CN115561768A - Vehicle-mounted laser radar data precision detection method based on single-point laser point cloud - Google Patents
Vehicle-mounted laser radar data precision detection method based on single-point laser point cloud Download PDFInfo
- Publication number
- CN115561768A CN115561768A CN202211151258.2A CN202211151258A CN115561768A CN 115561768 A CN115561768 A CN 115561768A CN 202211151258 A CN202211151258 A CN 202211151258A CN 115561768 A CN115561768 A CN 115561768A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- point
- vehicle
- laser
- data
- 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
- 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
- G01S17/02—Systems using the reflection of electromagnetic waves other than radio waves
- G01S17/06—Systems determining position data of a target
-
- 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
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
- G01S7/481—Constructional features, e.g. arrangements of optical elements
-
- 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
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
- G01S7/497—Means for monitoring or calibrating
Abstract
The invention provides a vehicle-mounted laser radar data precision detection method based on single-point laser point cloud, which relates to the technical field of laser point cloud. The invention realizes the classification and extraction of point clouds based on semantic features, can realize the automatic extraction and point contact of vector data, and is a vehicle-mounted laser radar data precision detection method with high efficiency and low cost.
Description
Technical Field
The invention relates to the technical field of laser point cloud, in particular to a vehicle-mounted laser radar data precision detection method based on single-point laser point cloud.
Background
With the rapid development of the automatic driving technology, the construction pace of the high-precision map is increasingly accelerated, the point cloud data of the road acquired by the vehicle-mounted laser radar vehicle becomes a main data source for high-precision map production, and the vehicle-mounted laser radar vehicle adopts 360-degree real-time drawing and has the advantages of more intuitively and rapidly showing and analyzing road conditions and accessory facilities of the road. The high-precision map is a centimeter-level precision map, so the precision of the vehicle-mounted laser radar point cloud data is equal to or higher than the precision, however, due to the influences of the driving speed of a collecting vehicle, sunlight interference, opposite interference, point cloud quality, time registration and the like, the vehicle-mounted laser radar point cloud can deviate, and the precision detection of the data is very important for ensuring the precision of the vehicle-mounted laser radar point cloud data.
The laser radar is mainly technically characterized in that the turn-back time of laser meeting an object point is analyzed by transmitting and receiving laser beams, the relative distance between a point location and a vehicle is calculated, the three-dimensional coordinate of the point location is obtained, and the laser radar has extremely high distance resolution and angle resolution. The vehicle-mounted laser radar has the advantages of large scanning range, high speed and low density. The single-point laser radar has long scanning time, high accuracy and high density, can comprehensively acquire the spatial point position of the surface of the measured target in a complex area, and has complementarity with the spatial point position. The data of the vehicle-mounted laser radar point cloud and the data of the single-point laser radar point cloud are used for carrying out registration modes of data space position, homonymous characteristic, classification characteristic extraction and the like, and the point cloud comparison mode based on normal vectors of two pieces of point cloud data is used for obtaining deviation values of homonymous points, so that the aim of correcting the radar laser point cloud precision by using the high-precision single-point laser point cloud is fulfilled.
The high-precision map is used as basic data for decision assistance and unmanned driving, the requirement on precision is high, and in order to guarantee the precision of the high-precision map, quality inspection of point cloud data of the vehicle-mounted laser radar is a key link. Various experts make some researches on the detection aspect of vehicle-mounted laser point cloud data to obtain certain achievements, and the common method comprises the steps of obtaining feature points by adopting a three-coordinate instrument and a total station, comparing the feature points with the feature points of the point cloud data, and calculating errors to obtain the vehicle-mounted laser point cloud data. However, the total station has the problems of slow aging, large workload, low safety, inaccurate point contact and the like when being compared with point cloud.
With the wide application of laser point cloud data, higher requirements are put on the precision and workload of the point cloud data, and therefore a method with high efficiency and low cost is urgently needed to verify the precision of the vehicle-mounted laser radar point cloud data.
The present application was made based on this.
Disclosure of Invention
In order to overcome the defects in the background art, the invention aims to provide a vehicle-mounted laser radar data precision detection method based on single-point laser point cloud with high efficiency and low cost so as to verify the precision of the vehicle-mounted laser radar point cloud data.
In order to achieve the purpose, the invention provides the following technical scheme:
a vehicle-mounted laser radar data precision detection method based on single-point laser point cloud comprises the following steps:
s100, acquiring vehicle-mounted laser point cloud data: acquiring road conditions of a high-precision map to be produced by using a vehicle-mounted laser radar acquisition vehicle to acquire corresponding vehicle-mounted laser point cloud data;
s200, acquiring single-point laser point cloud data: determining fixed coordinates, erecting a base station type three-dimensional laser scanner, and scanning the road condition of vehicle-mounted laser point cloud data needing to be checked and calibrated to obtain corresponding single-point laser point cloud data;
s300, automatic point cloud processing: analyzing mass point clouds in the early stage, and automatically denoising, classifying and simplifying on the basis of analysis data;
s400, automatic feature extraction: materializing the denoised, classified and simplified point cloud data, and extracting element features by combining a random sampling consistency algorithm with robust estimation;
s500, automatic matching calculation: after the characteristics are extracted, the point cloud normal vector is calculated by taking the single-point laser point cloud data as a reference, and the two pieces of point cloud data are compared based on the normal vector;
s600, providing an accuracy report: and according to the comparison result, taking the single-point laser point cloud data as standard data, and issuing a precision report after comparing the vehicle-mounted laser point cloud data with the single-point laser point cloud data.
As a preferred scheme of the present invention, the method for calculating the point cloud normal vector adopts a local surface fitting method to calculate, specifically:
assuming that the surface of the object is smooth everywhere, according to the differential geometry principle, the local area of each point in the point cloud data obtained by scanning can be represented by a plane, therefore, each point p in the point cloud data i The k nearest neighbors or k points within a certain radius can be obtained, and then the fitting plane P of these points is calculated according to the principle of least squares, the plane P being represented by the following mathematical equation:
wherein n is the normal vector of the plane P, d is the distance from the plane P to the origin of coordinates, argmin represents the variable value when the target function takes the minimum value, it can be known through calculation that the centroids of k adjacent points are located in the plane P, and the normal vector n satisfies, i.e. n is a unit vector, therefore, the problem of solving the normal vector can be converted into the problem of decomposing the eigenvalue of a semi-positive definite covariance matrix M, the minimum eigenvalue of the covariance matrix M is the solved normal vector n of the plane P, a formula is adopted, wherein T represents the matrix transposition,
means of expression
As a preferred embodiment of the present invention, in step S500, comparing two point cloud data based on a normal vector includes:
s510, acquiring two point cloud data after feature extraction;
s520, constructing a point cloud data index;
s530, selecting single-point laser point cloud data as a reference point cloud, taking an ith point, assuming that the coordinate of a certain point i in the reference point cloud is (X, Y, Z), and giving a radius R;
s540, searching point clouds with the radius of the point i as the center as the range R by using the point cloud data index, and calculating a normal vector N of the point cloud data in the range;
s550, taking a normal vector N of the point i as an axis of a cylinder of a search range, giving a radius r, and retrieving two pieces of point cloud data in the range by using point cloud data index;
s560, calculating the distance between the planes represented by the two point clouds to obtain the deformation quantity of the two point cloud data at the point i;
s570, performing the processing from the step S530 to the step S560 on all the points in the reference point cloud;
s580, deformation areas and deformation quantities of the two point cloud data are obtained.
As a preferred scheme of the present invention, in step S100, the vehicle-mounted lidar collection vehicle includes a lidar, a GPS, a camera, and an inertial measurement instrument, and the point cloud data acquired by the lidar, the GPS data, the image data acquired by the camera, and the position data acquired by the inertial measurement instrument are fused to obtain the vehicle-mounted lidar point cloud data.
As a preferred aspect of the present invention, in step S200, the point cloud data of the single-point laser is obtained by: the three-dimensional space coordinates of corresponding points are obtained through the emission and the return of laser beams, the array type geometric figure data of the three-dimensional surface of the terrain and the complex object are obtained in a point cloud mode, and the point position precision is controlled within 6 mm.
As a preferred embodiment of the present invention, in step S300, the denoising specifically is performed in the following manner: creating a statistical analysis filter, setting the filter to input point clouds to be processed, setting the number of adjacent points analyzed by each point, setting multiples of standard deviation, and if the distance of one point exceeds the average distance plus more than one standard deviation, marking the point as an outlier and removing the outlier.
As a preferred embodiment of the present invention, in step S300, the classification is specifically performed by: classifying the point cloud according to the characteristics, elevation, intensity and the like of the point cloud;
the simplification is specifically carried out in the following way: and establishing a rasterization topology relation of the point cloud based on the shape characteristics of the traffic elements, fitting a local paraboloid to estimate curvature values of all measuring points, and removing repeated points and redundant points in the point cloud main body.
As a preferable scheme of the present invention, between the steps S520 and S530, the method further includes the following steps: s521 determines whether the point cloud data needs to be simplified, if so, step S522 is performed to perform downsampling on the point cloud data, and then step S530 is performed, otherwise, step S530 is directly performed.
Compared with the prior art, the technical scheme of the invention has the following beneficial effects:
the invention establishes an automatic precision detection technology of single-point laser point cloud data and vehicle-mounted laser radar data, obtains the precision difference between two pieces of data mass characteristic point clouds by automatic denoising, classifying, repairing, characteristic extracting and normal vector comparison of the single-point laser point cloud data and the vehicle-mounted laser radar point cloud data obtained by field operation, and effectively ensures the precision and the accuracy of the vehicle-mounted laser radar point cloud data.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
FIG. 1 is a flow chart of a specific embodiment of a vehicle-mounted laser radar data precision detection method based on single-point laser point cloud according to the present invention;
FIG. 2 is a flow chart of a point cloud comparison method based on normal vectors in a specific embodiment of the vehicle-mounted laser radar data precision detection method based on single-point laser point cloud of the present invention;
FIG. 3 is a model diagram of normal vector calculation in a specific embodiment of the vehicle-mounted laser radar data accuracy detection method based on single-point laser point cloud of the present invention;
fig. 4 is a model diagram of deformation amount calculation in a specific embodiment of the vehicle-mounted laser radar data precision detection method based on single-point laser point cloud.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
High-precision maps, one of the technologies at the core of unmanned vehicles, have become an important basic technology in the intelligent internet automobile technology industry. Especially in the face of autonomous driving systems of the level above L3. Compared with the common map, the high-precision map has higher precision and generally needs to be controlled within 20CM, and the vehicle-mounted laser point cloud data is base data for producing high precision.
Referring to fig. 1, a method for detecting data accuracy of a vehicle-mounted laser radar based on a single-point laser point cloud in this embodiment includes the following steps:
s100, acquiring vehicle-mounted laser point cloud data: and acquiring the road condition of the high-precision map to be produced by using a vehicle-mounted laser radar acquisition vehicle so as to acquire corresponding vehicle-mounted laser point cloud data. The vehicle-mounted laser radar acquisition vehicle is provided with a laser radar, a GPS, a camera and an inertia measuring instrument, and point cloud data acquired by the laser radar, GPS data, image data acquired by the camera and position data acquired by the inertia measuring instrument are fused to obtain vehicle-mounted laser radar point cloud data. An inertial measurement unit is a device for vehicle-mounted laser scanning, and is used for position determination in a GPS unlocking state.
The method comprises the following specific steps:
the road condition of a high-precision map to be produced is collected by using a vehicle-mounted laser radar collecting vehicle, the vehicle speed is 40 kilometers per hour, the laser emission frequency is 200 Hz, the laser point cloud density is 60 ten thousand points per second, the image resolution is 3000 ten thousand pixels, and the inertial navigation null shift is 0.005 DEG/hour. Sequentially connecting equipment according to an operation flow, measuring a DMI (wheel encoder) offset component, connecting a GNSS receiver and a camera, and checking that the POS Mode state is Nav: aligned; the DMI state is ZUPD or OK; the RMS Accuracy value in Position decreases to around 1 m. After the static calibration is finished, the vehicle is started, an inertial navigation system (an inertial measurement instrument) is activated in a dynamic calibration mode, the yaw angle precision is improved, and the GPS receiving number is ensured to be more than 7 in the acquisition process.
When the number of the one-way passing lanes of the whole road is less than 3 lanes, the road is collected once, when the number of the one-way passing lanes is more than 3 lanes and is less than 6 lanes, the road is collected for 2 times, and the collection is performed on the road by analogy in sequence. The bidirectional traffic road is collected again and again according to the requirement of the number of lanes.
S200, acquiring single-point laser point cloud data: and determining fixed coordinates, erecting a base station type three-dimensional laser scanner, and scanning the road condition of the vehicle-mounted laser point cloud data needing to be calibrated to obtain corresponding single-point laser point cloud data. The base station type three-dimensional laser scanner performs full-automatic 3.5mm @25m precision stepping measurement from left to right and from top to bottom on a scanned object through a 200 ten thousand point/second high-speed laser scanning measurement method to obtain three-dimensional coordinates, reflection intensity and RGB information of the surface of the measured object and dense point cloud data, and scanned point cloud is accurately registered to the data through an ICP (inductively coupled plasma) and feature-based registration method, so that a high-precision data source is provided for high-precision navigation electronic map application. It should be noted that, the registration method based on ICP and feature is the prior art, and briefly, ICP finds a rotation matrix and a translation vector of two pieces of point clouds to form a target function, and the point clouds to be converted can be superposed on the target point cloud through rotation and translation; the feature-based registration is the matching of the same name points, and this embodiment is not described herein again.
The single-point laser point cloud data is obtained by the following method: the three-dimensional space coordinates of corresponding points are obtained through the emission and the return of laser beams, the array type geometric figure data of the three-dimensional surface of the terrain and the complex object are obtained in a point cloud mode, and the point position precision is controlled within 6 mm.
S300, automatic point cloud processing: analyzing mass point clouds in the early stage, and automatically denoising, classifying and simplifying on the basis of analysis data;
the denoising is specifically performed by the following method: the method comprises the steps of analyzing massive point clouds in the early stage, creating a statistical analysis filter, confirming adjacent points (the number of the adjacent points analyzed by each point is set) and standard deviation multiples, if the distance of one point exceeds the average distance and is more than one standard deviation, marking the point as an outlier, removing the outlier, inputting a numerical value into a detection tool, and importing vehicle-mounted laser point cloud data and single-point laser point cloud data.
The classification is specifically carried out in the following manner: in order to realize automatic extraction and point cutting of vector data, point cloud classification and extraction are realized on the basis of semantic features, point cloud is classified according to characteristics, elevation, strength and the like of the point cloud, and after classification, surfaces of standard structures such as buildings and the ground are easier to obtain, so that corresponding normal vectors are calculated, and the calculated amount is reduced.
The method comprises the steps of directly acquiring laser point cloud data, comparing two pieces of data at the later stage, preliminarily classifying point cloud data by adopting the three-dimensional space position and the projection density of the laser point cloud, and then classifying the semantic characteristics, the macroscopic characteristics and the local set characteristics of a road structure in detail.
The simplification is specifically carried out in the following way: based on the shape characteristics of the traffic elements, establishing a grid expansion relation of the point cloud, fitting a local paraboloid to estimate curvature values of all measuring points, and removing repeated points and redundant points in the point cloud main body.
S400, automatic feature extraction: materializing the denoised, classified and simplified point cloud data, and extracting element features by combining a random sampling consistency algorithm with robust estimation; the random sample consensus algorithm is a robust plane fitting method based on random sample consensus (RANSAC) in combination with robust estimation, which is not described herein.
S500, automatic matching calculation: after the characteristics are extracted, point cloud normal vectors are calculated by taking the single-point laser point cloud data as a reference, and two pieces of point cloud data are compared based on the normal vectors;
the method for calculating the point cloud normal vector adopts a local surface fitting method for calculation, and specifically comprises the following steps:
assuming that the surface of the object is smooth everywhere, according to the differential geometry principle, the local area of each point in the point cloud data obtained by scanning can be represented by a plane, therefore, each point p in the point cloud data i Then, k nearest neighbors or k points within a certain radius can be obtained, and then a fitting plane P of these points is calculated according to the principle of least squares, the plane P being represented by the following mathematical equation:
wherein n is a normal vector of the plane P, d is a distance from the plane P to an origin of coordinates, argmin represents a variable value when the target function takes a minimum value, it can be known through calculation that the centroid of k adjacent points is located in the plane P, and the normal vector n satisfies, i.e., n is a unit vector, so that the problem of solving the normal vector can be converted into the problem of decomposing the eigenvalue of a semi-positive definite covariance matrix M, the minimum eigenvalue of the covariance matrix M is the normal vector n of the plane P, a formula is adopted, wherein T represents the matrix transposition,
represents the mean value
Referring to fig. 2, comparing two point cloud data based on a normal vector specifically includes the following steps:
s510, acquiring two point cloud data after feature extraction;
s520, constructing a point cloud data index to accelerate the searching speed;
s521, selecting the point cloud data of the single-point laser as a reference point cloud, judging whether the point cloud data needs to be simplified, namely judging whether the point cloud data volume is large (whether the point spacing is smaller than 1 cm), if so, executing a step S522, then executing a step S530, and if not, directly executing the step S530;
s522, down-sampling the point cloud data to accelerate algorithm processing;
s530, selecting single-point laser point cloud data as a reference point cloud, taking the ith point, assuming that the i coordinate of a certain point in the reference point cloud is (X, Y, Z), and giving a radius R;
s540, using the point cloud data index to retrieve the point cloud with the radius of the point i as the center as the range of R, calculating the normal vector N of the point cloud data in the range, using the vector of the method as the direction reference of the subsequent calculation, and using the normal vector calculation method as shown in FIG. 3;
s550, taking a normal vector N passing through the point i as an axis of a search range cylinder, giving a radius r, and retrieving two point cloud data in the range by using a point cloud data index;
s560 calculates the distance between the planes represented by the two point clouds to obtain the deformation amount of the two point cloud data at the point i, and the deformation amount calculation method is shown in fig. 4.
S570, processing steps S530 to S560 are carried out on all points in the reference point cloud;
and S580, obtaining deformation areas and deformation quantities of the two point cloud data.
S600, providing an accuracy report: and according to the comparison result, taking the single-point laser point cloud data as standard data, and issuing a precision report after comparing the vehicle-mounted laser point cloud data with the single-point laser point cloud data.
According to the invention, the automatic precision detection technology of the single-point laser point cloud data and the vehicle-mounted laser radar data is established through the embodiment, and the precision difference between the two massive characteristic point clouds is obtained through the automatic denoising, classification, repairing, characteristic extraction and normal vector comparison of the single-point laser point cloud data and the vehicle-mounted laser radar point cloud data obtained by the field, so that the precision and the accuracy of the vehicle-mounted laser radar point cloud data are effectively ensured.
According to the embodiment of the invention, the precision detection efficiency and convenience of the data of the vehicle-mounted laser radar are improved, the data of a 100-kilometer urban area are taken as an example, about 400 calibration points are required to be obtained by manual dotting, the field time is about 52 hours, the field time is 16 hours, the total time is 68 hours, and the limit that the road surface characteristic points cannot be obtained, and the weather and traffic conditions exist. Through the matching check of the single-point laser point cloud data and the vehicle-mounted laser radar data, the field time is 6 hours, the field time is 0.5 hour, the total time is 6.5 hours, and the method is not limited by weather and traffic conditions.
The details of one of the features provided by the present invention are described above. The principles and embodiments of the present invention have been described herein using specific examples, which are presented only to assist in understanding the method and its core concepts of the present invention. It should be noted that, for those skilled in the art, without departing from the principle of the present invention, it is possible to make various improvements and modifications to the present invention, and those improvements and modifications also fall within the scope of the claims of the present invention.
Claims (10)
1. A vehicle-mounted laser radar data precision detection method based on single-point laser point cloud is characterized by comprising the following steps:
s100, vehicle-mounted laser point cloud data are obtained: acquiring road conditions of a high-precision map to be produced by using a vehicle-mounted laser radar acquisition vehicle to acquire corresponding vehicle-mounted laser point cloud data;
s200, acquiring single-point laser point cloud data: determining fixed coordinates, erecting a base station type three-dimensional laser scanner, and scanning road conditions needing to be checked and calibrated for vehicle-mounted laser point cloud data to obtain corresponding single-point laser point cloud data;
s300, automatic point cloud processing: analyzing mass point clouds in the early stage, and automatically denoising, classifying and simplifying on the basis of analyzing data;
s400, automatic feature extraction: materializing the denoised, classified and simplified point cloud data, and extracting element features by combining a random sampling consistency algorithm with robust estimation;
s500, automatic matching calculation: after the characteristics are extracted, point cloud normal vectors are calculated by taking the single-point laser point cloud data as a reference, and two pieces of point cloud data are compared based on the normal vectors;
s600, issuing an accuracy report: and according to the comparison result, taking the single-point laser point cloud data as standard data, and issuing a precision report after comparing the vehicle-mounted laser point cloud data with the single-point laser point cloud data.
2. The vehicle-mounted laser radar data precision detection method based on the single-point laser point cloud as claimed in claim 1, wherein the calculation method of the point cloud normal vector adopts a local surface fitting method for calculation, and specifically comprises the following steps:
assuming that the surface of the object is smooth everywhere, the local area of each point in the point cloud data obtained by scanning can be represented by a plane according to the differential geometry principle, so that each point p in the point cloud data i The k nearest neighbors or k points within a certain radius can be obtained, and then the fitting plane P of these points is calculated according to the principle of least squares, the plane P being represented by the following mathematical equation:
wherein n is the normal vector of the plane P, d is the distance from the plane P to the origin of coordinates, argmin represents the variable value when the target function takes the minimum value, it can be known through calculation that the centroids of k adjacent points are located in the plane P, and the normal vector n satisfies, i.e. n is a unit vector, therefore, the problem of solving the normal vector can be converted into the problem of decomposing the eigenvalue of a semi-positive definite covariance matrix M, the minimum eigenvalue of the covariance matrix M is the solved normal vector n of the plane P, and the method adopts the method such asThe following formula, where T represents a matrix transpose,
means of expression
3. The method for detecting data accuracy of vehicle-mounted lidar based on single-point laser point cloud as claimed in claim 1, wherein the step S500 of comparing two point cloud data based on normal vector specifically comprises the following steps:
s510, acquiring two point cloud data after feature extraction;
s520, constructing a point cloud data index;
s530, selecting single-point laser point cloud data as a reference point cloud, taking the ith point, assuming that the i coordinate of a certain point in the reference point cloud is (X, Y, Z), and giving a radius R;
s540, using the point cloud data index to retrieve the point cloud with the radius of the point i as the center as the range of R, and calculating the normal vector N of the point cloud data in the range;
s550, taking a normal vector N passing through the point i as an axis of a search range cylinder, giving a radius r, and retrieving two point cloud data in the range by using a point cloud data index;
s560, calculating the distance between the planes represented by the two point clouds to obtain the deformation quantity of the two point cloud data at the point i;
s570, performing the processing from the step S530 to the step S560 on all the points in the reference point cloud;
s580, deformation areas and deformation quantities of the two point cloud data are obtained.
4. The method for detecting data accuracy of vehicle-mounted lidar based on single-point laser point cloud as claimed in claim 1, wherein in step S100, the vehicle-mounted lidar collection vehicle has a lidar, a GPS, a camera and an inertial measurement instrument, and the point cloud data acquired by the lidar, the GPS data, the image data acquired by the camera and the position data acquired by the inertial measurement instrument are fused to obtain the vehicle-mounted lidar point cloud data.
5. The method for detecting data accuracy of vehicle-mounted laser radar based on single-point laser point cloud as claimed in claim 4, wherein in step S100, the vehicle-mounted laser radar collecting vehicle is statically calibrated, then the vehicle-mounted laser radar collecting vehicle is started, then the inertial measuring instrument is activated in a dynamic calibration mode, the yaw angle accuracy is improved, and the GPS receiving number is ensured to be more than 7 in the collecting process; the vehicle-mounted laser radar acquisition vehicle adopts the speed of 40 kilometers per hour, the laser emission frequency of 200 Hz, the laser point cloud density of 60 ten thousand points per second, the image resolution of 3000 ten thousand pixels and the inertial navigation null shift of 0.005 degrees per hour.
6. The method for detecting data accuracy of vehicle-mounted lidar based on single-point laser point cloud as claimed in claim 1, wherein in step S200, the base station type three-dimensional laser scanner performs full-automatic step measurement with 3.5mm @25m accuracy from left to right and from top to bottom on the scanned object by a method of high-speed laser scanning measurement at 200 ten thousand points/second to obtain three-dimensional coordinates, reflection intensity and RGB information of the surface of the measured object and dense point cloud data, and the scanned point cloud is accurately registered with the data by using an ICP-based and feature-based registration method.
7. The method for detecting data accuracy of vehicle-mounted lidar based on single-point laser point cloud according to claim 1, wherein in step S200, the single-point laser point cloud data is obtained by: three-dimensional space coordinates of corresponding points are obtained through emission and return of laser beams, array type geometric figure data of three-dimensional surfaces of landforms and complex objects are obtained in a point cloud mode, and point position accuracy is controlled within 6 mm.
8. The method for detecting data accuracy of vehicle-mounted lidar based on single-point laser point cloud as claimed in claim 1, wherein in step S300, the denoising is specifically performed by: creating a statistical analysis filter, setting the filter to input point clouds to be processed, setting the number of adjacent points analyzed by each point, setting multiples of standard deviation, and if the distance of one point exceeds the average distance plus more than one standard deviation, marking the point as an outlier and removing the outlier.
9. The method for detecting data accuracy of vehicle-mounted lidar based on single-point laser point cloud as claimed in claim 1, wherein in step S300, the classification is specifically performed by: classifying the point cloud according to the characteristics, elevation, intensity and the like of the point cloud;
the simplification is specifically carried out in the following way: based on the shape characteristics of the traffic elements, establishing a grid expansion relation of the point cloud, fitting a local paraboloid to estimate curvature values of all measuring points, and removing repeated points and redundant points in the point cloud main body.
10. The method for detecting data accuracy of vehicle-mounted lidar based on single-point laser point cloud of claim 1, wherein between the steps S520 and S530, the method further comprises the following steps: s521 determines whether the point cloud data needs to be simplified, if so, step S522 is performed to perform downsampling on the point cloud data, and then step S530 is performed, otherwise, step S530 is directly performed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211151258.2A CN115561768A (en) | 2022-09-19 | 2022-09-19 | Vehicle-mounted laser radar data precision detection method based on single-point laser point cloud |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211151258.2A CN115561768A (en) | 2022-09-19 | 2022-09-19 | Vehicle-mounted laser radar data precision detection method based on single-point laser point cloud |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115561768A true CN115561768A (en) | 2023-01-03 |
Family
ID=84740234
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211151258.2A Pending CN115561768A (en) | 2022-09-19 | 2022-09-19 | Vehicle-mounted laser radar data precision detection method based on single-point laser point cloud |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115561768A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116538996A (en) * | 2023-07-04 | 2023-08-04 | 云南超图地理信息有限公司 | Laser radar-based topographic mapping system and method |
-
2022
- 2022-09-19 CN CN202211151258.2A patent/CN115561768A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116538996A (en) * | 2023-07-04 | 2023-08-04 | 云南超图地理信息有限公司 | Laser radar-based topographic mapping system and method |
| CN116538996B (en) * | 2023-07-04 | 2023-09-29 | 云南超图地理信息有限公司 | Laser radar-based topographic mapping system and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111598823B (en) | Multisource mobile measurement point cloud data space-ground integration method and storage medium | |
| Holgado‐Barco et al. | Semiautomatic extraction of road horizontal alignment from a mobile LiDAR system | |
| Elkhrachy | Accuracy assessment of low-cost unmanned aerial vehicle (UAV) photogrammetry | |
| EP3637371B1 (en) | Map data correcting method and device | |
| EP3631494B1 (en) | Integrated sensor calibration in natural scenes | |
| Csanyi et al. | Improvement of lidar data accuracy using lidar-specific ground targets | |
| CN102208013B (en) | Landscape coupling reference data generation system and position measuring system | |
| KR101632486B1 (en) | Method for extracting curb of road using laser range finder and method for localizing of mobile robot using curb information of road | |
| CN102506824B (en) | Method for generating digital orthophoto map (DOM) by urban low altitude unmanned aerial vehicle | |
| US8180100B2 (en) | Plane detector and detecting method | |
| CN112740225B (en) | Method and device for determining road surface elements | |
| CN112446844B (en) | Point cloud feature extraction and registration fusion method | |
| CN107564046A (en) | It is a kind of based on a cloud and the secondary accurate extracting method of registering contour of building of UAV images | |
| CN114526745A (en) | Drawing establishing method and system for tightly-coupled laser radar and inertial odometer | |
| CN110864696A (en) | Three-dimensional high-precision map drawing method based on vehicle-mounted laser inertial navigation data | |
| CN114877838B (en) | Road geometric feature detection method based on vehicle-mounted laser scanning system | |
| CN111913169A (en) | Method, equipment and storage medium for correcting laser radar internal reference and point cloud data | |
| CN114829971A (en) | Laser radar calibration method and device and storage medium | |
| CN115561768A (en) | Vehicle-mounted laser radar data precision detection method based on single-point laser point cloud | |
| CN110046563B (en) | Power transmission line section elevation correction method based on unmanned aerial vehicle point cloud | |
| CN115063465A (en) | Unmanned vehicle driving road condition modeling method based on laser radar | |
| CN113822944A (en) | External parameter calibration method and device, electronic equipment and storage medium | |
| Zhou et al. | Crater identification simulation using LiDAR on Lunar rover | |
| Guo et al. | Novel Trajectory Optimization Algorithm of Vehicle-borne LiDAR Mobile Measurement System. | |
| CN110927765A (en) | Laser radar and satellite navigation fused target online positioning method |
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 |