |
CN112164145B (en) | Method for rapidly extracting indoor three-dimensional line segment structure based on point cloud data | |
CN108876861B (en) | Stereo matching method for extraterrestrial celestial body patrolling device | |
Ye et al. | Integrated image matching and segmentation for 3D surface reconstruction in urban areas | |
CN113887624A (en) | Improved feature stereo matching method based on binocular vision | |
CN115456937A (en) | Deformation detection method directly based on TLS high-density point cloud measured value | |
CN115909025A (en) | Terrain vision autonomous detection and identification method for small celestial body surface sampling point | |
CN106709432B (en) | Human head detection counting method based on binocular stereo vision | |
Parmehr et al. | Automatic registration of optical imagery with 3d lidar data using local combined mutual information | |
CN107765257A (en) | A kind of laser acquisition and measuring method based on the calibration of reflected intensity accessory external | |
CN116452764A (en) | Airborne laser radar point cloud hierarchical rarefaction method integrating terrain information | |
CN114067064A (en) | Target three-dimensional reconstruction method based on multi-view radar image | |
Su et al. | An Accurate Extraction Algorithm of the Indoor Boundary Features Based on Point Cloud Data | |
Xu et al. | Point cloud segmentation of gully based on characteristic difference using airborne lidar data | |
Yan et al. | A hierarchical image matching method for stereo satellite imagery | |
He et al. | An algorithm of reconstructing LiDAR building models based on isoheight |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |