CN112861674B - Point cloud optimization method based on ground characteristics and computer readable storage medium - Google Patents
Point cloud optimization method based on ground characteristics and computer readable storage medium Download PDFInfo
- Publication number
- CN112861674B CN112861674B CN202110115668.0A CN202110115668A CN112861674B CN 112861674 B CN112861674 B CN 112861674B CN 202110115668 A CN202110115668 A CN 202110115668A CN 112861674 B CN112861674 B CN 112861674B
- Authority
- CN
- China
- Prior art keywords
- point cloud
- matrix
- ground
- optimization method
- coordinate
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 26
- 238000005457 optimization Methods 0.000 title claims abstract description 22
- 238000003860 storage Methods 0.000 title claims abstract description 9
- 239000011159 matrix material Substances 0.000 claims abstract description 88
- 239000013598 vector Substances 0.000 claims abstract description 17
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 7
- 238000004590 computer program Methods 0.000 claims description 3
- 238000010276 construction Methods 0.000 abstract description 2
- 238000013507 mapping Methods 0.000 description 4
- 238000006243 chemical reaction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
- G06T17/05—Geographic models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
- G06T17/20—Finite element generation, e.g. wire-frame surface description, tesselation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4038—Image mosaicing, e.g. composing plane images from plane sub-images
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Geometry (AREA)
- Data Mining & Analysis (AREA)
- Computer Graphics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Multimedia (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Remote Sensing (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Radar Systems Or Details Thereof (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a point cloud optimization method based on ground characteristics and a computer-readable storage medium, wherein the point cloud optimization method comprises the following steps: constructing a coordinate matrix; constructing an average three-dimensional coordinate vector; obtaining a first reference matrix according to the coordinate matrix and the average three-dimensional coordinate vector; multiplying the first reference matrix with a transposed matrix thereof to obtain a second reference matrix, and performing singular value decomposition on the second reference matrix to obtain a right singular matrix; and constructing a third reference matrix according to the right singular matrix and processing the data in the first point cloud set to obtain a second point cloud set. The point cloud optimization method corrects the height error of the laser point cloud map, improves the accuracy of the point cloud, and further reduces the map construction error of the high-precision map.
Description
Technical Field
The invention relates to the technical field of unmanned aerial vehicle, in particular to a point cloud optimization method based on ground characteristics and a computer readable storage medium.
Background
In the unmanned field, high-precision maps play an extremely important role, because unmanned vehicles can be operated normally only with reference to the element semantics in the high-precision maps. The high-precision map is a map with rich semantics, and the map with high precision. At present, the high-precision map is not well defined in the industry, so that the manufacturing modes of the high-precision map are different from company to company. Most of the high-precision maps are manufactured by fusing various sensor data, such as multi-line laser radar, cameras, imu and the like, and some of the high-precision maps are manufactured by only adopting partial sensor data. The patent is mainly aimed at optimizing and improving the part of the multi-line laser radar for making the high-precision map.
The idea of multi-line laser radar mapping is mainly to generate a point cloud map by feature registration among radar data frames and then splicing point cloud data of all frames together. First, the laser radar data itself has a sensor error at present; second, there is an error in the registration of the point cloud keyframes, and the accumulation of the errors can also cause the deviation between the final point cloud map and the actual scene, especially the longer road surface splice, which may generate a larger ground elevation difference. The accumulated elevation difference makes the originally flat road surface look like a slope, which is different from the actually collected scene.
Disclosure of Invention
In view of the above, the present invention provides a point cloud optimization method and a computer readable storage medium based on ground characteristics, which can improve the accuracy of a point cloud map. The technical proposal is as follows:
In one aspect, the invention provides a point cloud optimization method based on ground characteristics, which comprises the following steps:
s1, collecting coordinate data of a plurality of groups of ground points, wherein the plurality of groups of coordinate data form a first point cloud set;
S2, constructing a coordinate matrix by using three-dimensional coordinates of all the ground points, calculating average three-dimensional coordinates according to the three-dimensional coordinates of all the ground points to obtain average three-dimensional coordinate vectors, and respectively subtracting the average three-dimensional coordinate vectors from sub-vectors corresponding to each ground point in the coordinate matrix to obtain a first reference matrix;
S3, multiplying the first reference matrix with a transposed matrix thereof to obtain a second reference matrix, and performing singular value decomposition on the second reference matrix to obtain a right singular matrix;
s4, replacing the upper left corner of the fourth-order identity matrix with the inverse matrix of the right singular matrix to obtain a third reference matrix, and respectively multiplying the plurality of bit gesture matrixes in the first point cloud set by the third reference matrix to obtain a plurality of correction bit gesture matrixes which form a second point cloud set.
Further, in step S1, the collecting coordinate data of the plurality of sets of ground points includes: and acquiring a plurality of frames of ground images, marking coordinate data of ground points in the plurality of frames of ground images, wherein at least two frames of ground images are partially or completely overlapped in the plurality of frames of ground images.
Further, the multi-frame ground images are collected at different times.
Further, the bit gesture matrix is a fourth-order square matrix, and each bit gesture matrix comprises three-dimensional coordinate data and azimuth angle data of the corresponding ground point.
Further, the coordinate data are acquired by a laser radar.
Further, the point cloud optimization method further comprises the following steps:
s5, splicing the data in the second point cloud set to obtain a point cloud map.
In another aspect, the present invention provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of a point cloud optimization method as described above.
The invention has the following technical effects:
the data in the point cloud set is optimized, so that the spliced point cloud data more accords with an actual scene, the height error of the laser point cloud map is corrected to a certain extent, the accuracy of the point cloud map is improved, and the map building error of the high-precision map is further reduced.
Drawings
FIG. 1 is a schematic diagram of a point cloud optimization method according to an embodiment of the present invention;
Detailed Description
The technical scheme of the invention is further described below with reference to the attached drawings and specific embodiments. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
In one embodiment of the present invention, the present invention provides a point cloud optimization method based on ground characteristics, the point cloud optimization method comprising the steps of:
s1, collecting coordinate data of a plurality of groups of ground points, wherein the plurality of groups of coordinate data form a first point cloud set;
The coordinate data may be acquired by a lidar, which may acquire images while being displaced, so that the acquired images differ in time and may or may not include timing information in accordance with a certain timing.
S2, constructing a coordinate matrix by using three-dimensional coordinates of all points, calculating average three-dimensional coordinates according to the three-dimensional coordinates of all points to obtain an average three-dimensional coordinate vector, and respectively subtracting the average three-dimensional coordinate vector from a sub-vector corresponding to each ground point in the coordinate matrix to obtain a first reference matrix;
The coordinate matrix may be set as follows: each column vector is a three-dimensional coordinate of a ground point, N ground points are needed to finish splicing when a point cloud map is constructed, and a coordinate matrix is 3*N; the column vectors in the first reference matrix are subtracted from the average three-dimensional coordinate vector in the coordinate matrix, and the first reference matrix is also a 3*3 matrix.
S3, multiplying the first reference matrix with a transposed matrix thereof to obtain a second reference matrix, and performing singular value decomposition on the second reference matrix to obtain a right singular matrix;
The transposed matrix of the first reference matrix is also a matrix, so that the second reference matrix is also a 3*3 matrix, and correspondingly, the singular value decomposition is performed on the second reference matrix to obtain three matrices, namely a left singular matrix, a characteristic value matrix and a right singular matrix, which are all 3*3 matrices. Since the second reference matrix is a square matrix, it is the eigenvalue decomposition that is actually performed on the second reference matrix.
S4, replacing the upper left corner of the fourth-order identity matrix with the inverse matrix of the right singular matrix to obtain a third reference matrix, and respectively multiplying the plurality of bit gesture matrixes in the first point cloud set by the third reference matrix to obtain a plurality of correction bit gesture matrixes which form a second point cloud set.
The third reference matrix is 4*4 matrix, and correspondingly, the bit gesture matrix is 4*M matrix, and the size of M is determined according to the information protected by the bit gesture matrix. In one embodiment of the present invention, the bit gesture matrix is a fourth-order square matrix, and each bit gesture matrix includes three-dimensional coordinate data and azimuth data of a corresponding ground point. In practice, in step S4, a fourth order third reference matrix is constructed, which is because the bit pose matrix has four column vectors, and thus the third reference matrix must have four row vectors when multiplied by the bit pose matrix.
Through the steps, the second point cloud set is obtained from the first point cloud set, and the data of the second point cloud set is more excellent than the first point cloud set in terms of eliminating errors.
In one embodiment of the present invention, the point cloud optimization method further includes the steps of:
s5, splicing the data in the second point cloud set to obtain a point cloud map.
The term "ground point" in the present invention does not refer to a specific point on the ground, and because the radar acquired data itself has a certain error, the points acquired in the preset range are all considered as ground points, i.e. points near the ground level are all acquired and processed, so as to obtain a more accurate point cloud map later.
In step S1, the collecting coordinate data of a plurality of groups of ground points includes: and acquiring a plurality of frames of ground images, marking coordinate data of ground points in the plurality of frames of ground images, wherein at least two frames of ground images are partially or completely overlapped in the plurality of frames of ground images. Therefore, by identifying and matching the characteristic points of the overlapping part, a plurality of images can be spliced together to obtain the point cloud map.
The point cloud map reflects the ground conditions, and for the unmanned system, the driving strategy can be determined according to the point cloud map. In other words, the data in the first point cloud set are enough to be spliced to obtain a point cloud map; however, since the data in the first point cloud set is directly acquired and only subjected to simple preliminary processing, some errors exist, and when the data in the first point cloud set is applied to splice point cloud maps, when images of different frames are spliced together, errors in height and angle can be generated, so that a height difference and a gradient which are not present in reality appear in the point cloud maps.
At present, no matter which mapping method is used, the errors of the sensor and the errors of the solutions cannot be eliminated, and the optimization can be improved only through an effective method, so that the mapping result is more in line with the actual scene. The data in the second point cloud set is processed to eliminate errors, and the spliced point cloud map correspondingly has no false height difference and false gradient. That is, from the normalization thought, the point sets collected near the ground height are considered to belong to the ground data, and then the point clouds near the ground are extracted and processed, so that an accurate ground characteristic and a conversion matrix can be obtained, and the pose information of the point clouds of each key frame is updated according to the ground characteristic, so that the effect of optimizing the point clouds is achieved; the point cloud pavement information is smoother and more similar to a real scene, and the error of registration calculation is greatly improved.
It should be noted that the invention is only suitable for multi-line laser mapping of low-speed and road leveling scenes, and is not suitable for scenes with continuous and obvious fluctuation on the ground.
FIG. 1 represents an embodiment of the present invention in which the term "constructing a feature matrix of a point cloud" refers to obtaining a second reference matrix; the SVD decomposition to obtain the ground direction and the conversion matrix is equivalent to the steps S3 and S4 to finally obtain the corrected bit posture matrix; in this example, the "update point cloud posture" corresponds to replacing the first point cloud with the second point cloud.
In one embodiment of the invention, the invention provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of a point cloud optimization method as described above. The storage medium may be a usb disk, a hard disk, etc.
The point cloud optimization method effectively solves the problem that the point cloud data and the actual scene have elevation deviation in the multi-line laser radar image construction process.
The foregoing description is only of the preferred embodiments of the present invention and is not intended to limit the scope of the claims, and all equivalent structures or equivalent processes using the descriptions and drawings of the present invention are directly or indirectly applied to other related technical fields, which are also included in the scope of the present invention.
Claims (6)
1. The point cloud optimization method based on the ground characteristics is characterized by comprising the following steps of:
s1, collecting coordinate data of a plurality of groups of ground points, wherein the plurality of groups of coordinate data form a first point cloud set;
S2, constructing a coordinate matrix by using three-dimensional coordinates of all the ground points, calculating average three-dimensional coordinates according to the three-dimensional coordinates of all the ground points to obtain average three-dimensional coordinate vectors, and respectively subtracting the average three-dimensional coordinate vectors from sub-vectors corresponding to each ground point in the coordinate matrix to obtain a first reference matrix, wherein the sub-vectors corresponding to the ground points are the three-dimensional coordinates of the ground points;
S3, multiplying the first reference matrix with a transposed matrix thereof to obtain a second reference matrix, and performing singular value decomposition on the second reference matrix to obtain a right singular matrix;
S4, replacing the upper left corner of the fourth-order identity matrix with an inverse matrix of the right singular matrix to obtain a third reference matrix, and respectively multiplying a plurality of bit gesture matrixes in the first point cloud set by the third reference matrix to obtain a plurality of correction bit gesture matrixes which form a second point cloud set;
s5, splicing the data in the second point cloud set to obtain a point cloud map.
2. The method of optimizing a point cloud as claimed in claim 1, wherein in step S1, said collecting coordinate data of a plurality of sets of ground points includes: and acquiring a plurality of frames of ground images, marking coordinate data of ground points in the plurality of frames of ground images, wherein at least two frames of ground images are partially or completely overlapped in the plurality of frames of ground images.
3. The method of point cloud optimization as claimed in claim 2, wherein said plurality of frames of ground images are acquired at different times.
4. The point cloud optimization method of claim 1, wherein the bit pose matrices are fourth-order square matrices, and each bit pose matrix includes three-dimensional coordinate data and azimuth data of a corresponding ground point.
5. The point cloud optimization method of claim 1, wherein the coordinate data is acquired by a lidar.
6. A non-transitory computer readable storage medium, having stored thereon a computer program which, when executed by a processor, implements the steps of the point cloud optimization method according to any of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110115668.0A CN112861674B (en) | 2021-01-28 | 2021-01-28 | Point cloud optimization method based on ground characteristics and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110115668.0A CN112861674B (en) | 2021-01-28 | 2021-01-28 | Point cloud optimization method based on ground characteristics and computer readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112861674A CN112861674A (en) | 2021-05-28 |
| CN112861674B true CN112861674B (en) | 2024-09-06 |
Family
ID=75986896
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110115668.0A Active CN112861674B (en) | 2021-01-28 | 2021-01-28 | Point cloud optimization method based on ground characteristics and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112861674B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105678833A (en) * | 2016-01-11 | 2016-06-15 | 塔里木大学 | Point cloud geometrical data automatic splicing algorithm based on multi-view image three-dimensional modeling |
| CN105913489A (en) * | 2016-04-19 | 2016-08-31 | 东北大学 | Indoor three-dimensional scene reconstruction method employing plane characteristics |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107403456B (en) * | 2017-07-28 | 2019-06-18 | 北京大学深圳研究生院 | A kind of point cloud genera compression method based on KD tree and optimization figure transformation |
| CN108665491B (en) * | 2018-03-22 | 2022-04-12 | 西安电子科技大学 | Rapid point cloud registration method based on local reference points |
| US10694103B2 (en) * | 2018-04-24 | 2020-06-23 | Industrial Technology Research Institute | Building system and building method for panorama point cloud |
| CN109859256B (en) * | 2019-03-13 | 2023-03-31 | 大连理工大学 | Three-dimensional point cloud registration method based on automatic corresponding point matching |
| CN110363838B (en) * | 2019-06-06 | 2020-12-15 | 浙江大学 | Large-visual-field image three-dimensional reconstruction optimization method based on multi-spherical-surface camera model |
| CN110728751A (en) * | 2019-06-19 | 2020-01-24 | 武汉科技大学 | Construction method of indoor 3D point cloud semantic map |
| CN110827403B (en) * | 2019-11-04 | 2023-08-25 | 北京易控智驾科技有限公司 | Construction method and device of mine three-dimensional point cloud map |
| CN112162561A (en) * | 2020-10-13 | 2021-01-01 | 上海美迪索科电子科技有限公司 | Map construction optimization method, device, medium and equipment |
-
2021
- 2021-01-28 CN CN202110115668.0A patent/CN112861674B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105678833A (en) * | 2016-01-11 | 2016-06-15 | 塔里木大学 | Point cloud geometrical data automatic splicing algorithm based on multi-view image three-dimensional modeling |
| CN105913489A (en) * | 2016-04-19 | 2016-08-31 | 东北大学 | Indoor three-dimensional scene reconstruction method employing plane characteristics |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112861674A (en) | 2021-05-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110853075B (en) | Visual tracking positioning method based on dense point cloud and synthetic view | |
| CN113657224B (en) | Method, device and equipment for determining object state in vehicle-road coordination | |
| CN111862673B (en) | Parking lot vehicle self-positioning and map construction method based on top view | |
| CN109752701A (en) | A kind of road edge detection method based on laser point cloud | |
| DE112018000605T5 (en) | Information processing apparatus, data management apparatus, data management system, method and program | |
| CN112258409B (en) | Method and device for recovering absolute scale of unmanned monocular camera | |
| CN112232275B (en) | Obstacle detection method, system, equipment and storage medium based on binocular recognition | |
| CN112700486B (en) | Method and device for estimating depth of road surface lane line in image | |
| CN113834492A (en) | Map matching method, system, device and readable storage medium | |
| CN113327296B (en) | Laser radar and camera online combined calibration method based on depth weighting | |
| CN112837383A (en) | Camera and laser radar recalibration method and device and computer readable storage medium | |
| CN113156407B (en) | Vehicle-mounted laser radar external parameter joint calibration method, system, medium and device | |
| CN114037762B (en) | Real-time high-precision positioning method based on registration of image and high-precision map | |
| CN115451964B (en) | Ship scene simultaneous mapping and positioning method based on multi-mode mixing characteristics | |
| CN112258582B (en) | Camera attitude calibration method and device based on road scene recognition | |
| KR20230003803A (en) | Automatic calibration through vector matching of the LiDAR coordinate system and the camera coordinate system | |
| CN111862236B (en) | Self-calibration method and system for fixed-focus binocular camera | |
| CN112861674B (en) | Point cloud optimization method based on ground characteristics and computer readable storage medium | |
| CN107330934B (en) | Low-dimensional cluster adjustment calculation method and system | |
| CN112116644B (en) | Obstacle detection method and device based on vision and obstacle distance calculation method and device | |
| CN116358486A (en) | Target ranging method, device and medium based on monocular camera | |
| KR20140029794A (en) | Image stabilization method and system using curve lane model | |
| CN110728684A (en) | Map construction method and device, storage medium and electronic equipment | |
| CN117029870A (en) | Laser odometer based on road surface point cloud | |
| CN114419118A (en) | Three-dimensional point cloud registration method, mobile device and storage medium |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |