CN111127530A - Accurate registration method for three-dimensional laser point clouds of multi-phase roads - Google Patents

Accurate registration method for three-dimensional laser point clouds of multi-phase roads Download PDF

Info

Publication number
CN111127530A
CN111127530A CN201911341604.1A CN201911341604A CN111127530A CN 111127530 A CN111127530 A CN 111127530A CN 201911341604 A CN201911341604 A CN 201911341604A CN 111127530 A CN111127530 A CN 111127530A
Authority
CN
China
Prior art keywords
point cloud
data
point
stage
calculating
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.)
Granted
Application number
CN201911341604.1A
Other languages
Chinese (zh)
Other versions
CN111127530B (en
Inventor
刘如飞
卢秀山
邢恺强
马新江
杨雷
杨凯
杨继奔
王飞
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shandong University of Science and Technology
Original Assignee
Shandong University of Science and Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shandong University of Science and Technology filed Critical Shandong University of Science and Technology
Priority to CN201911341604.1A priority Critical patent/CN111127530B/en
Publication of CN111127530A publication Critical patent/CN111127530A/en
Application granted granted Critical
Publication of CN111127530B publication Critical patent/CN111127530B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
    • G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
    • 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
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Processing Or Creating Images (AREA)

Abstract

The invention discloses a three-dimensional laser point cloud accurate registration method for a multi-phase road, which comprises the steps of segmenting point cloud data according to a certain distance to obtain a plurality of segments of data, then registering the plurality of segments of data respectively, firstly correcting deviation perpendicular to the driving lane direction, secondly correcting deviation in the driving lane direction, and finally correcting elevation to obtain finally registered data. The invention carries out sectional processing on the large scene point cloud data, simultaneously adopts independent ground objects and road marked lines as registration elements, optimizes the registration process and solves the problem of low speed of a point-to-point registration mode of mass point cloud data.

Description

Accurate registration method for three-dimensional laser point clouds of multi-phase roads
Technical Field
The invention relates to the technical field of fine registration of large-scene point clouds of a mobile measurement system, in particular to a precise registration method of a multi-phase road three-dimensional laser point cloud.
Background
Due to the development of social economy, roads and bridges in China are more and more widely constructed, road and bridge data contained in large scene point cloud data are more and more abundant, and along with the increase of service life, the method is very important for monitoring the change in the scene in order to ensure the driving safety.
The fundamental premise of change detection is to realize high-precision registration of multi-stage point cloud data, and at present, the high-precision registration of the multi-stage point cloud data at home and abroad mainly depends on manual interactive selection of homonymous independent ground objects for high-precision registration to obtain all parameters required by the registration. The method can greatly reduce the time required by registration and the data volume of the registration element, and improve the registration precision and efficiency.
In the process of implementing the invention, the inventor finds that the prior method has the following defects: in large scene point cloud data, due to the large data volume, it is difficult to directly use all point cloud data to perform point-to-point registration, and in scenes such as highways, necessary independent objects are extremely lacking as registration elements, so that the requirement of high-precision registration cannot be met.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention provides a three-dimensional laser point cloud accurate registration method for a multi-phase road.
The technical scheme adopted by the invention is that the method for accurately registering the three-dimensional laser point clouds of the multi-stage road specifically comprises the following steps:
a first step; acquiring multi-stage point cloud data of a monitored road, selecting any two-stage original point cloud data, and performing equidistant segmentation processing on the original point cloud data to generate a plurality of sections of point cloud data;
the second step is that: and (3) carrying out deviation correction on each segmented point cloud data in a direction vertical to a driving lane:
the third step: carrying out deviation correction on the driving lane direction of each segmented point cloud data;
the fourth step: performing elevation correction on each segmented point cloud data to finally obtain a complete registration result of the two-stage data;
the fifth step: and selecting two periods of original point cloud data, repeating the steps from the first step to the fourth step, and so on to obtain a complete registration result of the multi-period data.
Further: the second step specifically comprises the following steps:
2.1: aiming at each single section of data, firstly selecting two periods of point clouds of same-name pavement markings, and solving abrupt change points between the markings and the ground points on the single scanning line containing the point clouds of the markings by using a Sigmoid mathematical function model according to abrupt change characteristics of point cloud intensity values on the single scanning line between the ground points and the markings and taking the abrupt change points as marking boundary points;
and 2.2, calculating each centerline point of the reticle according to the solved reticle boundary points, fitting a straight line by using a least square method to serve as the reticle centerline, solving the reticle centerline of the two periods of the same-name reticles, and solving the deviation of the two periods of the reticles in the direction vertical to the road driving direction according to the reticle centerline.
Further, step 2.1 comprises the following sub-steps:
2.1.1: setting the distance, and performing data segmentation on the point cloud data to be registered and the template point cloud data;
2.1.2: manually and uniformly selecting a plurality of homonymous marked lines of the two-stage point cloud data;
2.1.3: selecting a homonymous marking, and calculating marking boundary points on each scanning line of the two-stage marking by using a Sigmoid function model according to the sudden change characteristics of point cloud intensity values on the scanning lines containing marking point clouds between ground points and marking points, wherein the function model is as follows:
Figure BDA0002332421030000021
wherein, S (x) is the intensity value corresponding to each reticle boundary point;
x is the x, y coordinate value of each reticle boundary point, so as to obtain the invented productCalculating the boundary point of each marking line as-b2；
B in the formula1、b3、b4Respectively representing the coefficients of the formula, wherein b1Scale factor representing the function curve in x-axis, b3Scale factor representing the function curve in y-axis, b4Represents the translation of the function curve relative to the origin (0,0) on the y-axis; e is the natural coefficient of the mathematical noun.
Further, step 2.2 comprises the following sub-steps:
2.2.1: calculating the central point of each marking line according to the boundary points of each marking line obtained by calculation;
2.2.2: fitting the central line of the marked line according to a least square fitting method by using the central point of each marked line:
Figure BDA0002332421030000031
Figure BDA0002332421030000032
wherein x and y are coordinates of the central point of each marked line participating in calculation, i is a coordinate base mark participating in calculation, n is the number of the coordinates participating in calculation,
Figure BDA0002332421030000033
for the slope of the fitted centerline to be,
Figure BDA0002332421030000034
is the fitted centerline intercept;
2.2.3: calculating the distance between the lines in the marked lines with the same name according to a point-to-straight line distance formula, and calculating the offset of the marked line to be registered in each marked line with the same name in the x and y directions compared with the original marked line;
Figure BDA0002332421030000035
Figure BDA0002332421030000036
in the formula, k and d respectively represent the slope and intercept of a linear equation where the midline of the original marked line in a pair of marked lines with the same name is located, and x and y are three-dimensional coordinates of points on the midline of the cloud marked line of the point to be registered;
2.2.4: and (4) performing the 2.2.1-2.2.3 steps of calculation on each selected homonymous marked line, calculating the average value of the translation amount of each homonymous marked line, and translating the cloud point to be registered in the x and y directions to finish the first correction.
Further: the third step specifically comprises the following steps:
3.1: and (3) interactively selecting multiple same-name independent feature data on two sides of a two-stage road as registration elements:
3.2: respectively registering multiple homonymous independent objects on two sides of the two-period data road by using an ICP (inductively coupled plasma) algorithm, calculating offsets OffsetX and OffsetY in x and y directions, and calculating an average value
Figure BDA0002332421030000041
Figure BDA0002332421030000042
3.3 respectively calculating the error б according to the calculated offset of the multiple groups of the same-name independent objects in the x and y directionsx，бyAnd will be within range
Figure BDA0002332421030000043
Recalculating the average value of the internal offset to be used as a second correction registration offset;
3.4: and translating the cloud point to be registered in the x and y directions to finish the second correction.
Further, the fourth step specifically comprises the following steps:
4.1: for two-stage data after data segmentation, ground point cloud data is obtained by using a road filtering algorithm, and a unified grid index is established for the two-stage homonymous road point cloud data, wherein the specific method comprises the following steps:
4.1.1: acquiring a common minimum external rectangle of the point cloud plane projection of the two-stage road surface, namely: xmin, ymin, xmax, ymax;
4.1.2: setting a certain grid step length, and dividing grids for two-stage data:
Figure BDA0002332421030000044
Figure BDA0002332421030000045
ID＝(RID-1)×Col+CID
in the formula: ID is the index number of the grid, Row and Col are the number of rows and columns of the grid, RID and CID are the number of rows and columns, and xi、yiRespectively, the coordinates of the point cloud, and StepX and StepY respectively are the length and width of the grid.
4.2: calculating a normal vector of each grid point cloud in the two-stage pavement point cloud data by using a Principal Component Analysis (PCA) algorithm; and then calculating the size of an included angle theta between a normal vector of each grid point cloud of the two-stage road surface point cloud data and a normal vector (0,0,1) of a point cloud z axis:
Figure BDA0002332421030000051
Figure BDA0002332421030000052
4.3: respectively carrying out statistical analysis on the point cloud data of the two-stage road surface and determining an invariant region (namely a stable region), wherein the method specifically comprises the following steps:
4.3.1: setting the same included angle interval step length, and counting the number of corresponding grids and the corresponding ID number in each interval;
4.3.2: counting and analyzing the upper limit and the lower limit of the interval with the maximum grid number, taking the interval as a standard interval, regarding a point cloud area in the standard interval as a stable area in the two-stage data, namely an area without serious deformation, and recording an ID number corresponding to the grid in the interval;
4.4: and counting the ID numbers of the common grids of the two-stage same-name ground data, dividing the grids under the same standard, and then inevitably generating overlapped grid IDs, respectively calculating the average elevation of the point clouds in the common grids of the two-stage same-name ground data, calculating the elevation deviation of the two-stage data according to the difference of the average elevations, and finally correcting.
Compared with the prior art, the invention has the beneficial effects that:
(1) the method carries out sectional processing on the large scene point cloud data, simultaneously adopts independent ground objects and road marked lines as registration elements, optimizes the registration process and solves the problem of low speed of a point-to-point registration mode of mass point cloud data;
(2 the invention utilizes the road marking as a part of the registration primitive, can directly adopt the manual interactive extraction road marking as the registration primitive, overcome the extremely deficient regional registration result lower precision disadvantage of independent ground feature, have realized the registration of more simple and efficient.
(3) The invention utilizes principal component analysis and statistical analysis during elevation correction, so that the robustness of the algorithm is stronger, the convenience and the reliability of data processing are improved, and the defects of surveying and mapping auxiliary tools such as a target and the like during general elevation correction are overcome.
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 some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a flow chart of a method for accurately registering a three-dimensional laser point cloud of a multi-phase road according to the invention;
FIG. 2 is a flow chart of progressive refined registration data based on vehicle-mounted three-dimensional laser point cloud in the present invention;
fig. 3 is a schematic diagram of offset correction before and after the cloud data of the point to be registered is perpendicular to the driving lane direction, wherein fig. 3a and 3b are a top view and a cross-sectional view before correction, and fig. 3c and 3d are a top view and a cross-sectional view after correction;
figure 4 is a schematic diagram of the calculation of reticle boundary points using Sigmoid function,
fig. 5a is an original point cloud picture of point cloud data to be registered, and fig. 5b is a point cloud picture after first correction of the point cloud data to be registered, which is vertical to the driving lane direction;
fig. 6 is a schematic top view of a pre-correction and post-correction marked line and a schematic top view of an independent ground object in the driving lane direction of cloud data of a point to be registered, wherein fig. 6a and 6b are a schematic top view of the cloud data marked line of the point to be registered before correction and a schematic top view of the independent ground object respectively, and fig. 6c and 6d are a schematic top view of the cloud data marked line of the point to be registered after correction and a schematic top view of the independent ground object.
FIG. 7 is a diagram of results before and after offset correction of cloud data driving lane directions of points to be registered; where fig. 7a and 7b are an example of a graph of the results after a first correction, and fig. 7c and 7d are graphs of the results after a second correction for the examples given in fig. 7a and 7 b.
Fig. 8 is a schematic view of elevation correction of cloud data of a point to be registered, where fig. 8a is a cross-sectional view of the point cloud data before correction, and fig. 8b is a cross-sectional view of the point cloud data after correction;
FIG. 9 is a schematic diagram of determining a standard interval by performing statistical analysis on two-stage road surface point cloud data respectively;
fig. 10 is a diagram of a result of elevation correction of cloud data of a point to be registered.
Detailed Description
The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings so that the advantages and features of the present invention can be more easily understood by those skilled in the art, and the scope of the present invention will be more clearly and clearly defined.
As shown in fig. 1-2, the method for accurately registering the three-dimensional laser point clouds of the multi-phase road mainly comprises the following steps:
the first step is as follows: the method comprises the steps of preprocessing original point cloud data, wherein a certain distance length is set, and the two-stage homonymous original point cloud data are respectively subjected to segmentation processing to generate a plurality of sections of point cloud data, so that the subsequent processing is facilitated;
the second step is that: taking the single segment data as an example, firstly, the deviation correction perpendicular to the driving lane direction is performed:
as shown in fig. 3, the rectangle in the figure represents a road marking, and the purpose of this step is to convert the two-stage point cloud (point cloud to be registered and template point cloud) from the state of 3a, 3b to the state of 3c, 3d, so as to realize the offset correction perpendicular to the driving lane direction. The specific process is as follows:
2.1: and uniformly selecting a plurality of homonymous marked lines of the two-stage point cloud data in an interactive manner, and numbering and managing the marked lines in groups.
2.2: and calculating the boundary points of the marked lines, and calculating the coordinates of the center points of the marked lines according to the boundary points of the marked lines obtained by calculation.
Taking a single scan line as an example, the x, y coordinate values and the intensity values of each point on the scan line have the corresponding relationship as shown in fig. 4, wherein the abscissa axis is the x or y coordinate value, and the ordinate is the intensity value of the point. Fig. 4 reflects the intensity value variation trend of point cloud on the scanning line from the ground point to the reticle point, and then from the reticle point to the ground point, so as to obtain the intensity value variation trend, as shown in fig. 4, the variation curve can be linearly fitted by using Sigmoid function, and the corresponding point with the maximum variation rate is calculated as the boundary point at the two ends of the reticle:
Figure BDA0002332421030000081
wherein S (x) is the intensity value corresponding to each point, x is the x or y coordinate value of the point, and the x or y coordinate value of the boundary point of the reticle calculated by the method is-b2Therefore, the boundary point coordinates of the marked line on the single scanning line can be obtained.
And (4) performing the calculation processing on each scanning line in the manually selected marking line point cloud, so that the boundary point of the whole marking line can be obtained.
Calculating the coordinates of the center points of the marking lines according to the calculated boundary points of the marking lines:
Figure BDA0002332421030000082
Figure BDA0002332421030000083
wherein the content of the first and second substances,
Figure BDA0002332421030000084
and
Figure BDA0002332421030000085
is the abscissa of the left and right boundary points of the marked line,
Figure BDA0002332421030000086
and
Figure BDA0002332421030000087
is the ordinate of the left and right boundary points of the marked line,
Figure BDA0002332421030000088
and
Figure BDA0002332421030000089
and coordinates of the middle point of the marked line.
2.3: and fitting the center line of the marked line.
And fitting the central line of the marked line according to a least square fitting method by using the central point of each marked line, wherein a central line equation is calculated as follows:
Figure BDA00023324210300000810
Figure BDA00023324210300000811
the method is used for solving the center line of the homonymous marked line of the two-stage point cloud data.
2.4: and calculating the offset of the two-stage marked line in the direction vertical to the road driving direction.
Firstly, calculating the distance between the central lines of the homonymous marked lines of the two-stage point cloud data according to a distance formula from a point to a straight line, wherein the distance calculation method from the point to the straight line is as follows:
Figure BDA0002332421030000091
wherein x and y are three-dimensional coordinates of points on the center line of the cloud marked line of the point to be registered, d is the distance from the center line of the cloud marked line of the point to be registered to the center line of the point cloud marked line of the template point, and k and b respectively represent the slope and intercept of a linear equation.
Calculating the offset in the x and y directions according to the distance between the lines in the same-name graticule and a graticule equation:
Figure BDA0002332421030000092
Figure BDA0002332421030000093
calculating the translation amount among a plurality of groups of same-name mark lines which are manually and interactively selected in the point cloud data, and obtaining the translation amount by averaging
Figure BDA0002332421030000094
And
Figure BDA0002332421030000095
2.5: respectively translating the x and y coordinates of the cloud data of the section of point to be registered
Figure BDA0002332421030000096
And
Figure BDA0002332421030000097
and the first correction, i.e. the correction perpendicular to the road running direction, is completed, and the result of the correction is shown in fig. 5.
The third step: taking single segment data as an example, then performing deviation correction of the driving lane direction:
as shown in fig. 6a to 6d and fig. 7a to 7d, the rectangles in the drawings represent road markings, and the circles represent street lamps or other independent objects, and the purpose of this step is to correct the deviation of the driving lane direction by two-stage point clouds (point cloud to be registered and template point cloud) from 6a, 6b, 7a, 7b to 6c, 6d, 7c and 7 d.
3.1: and uniformly selecting a plurality of homonymous independent ground objects of the two-stage point cloud data in an interactive manner, and numbering and managing the independent ground objects in groups.
3.2: and (3) error-considered multi-element ICP point cloud registration.
And registering the point cloud to be registered and the template point cloud for each group of independent objects by utilizing an ICP (inductively coupled plasma) algorithm. The core idea of the ICP algorithm is as follows: respectively finding out the nearest point (P) in the point cloud P to be registered and the target point cloud Q according to certain constraint conditionsi,qi) Then, optimal matching parameters R and t are calculated so that the error function is minimized. The error function is E (R, t) is:
Figure BDA0002332421030000101
where n is the number of nearest neighbor point pairs, piFor a point in the point cloud P to be registered, qiNeutralization p in target point cloud QiAnd R is a rotation matrix and t is a translation vector.
ICP algorithm step:
① Point set P in Point cloud P to be registeredi∈P；
② find the corresponding point set Q in the target point cloud QiBelongs to Q, so that | | | pi-qi||＝min；
③, calculating a rotation matrix R and a translation matrix t to minimize an error function, wherein the rotation matrix R is an identity matrix because the registration between two phase point clouds in the registration method is regarded as rigid transformation, namely:
Figure BDA0002332421030000102
④ pairs of piCarrying out rotation and translation transformation by using the rotation matrix R and the translation matrix t obtained in the previous step to obtain a new corresponding point set pi′＝{pi′＝Rpi+t,pi∈P}；
⑤ calculating pi' and corresponding point set qiAverage distance d of (d):
Figure BDA0002332421030000103
⑥ if d is less than a given threshold or greater than a preset maximum iteration number, stopping iteration calculation, otherwise, returning to step 2 until meeting convergence conditionx，ty，tz]T。
3.3: calculating the translation amount of a plurality of independent objects with the same name selected by manual interaction in the point cloud data
Figure BDA0002332421030000106
Is worth averaging
Figure BDA0002332421030000104
And
Figure BDA0002332421030000105
according to the calculated offset of the x and y directions of the multiple groups of homonymous independent objects
Figure BDA0002332421030000111
Separately calculate median error бx，бy：
Figure BDA0002332421030000112
Figure BDA0002332421030000113
Will be in range
Figure BDA0002332421030000114
Inner offset recalculating average
Figure BDA0002332421030000115
And
Figure BDA0002332421030000116
as a second correction registration offset;
3.4: respectively translating the x and y coordinates of the cloud data of the section of point to be registered
Figure BDA0002332421030000117
And
Figure BDA0002332421030000118
and (4) completing the second correction, namely the correction of the driving lane direction, and the correction results are shown in figures 7c-7 d.
And fourthly, taking single segmented data as an example, and finally performing elevation deviation correction:
4.1: as shown in fig. 8a-8b, fig. 8a is a cross-sectional view of two-stage point cloud data of the same name of road surface, which aims to convert the two-stage point cloud (point cloud to be registered and template point cloud) from the state of fig. 8a to the state of fig. 8b, and implement elevation deviation correction.
4.2: and establishing a unified grid index for the point cloud data of the two-stage same-name pavement.
Acquiring a common minimum external rectangle of the point cloud plane projection of the two-stage road surface, namely: xmin, ymin, xmax, ymax.
Setting a certain grid step length, dividing the grid into two-stage data, numbering the grid and establishing an index:
Figure BDA0002332421030000119
Figure BDA00023324210300001110
ID＝(RID-1)×Col+CID
wherein, ID is the index number of the grid, Row and Col are the number of rows and columns of the grid, and RID and CID are the number of rows and columns.
4.3: and calculating the normal vector of each grid point cloud in the two-stage road surface point cloud data by using a PCA algorithm.
The PCA algorithm comprises the following steps:
①, forming a matrix with n rows and 3 columns by the x, y and z coordinates of the point clouds in each grid network according to columns;
② zero-averaging each row of the matrix, i.e. subtracting the average of this row;
③ a covariance matrix is constructed:
Figure BDA0002332421030000121
④ find the minimum eigenvalue of the covariance matrix and the corresponding eigenvector as normal vectors, normalx, normaly, normalz.
⑤, calculating the size of an included angle theta between the normal vector of each grid point cloud of the two-stage road surface point cloud data and the normal vector (0,0,1) of the z axis of the point cloud:
Figure BDA0002332421030000122
Figure BDA0002332421030000123
4.4: respectively carrying out statistical analysis on the point cloud data of the two-stage road surface:
as shown in fig. 9, setting the same included angle interval step length, and counting the number of grids corresponding to each interval and the corresponding ID number;
meanwhile, the upper limit and the lower limit of the interval with the maximum grid number are statistically analyzed, the interval is taken as a standard interval, a point cloud area in the standard interval is taken as a stable area in the two-stage data, namely, an area without serious deformation, and the ID number corresponding to the grid in the interval is recorded;
counting the common grid ID numbers of two-stage same-name ground data, dividing the grids under the same standard, and then inevitably generating overlapped grid IDs, for example, the ID numbers in the standard interval of the point cloud to be registered are 1, 2, 3, 4 and 5, the ID numbers in the standard interval of the template point cloud are 3, 4, 5, 6 and 7, and then the common grid ID numbers are 3, 4 and 5.
4.5: respectively calculating the average elevation of point clouds in a common grid network of two periods of same-name ground data, and calculating the elevation deviation of the two periods of data according to the difference of the average elevations of the two periods of data;
4.6: and performing final correction according to the elevation deviation of the obtained two-stage data, wherein the correction result of the step is shown in fig. 10.
The fifth step: and performing the second to fifth steps of registration scheme on each segmented data to finally obtain a complete registration result of the two-stage data.
The above is only a specific embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that are not thought of through the inventive work should be included in the scope of the present invention. Therefore, the protection scope of the present invention should be defined by the claims.

Claims (6)

1. A three-dimensional laser point cloud accurate registration method for a multi-phase road is characterized by comprising the following steps:
a first step; acquiring multi-stage point cloud data of a monitored road, selecting any two-stage original point cloud data, and performing equidistant segmentation processing on the original point cloud data to generate a plurality of sections of point cloud data;
the second step is that: and (3) carrying out deviation correction on each segmented point cloud data in a direction vertical to a driving lane:
the third step: carrying out deviation correction on the driving lane direction of each segmented point cloud data;
the fourth step: performing elevation correction on each segmented point cloud data to finally obtain a complete registration result of the two-stage data;
the fifth step: and selecting two periods of original point cloud data, repeating the steps from the first step to the fourth step, and so on to obtain a complete registration result of the multi-period data.
2. The multi-stage road three-dimensional laser point cloud accurate registration method as claimed in claim 1, wherein the second step specifically comprises the following steps:
2.1: aiming at each single section of data, firstly selecting two periods of point clouds of same-name pavement markings, and solving abrupt change points between the markings and the ground points on the single scanning line containing the point clouds of the markings by using a Sigmoid mathematical function model according to abrupt change characteristics of point cloud intensity values on the single scanning line between the ground points and the markings and taking the abrupt change points as marking boundary points;
and 2.2, calculating each centerline point of the reticle according to the solved reticle boundary points, fitting a straight line by using a least square method to serve as the reticle centerline, solving the reticle centerline of the two periods of the same-name reticles, and solving the deviation of the two periods of the reticles in the direction vertical to the road driving direction according to the reticle centerline.
3. The method for accurately registering the multi-phase road three-dimensional laser point cloud as claimed in claim 2, wherein the step 2.1 comprises the following sub-steps:
2.1.1: setting the distance, and performing data segmentation on the point cloud data to be registered and the template point cloud data;
2.1.2: manually and uniformly selecting a plurality of homonymous marked lines of the two-stage point cloud data;
2.1.3: selecting a homonymous marking, and calculating marking boundary points on each scanning line of the two-stage marking by using a Sigmoid function model according to the sudden change characteristics of point cloud intensity values on the scanning lines containing marking point clouds between ground points and marking points, wherein the function model is as follows:
Figure FDA0002332421020000021
wherein, S (x) is the intensity value corresponding to each reticle boundary point;
x is the coordinate value of x, y of each marking boundary point, so as to calculate that each marking boundary point is-b2；
B in the formula1、b3、b4Respectively representing the coefficients of the formula, wherein b1Scale factor representing the function curve in x-axis, b3Scale factor representing the function curve in y-axis, b4Represents the translation of the function curve relative to the origin (0,0) on the y-axis; e is the natural coefficient of the mathematical noun.
4. The method for accurately registering the multi-stage road three-dimensional laser point cloud as claimed in claim 2, wherein the step 2.2 comprises the following sub-steps:
2.2.1: calculating the central point of each marking line according to the boundary points of each marking line obtained by calculation;
2.2.2: fitting the central line of the marked line according to a least square fitting method by using the central point of each marked line:
Figure FDA0002332421020000022
Figure FDA0002332421020000023
wherein x and y are coordinates of the central point of each marked line participating in calculation, i is a coordinate base mark participating in calculation, n is the number of the coordinates participating in calculation,
Figure FDA0002332421020000031
for the slope of the fitted centerline to be,
Figure FDA0002332421020000032
is the fitted centerline intercept;
2.2.3: calculating the distance between the lines in the marked lines with the same name according to a point-to-straight line distance formula, and calculating the offset of the marked line to be registered in each marked line with the same name in the x and y directions compared with the original marked line;
Figure FDA0002332421020000033
Figure FDA0002332421020000034
in the formula, k and d respectively represent the slope and intercept of a linear equation where the midline of the original marked line in a pair of marked lines with the same name is located, and x and y are three-dimensional coordinates of points on the midline of the cloud marked line of the point to be registered;
2.2.4: and (4) performing the 2.2.1-2.2.3 steps of calculation on each selected homonymous marked line, calculating the average value of the translation amount of each homonymous marked line, and translating the cloud point to be registered in the x and y directions to finish the first correction.
5. The multi-stage road three-dimensional laser point cloud accurate registration method as claimed in claim 1, wherein the third step specifically comprises the following steps:
3.1: and (3) interactively selecting multiple same-name independent feature data on two sides of a two-stage road as registration elements:
3.2: respectively registering multiple homonymous independent objects on two sides of the two-period data road by using an ICP (inductively coupled plasma) algorithm, calculating offsets OffsetX and OffsetY in x and y directions, and calculating an average value
Figure FDA0002332421020000035
Figure FDA0002332421020000036
3.3: respectively calculating the errors 6 in the x and y directions according to the calculated offset of the plurality of groups of the same-name independent objectsx，6yAnd will be within range
Figure FDA0002332421020000041
Recalculating the average value of the internal offset to be used as a second correction registration offset;
3.4: and translating the cloud point to be registered in the x and y directions to finish the second correction.
6. The multi-stage road three-dimensional laser point cloud accurate registration method as claimed in claim 1, wherein the fourth step specifically comprises the following steps:
4.1: for two-stage data after data segmentation, ground point cloud data is obtained by using a road filtering algorithm, and a unified grid index is established for the two-stage homonymous road point cloud data, wherein the specific method comprises the following steps:
4.1.1: acquiring a common minimum external rectangle of the point cloud plane projection of the two-stage road surface, namely: xmin, ymin, xmax, ymax;
4.1.2: setting a certain grid step length, and dividing grids for two-stage data:
Figure FDA0002332421020000042
Figure FDA0002332421020000043
ID＝(RID-1)×Col+CID
in the formula: ID is the index number of the grid, Row and Col are the number of rows and columns of the grid, RID and CID are the number of rows and columns, and xi、yiRespectively, the coordinates of the point cloud, and StepX and StepY respectively represent the length and width of the grid;
4.2: calculating a normal vector of each grid point cloud in the two-stage pavement point cloud data by using a Principal Component Analysis (PCA) algorithm; and then calculating the size of an included angle theta between a normal vector of each grid point cloud of the two-stage road surface point cloud data and a normal vector (0,0,1) of a point cloud z axis:
Figure FDA0002332421020000051
Figure FDA0002332421020000052
4.3: respectively carrying out statistical analysis on the point cloud data of the two-stage road surface and determining an invariant region, wherein the method specifically comprises the following steps:
4.3.1: setting the same included angle interval step length, and counting the number of corresponding grids and the corresponding ID number in each interval;
4.3.2: counting and analyzing the upper limit and the lower limit of the interval with the maximum grid number, taking the interval as a standard interval, regarding a point cloud area in the standard interval as a stable area in the two-stage data, namely an area without serious deformation, and recording an ID number corresponding to the grid in the interval;
4.4: and counting the ID numbers of the common grids of the two-stage same-name ground data, dividing the grids under the same standard, and then inevitably generating overlapped grid IDs, respectively calculating the average elevation of the point clouds in the common grids of the two-stage same-name ground data, calculating the elevation deviation of the two-stage data according to the difference of the average elevations, and finally correcting.
CN201911341604.1A 2019-12-24 2019-12-24 Multi-period road three-dimensional laser point cloud accurate registration method Active CN111127530B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911341604.1A CN111127530B (en) 2019-12-24 2019-12-24 Multi-period road three-dimensional laser point cloud accurate registration method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911341604.1A CN111127530B (en) 2019-12-24 2019-12-24 Multi-period road three-dimensional laser point cloud accurate registration method

Publications (2)

Publication Number Publication Date
CN111127530A true CN111127530A (en) 2020-05-08
CN111127530B CN111127530B (en) 2023-06-20

Family

ID=70501445

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911341604.1A Active CN111127530B (en) 2019-12-24 2019-12-24 Multi-period road three-dimensional laser point cloud accurate registration method

Country Status (1)

Country Link
CN (1) CN111127530B (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112465765A (en) * 2020-11-24 2021-03-09 山东科技大学 Road surface depth information model construction method based on vehicle-mounted mobile laser point cloud
CN112785631A (en) * 2020-12-31 2021-05-11 杭州鲁尔物联科技有限公司 Point cloud data registration method based on DLG
CN113034685A (en) * 2021-03-18 2021-06-25 北京百度网讯科技有限公司 Method and device for superposing laser point cloud and high-precision map and electronic equipment
CN113468282A (en) * 2021-05-28 2021-10-01 深圳市跨越新科技有限公司 Construction method and device of freight car site track, terminal and storage medium
CN113763438A (en) * 2020-06-28 2021-12-07 北京京东叁佰陆拾度电子商务有限公司 Point cloud registration method, device, equipment and storage medium

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100209013A1 (en) * 2009-02-13 2010-08-19 Harris Corporation Registration of 3d point cloud data to 2d electro-optical image data
CN103679741A (en) * 2013-12-30 2014-03-26 北京建筑大学 Method for automatically registering cloud data of laser dots based on three-dimensional line characters
CN106204547A (en) * 2016-06-29 2016-12-07 山东科技大学 The method automatically extracting shaft-like atural object locus from Vehicle-borne Laser Scanning point cloud
CN109544607A (en) * 2018-11-24 2019-03-29 上海勘察设计研究院(集团)有限公司 A kind of cloud data registration method based on road mark line
CN109872352A (en) * 2018-12-29 2019-06-11 中国科学院遥感与数字地球研究所 Power-line patrolling LiDAR data autoegistration method based on shaft tower characteristic point
CN109900298A (en) * 2019-03-01 2019-06-18 武汉光庭科技有限公司 A kind of vehicle location calibration method and system

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100209013A1 (en) * 2009-02-13 2010-08-19 Harris Corporation Registration of 3d point cloud data to 2d electro-optical image data
CN103679741A (en) * 2013-12-30 2014-03-26 北京建筑大学 Method for automatically registering cloud data of laser dots based on three-dimensional line characters
CN106204547A (en) * 2016-06-29 2016-12-07 山东科技大学 The method automatically extracting shaft-like atural object locus from Vehicle-borne Laser Scanning point cloud
CN109544607A (en) * 2018-11-24 2019-03-29 上海勘察设计研究院(集团)有限公司 A kind of cloud data registration method based on road mark line
CN109872352A (en) * 2018-12-29 2019-06-11 中国科学院遥感与数字地球研究所 Power-line patrolling LiDAR data autoegistration method based on shaft tower characteristic point
CN109900298A (en) * 2019-03-01 2019-06-18 武汉光庭科技有限公司 A kind of vehicle location calibration method and system

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
何培培;万幼川;杨威;秦家鑫;: "基于线特征的城区激光点云与影像自动配准" *
卢秀山: "一种三维激光点云中建筑物立面渐进分割方法" *
马新江;刘如飞;蔡永宁;王鹏;: "一种基于路缘特征的点云道路边界提取方法" *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113763438A (en) * 2020-06-28 2021-12-07 北京京东叁佰陆拾度电子商务有限公司 Point cloud registration method, device, equipment and storage medium
CN113763438B (en) * 2020-06-28 2024-04-19 北京京东叁佰陆拾度电子商务有限公司 Point cloud registration method, device, equipment and storage medium
CN112465765A (en) * 2020-11-24 2021-03-09 山东科技大学 Road surface depth information model construction method based on vehicle-mounted mobile laser point cloud
CN112785631A (en) * 2020-12-31 2021-05-11 杭州鲁尔物联科技有限公司 Point cloud data registration method based on DLG
CN113034685A (en) * 2021-03-18 2021-06-25 北京百度网讯科技有限公司 Method and device for superposing laser point cloud and high-precision map and electronic equipment
CN113468282A (en) * 2021-05-28 2021-10-01 深圳市跨越新科技有限公司 Construction method and device of freight car site track, terminal and storage medium

Also Published As

Publication number Publication date
CN111127530B (en) 2023-06-20

Similar Documents

Publication Publication Date Title
CN111127530A (en) Accurate registration method for three-dimensional laser point clouds of multi-phase roads
WO2022143796A1 (en) Calibration method and calibration device for line structured light measurement system, and system
US11900634B2 (en) Method for adaptively detecting chessboard sub-pixel level corner points
CN105783779B (en) The real-time form identification of rail profile and distortion calibration method based on three layers of matching
CN104732553B (en) A kind of Feature Points Extraction based on many laser assisted targets
CN110276732B (en) Mountain area point cloud cavity repairing method considering topographic characteristic line elements
CN107578400A (en) A kind of contact net device parameter detection method of BIM and three-dimensional point cloud fusion
CN112906694A (en) Reading correction system and method for inclined pointer instrument image of transformer substation
CN102565810B (en) Method for extracting land utilization landmark boundary outline on remote sensing image
CN113724193B (en) PCBA part size and clearance high-precision visual measurement method
CN113052881A (en) Automatic registration method for extracting pole point in indoor three-dimensional point cloud
CN113327276B (en) Mobile measurement-oriented general mass point cloud data registration method
CN107798698A (en) Structured light strip center extracting method based on gray-level correction and adaptive threshold
CN108613630A (en) The two linear pipe bubbles based on image processing techniques deviate measuring method
CN104574415A (en) Target space positioning method based on single camera
CN109583365A (en) Method for detecting lane lines is fitted based on imaging model constraint non-uniform B-spline curve
CN111145157A (en) Road network data automatic quality inspection method based on high-resolution remote sensing image
CN115018249B (en) Subway station construction quality evaluation method based on laser scanning technology
CN105701776A (en) Lens distortion correcting method and system used for automatic optical detection
CN109115127B (en) Sub-pixel peak point extraction algorithm based on Bezier curve
CN114399677A (en) Pointer instrument identification method based on text region reading
CN110232697B (en) Light spot edge fitting method
CN106407973A (en) Robust recognition method for AR code marked on cylinder
CN116503462A (en) Method and system for quickly extracting circle center of circular spot
CN114612412A (en) Processing method of three-dimensional point cloud data, application of processing method, electronic 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