CN101387700B - Data fusing method and system based on multi-laser scanner - Google Patents
Data fusing method and system based on multi-laser scanner Download PDFInfo
- Publication number
- CN101387700B CN101387700B CN2008101713176A CN200810171317A CN101387700B CN 101387700 B CN101387700 B CN 101387700B CN 2008101713176 A CN2008101713176 A CN 2008101713176A CN 200810171317 A CN200810171317 A CN 200810171317A CN 101387700 B CN101387700 B CN 101387700B
- Authority
- CN
- China
- Prior art keywords
- cluster
- combination
- directed edge
- center
- axle
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Length Measuring Devices By Optical Means (AREA)
Abstract
The invention discloses a data fusion method based on multiple laser scanners and a system. The data fusion method comprises: predefining a plane profile model for a mobile target; for each scan data frame, clustering the laser points of each data frame; in all clusters attained by different laser scanners, and for each cluster group which mutual distance is in a preset value, all direction vectors of the cluster group are matched with the plane profile model of the mobile target, combining the cluster group into a mobile target. The invention utilizes the character that the direction vectors of the laser points measured by the laser scanners at any part and direction in one plane are matched with the plane profile model of the mobile target, to fuse the scanned data to improve the data fusion accuracy of multiple laser scanners.
Description
Technical field
The present invention relates to the laser scanner technique field, be specifically related to data fusion method and system based on multi-laser scanner.
Background technology
Laser scanning and ranging is in being usually used in the traffic data collection system.The principle of laser ranging is: the laser head emission of lasering beam, when laser beam runs into barrier, reflected by barrier; Laser head receives the laser beam that is reflected, and by calculating from being transmitted into the mistiming of reception, and multiply by the light velocity, again divided by 2, is the distance of laser head to barrier.The principle of laser scanning and ranging is: the eyeglass of a high speed rotating is arranged in the laser scanner, and laser head is got to laser beam on the eyeglass earlier, reflects away again; By the rotation of eyeglass, can adjust the angle of laser beam; By the angle of laser beam, distance value can be converted into coordinate points.By adjusting the rotational speed and the scope of eyeglass, can control scope, sampling density and the frame per second of laser scanning.
Just eyeglass is around the rotation of axle for the line sweep laser scanner, and sense of rotation is fixed, and for turning clockwise, or is rotated counterclockwise.The distance value that measures can be exchanged into 2 dimension coordinates point, is called for short laser spots.Laser scanning can obtain a series of laser spots, is generally 1 frame.The parameter of laser scanner can be as follows: sweep limit: 180 degree, sampling density: 0.5 degree/point, 37Hz (per second 37 frames), maximum measure distance scope 60m.
When laser scanner is applied in the traffic data collection system, several laser scanners can be arranged at the roadside, horizontal scanning is arranged on laser scanner usually on the about 40cm height of level face of road surface and just can covers the object measured zone.As shown in Figure 1, two laser scanners 1,2 are placed on the top of a crossing respectively, at each constantly, laser scanner 1,2 is with the moving target that scans: automobile, bicycle, people's laser spots sends to server, in Fig. 1, the laser spots that solid black point measures for laser scanner 1, the laser spots that black dessert point measures for laser scanner 2.As can be seen, owing to reason such as blocking, the pedestrian can only be scanned by laser scanner 2 sometime, and bicycle can only be scanned by laser scanner 1.
A laser scanner can only observe or part side of moving target, the fusion of the scan-data by a plurality of laser scanners, can measure moving target around, thereby can improve the precision that object identification and motion state are analyzed effectively.The data of different laser scanner measurements are sent to server, server comes out different laser scanners at the data pick-up that synchronization measures, and integration carries out data fusion in global coordinate system, and such fused data can measure the face profile of moving target; Simultaneously, the about 30Hz of scanning frame per second can catch moving target fast.
Fig. 2 has provided the flow process of existing Data Fusion based on multi-laser scanner, and as shown in Figure 2, its concrete steps are as follows:
Step 201: every frame scan data are carried out the background difference.
Step 202: the laser spots that obtains through the background subtraction branch in every frame scan data is carried out cluster.
Step 203: each cluster is carried out the KL conversion respectively, and the distribution of determining the laser dot-matrix in this cluster is point (0), line (1) or face (2).
Fig. 3 has provided 0,1,2 laser dot-matrix distribution schematic diagram.
Step 204: the laser dot-matrix according to each cluster distributes, and each cluster is mated with the rectangle profile respectively.
Fig. 3 is existing three kinds of laser dot-matrix distribution schematic diagrams, and as shown in Figure 3, laser dot-matrix is distributed as 0,1,2 all available rectangle outline of cluster.
Step 205: according to nearest neighbour method, will gather apart from the rectangle profile in preset value is a class, as a moving target.
But in practice, along with the variation of complicated, the measuring object of measurement environment, the fusion treatment difficulty of data increases thereupon.Such as; owing to block; the reflection characteristic of Facing material; reasons such as clock synchronization error; the measurement data of same object can be divided into several; non-overlapping copies; like this; same moving target will be judged into a plurality of moving targets; Fig. 4 is the exemplary plot of existing Data Fusion based on multi-laser scanner; as shown in Figure 4; laser scanner 1; 2; the cluster of 3 certain moving target that scans is respectively shown in the left figure among Fig. 4; Ideal Match result is combined as a moving target with all clusters among the figure; but; according to nearest neighbour method; usually can obtain as the combined result among the right figure among Fig. 4, promptly a moving target is judged to be broken into 3 moving targets.In addition, when making that when different moving target close proximity measurement data is overlapping, then they can be used as same moving target.As seen, both of these case all can cause the mistake combination of moving target.
Summary of the invention
The invention provides a kind of data fusion method and system, to improve the degree of accuracy of data fusion based on multi-laser scanner.
Technical scheme of the present invention is achieved in that
A kind of data fusion method based on multi-laser scanner, the face profile model of pre-defined moving target, this method comprises:
For every frame scan data, the laser spots in these frame data is carried out cluster;
In all clusters that different laser scanner obtained, for the every group cluster of phase mutual edge distance in preset value, if all direction vectors in this group cluster and the face profile Model Matching of moving target then should be organized cluster and be combined as a moving target.
The face profile model of described moving target is an oriented polygon, polygonal each bar limit is a directed edge, the direction vector on each bar limit constitutes counterclockwise or a turn clockwise closed hoop, and, if axle of extracting out from cluster and oriented polygonal directed edge coupling are then with this support vector as this directed edge.
Described for the every group cluster of phase mutual edge distance in preset value, if this group in cluster all direction vectors and the face profile Model Matching of moving target, then should organize cluster and be combined as a moving target and comprise:
In all clusters that different laser scanner obtained, select a cluster that does not add any combination, in the distance with this cluster is combination in the preset value, select a combination, each the bar axle that to from this cluster, extract out respectively with each directed edge coupling of selected combination, if every axle extracting out from this cluster all mates with a directed edge respectively, then this cluster is joined in this combination.
Described each the bar axle that will extract out from this cluster with each directed edge coupling of selected combination is respectively: for every axle extracting out from this cluster, calculate this and the differential seat angle of each bar directed edge of selected combination respectively, if with the differential seat angle of a directed edge less than preset value, think that then this axle and this directed edge mate on.
When every axle from this cluster, extracting out all respectively with directed edge coupling, and the major axis of extracting out from cluster is during greater than the length of side of the longest directed edge of combination, described this cluster is joined in this combination comprises:
The major axis that to extract out from described cluster should rotate according to preset direction on the base limit, respectively each directed edge of other that is newly made up as the basic limit of new combination;
All laser spots in described cluster and the described combination are projected on each directed edge of new combination, the length of side of each directed edge that is newly made up, the center of calculating all laser spots in described cluster and the described combination, this center is the center of new combination;
In all direction vectors in the cluster in described cluster and described combination, the support vector of each directed edge of the new combination of search if new combination has a pair of adjacent directed edge that support vector is all arranged, determines that then this adjacent directed edge has angle point;
Upgrade described combination with new combination.
When every axle from this cluster, extracting out all respectively with directed edge coupling, and the length of every axle extracting out from cluster is when all being not more than the length of side of the longest directed edge of combination, described this cluster is joined in this combination comprises:
All laser spots in described cluster and the described combination are projected on each directed edge of described combination, obtain the new length of side of each directed edge of described combination, calculate the center of all laser spots in described cluster and the described combination, this center is the new center of described combination;
If find the axial projection from described cluster, extract out to directed edge be adjacent the limit and become and all have support vector by not all having support vector, then determine and should produce angle point adjacent directed edge.
When any one axle of extracting out from described cluster was not gone up with any directed edge coupling, described method further comprised: for this cluster itself generates a combination.
Described is that combination of this cluster generation itself comprises:
The major axis that to extract out from described cluster should rotate according to preset direction on the base limit, respectively each directed edge of other that is newly made up as the basic limit of the combination that will generate;
All laser spots in the described cluster are projected on each directed edge that will generate combination, obtain generating the length of side of each directed edge of combination, the center of calculating all laser spots in the described cluster, this center is the center that will generate combination;
In all direction vectors in described cluster, search will generate the support vector of each directed edge of combination, if there is a pair of adjacent directed edge that support vector is all arranged, determines that then this adjacent directed edge has angle point.
Describedly after being joined in this combination, this cluster further comprises: judge whether described cluster is joined the new combination that is generated in the described combination credible, if replace described combination with this new combination; Otherwise, discarded should newly the combination.
The face profile model of described moving target is an oriented polygon,
Described judgement joins the new combination that generated in the described combination credible comprising whether with described cluster: for each laser spots in all clusters in the new combination, determine the corresponding sides of axle in new combination in the cluster at this laser spots place, calculate the projector distance of this laser spots to these corresponding sides, calculate the variance of the distance of all laser spots correspondences, if it is insincere that this variance yields less than default variance, is then determined newly to make up; Otherwise, determine newly to make up credible.
A kind of data fusion system based on multi-laser scanner, this system comprises:
The cluster module for every frame scan data, is carried out cluster to the laser spots in these frame data, and cluster result is outputed to data fusion module;
Data fusion module, in all clusters that different laser scanner obtained, for the every group cluster of phase mutual edge distance in preset value, if all direction vectors in this group cluster and the face profile Model Matching of predefined moving target then should be organized cluster and be combined as a moving target.
Described data fusion module comprises:
Matching module, in all clusters, select a cluster that does not add any combination, in the distance with this cluster is combination in the preset value, select a combination, each the bar axle that to from this cluster, extract out respectively with each directed edge coupling of selected combination, if every axle extracting out all mates with a directed edge respectively, then this cluster sign and combination sign are sent to Fusion Module from this cluster;
Fusion Module receives cluster sign and combination sign that matching module is sent, and the corresponding cluster of this cluster sign is joined in the corresponding combination of this combination sign.
Described Fusion Module comprises:
The limit determination module, the major axis that will extract out from described cluster should rotate according to preset direction on the base limit, respectively each directed edge of other that is newly made up as the basic limit of new combination;
The length of side and center determination module, all laser spots in described cluster and the described combination are projected on each directed edge of new combination, the length of side of each directed edge that is newly made up, the center of calculating all laser spots in described cluster and the described combination, this center is the center of new combination;
The angle point determination module, in all direction vectors in the cluster in described cluster and described combination, the support vector of each directed edge of the new combination of search if there is a pair of adjacent directed edge that support vector is all arranged, determines that then this adjacent directed edge has angle point.
Described matching module further comprises a module, this module is used for after finding have any axle of extracting out from cluster not mate with any directed edge, this cluster sign is sent to Fusion Module, is that this cluster itself generates a combination with the indication Fusion Module.
Described Fusion Module comprises:
The limit determination module, the major axis that will extract out from described cluster should rotate according to preset direction on the base limit, respectively each directed edge of other that is newly made up as the basic limit of the combination that will generate;
The length of side and center determination module, all laser spots in the described cluster are projected on each directed edge that will generate combination, obtain generating the length of side of each directed edge of combination, the center of calculating all laser spots in the described cluster, this center is the center that will generate combination;
The angle point determination module, in all direction vectors in described cluster, search will generate the support vector of each directed edge of combination, if there is a pair of adjacent directed edge that support vector is all arranged, determines that then this adjacent directed edge has angle point.
Described data fusion module further comprises a module, and this module is used for judging whether described cluster is joined the new combination that described combination generates credible, if replace described combination with this new combination; Otherwise, discarded should newly the combination.
Compared with prior art, the face profile model of pre-defined moving target among the present invention; For every frame scan data, the laser spots in these frame data is carried out cluster; In all clusters that different laser scanner obtained, for the every group cluster of phase mutual edge distance in preset value, if all direction vectors in this group cluster and the face profile Model Matching of moving target then should be organized cluster and be combined as a moving target.The present invention by utilize anywhere, direction are placed in same plane laser scanner measurement to the direction vector of laser spots and the characteristic that is complementary of the face profile model of moving target, scan-data is merged, improved the degree of accuracy of the data fusion under the multi-laser scanner.
Description of drawings
Fig. 1 is applied to synoptic diagram in the traffic data collection system for laser scanner;
Fig. 2 is the process flow diagram of existing Data Fusion based on multi-laser scanner;
Fig. 3 is existing three kinds of laser dot-matrix distribution schematic diagrams;
Fig. 4 is the exemplary plot of existing Data Fusion based on multi-laser scanner;
Three kinds of laser dot-matrix distribution schematic diagrams that Fig. 5 provides for the embodiment of the invention;
The direction vector characteristic exemplary plot of all clusters of the moving target that Fig. 6 provides for the embodiment of the invention;
The processing flow chart that Fig. 7 provides for the embodiment of the invention based on the data fusion of multi-laser scanner;
The process flow diagram that one or more cluster is generated combination that Fig. 8 provides for the embodiment of the invention;
The flow process that a cluster is added to a new combination of generation in the existing combination that Fig. 9 provides for the embodiment of the invention;
Whether the definite newly-generated combination that Figure 10 provides for the embodiment of the invention believable process flow diagram;
The composition diagram that Figure 11 provides for the embodiment of the invention based on the data fusion system of multi-laser scanner.
Embodiment
By three kinds of laser dot-matrixes are distributed: observe and can find for 2,1,0: the laser spots that measures is sequential, three kinds of laser dot-matrix distribution schematic diagrams that Fig. 5 provides for the embodiment of the invention, as shown in Figure 5:
For 2, the laser spots that s measures at first for this cluster; The laser spots that e measures at last for this cluster; C is the flex point of the laser spots in this cluster.Definition: axle 1:u1 is the direction vector of e → c, and len1 is the length of side of ec; Axle 2:u2 is the direction vector of c → s, and len2 is the length of side of cs.
For 1, the laser spots that s measures at first for this cluster; The laser spots that e measures at last for this cluster.Definition: axle 1:u1 is the direction vector of e → s, and len1 is the length of side of es.
For 0, p is a central point, and len1 is that es is long.
Can find the cluster analysis that belongs to same moving target: the direction vector of all clusters has constituted a counterclockwise ring again, shown in the direction vector characteristic exemplary plot of all clusters of a moving target that provides as the embodiment of the invention among Fig. 6.
Can know by inference: as many laser scanners during same targeted scans, if target has the individual side of n (n is a positive integer), then when n side of this target all then by laser scanner scans, all laser scanner scans to direction vector can constitute the profile of an oriented closure, this closed profile is the face profile of target; And have only the part side then when this target by laser scanner scans, then laser scanner scans to each direction vector respectively with the face profile of target in a directed edge coupling.
Characteristic according to the cluster of above-mentioned moving target, among the present invention, the face profile model of a pre-defined moving target, for every frame scan data, laser spots in these frame scan data is carried out cluster, for the every group cluster of phase mutual edge distance, if the direction vector of all clusters and the face profile Model Matching of moving target then should be organized cluster and be combined as a moving target less than preset value.
In practice, the face profile model of moving target can be defined as an oriented n limit shape (n is a positive integer), and each bar limit of n limit shape is a directed edge, and the direction vector on each bar limit constitutes counterclockwise or a turn clockwise closed hoop; And, if certain the bar directed edge coupling in the face profile model of certain bar axle of a cluster and moving target then claims this to be the support vector of this directed edge; If for a pair of adjacent directed edge of oriented n limit shape, this all has support vector respectively to adjacent directed edge, can determine that then this has angle point to adjacent directed edge.Below be that rectangle (n=4) is an example with the face profile model of moving target, the present invention is described in detail.When n gets other value, can be achieved with reference to the embodiment of n=4.
The present invention is further described in more detail below in conjunction with drawings and the specific embodiments.
The processing flow chart that Fig. 7 provides for the embodiment of the invention based on the data fusion of multi-laser scanner, as shown in Figure 7, its concrete steps are as follows:
Step 701: for every frame scan data, the laser spots in these frame data is carried out cluster, then each cluster is carried out the KL conversion, the distribution of determining the laser dot-matrix in this cluster is point (0), line (1) or face (2).
Step 702: the laser dot-matrix according to each cluster distributes, and each cluster is mated with the rectangle profile respectively.
Step 701,702 can adopt existing techniques in realizing.
Step 703: in all clusters, seek a cluster ci who is not combined.
I is a positive integer, the sequence number of expression cluster.
Step 704: in the neighbour of ci, find a combination gj.
J is a positive integer, the sequence number of expression combination.
Here, can preestablish a diameter d, in this step, in the diameter d scope of ci, according to the distance of ci from the close-by examples to those far off, seek combination gj successively.The value of d can be determined by experience, also can perhaps be determined jointly by the two by a large amount of sample learnings is determined.
Step 705: ci is added among the combination gj, generates a new combination gj '.
Step 706: judge whether gj ' is credible, if, execution in step 710; Otherwise, execution in step 707.
Step 707: discarded gj '.
Step 708: judge whether that all combinations among the neighbour of ci are all searched out, if, execution in step 709; Otherwise, return step 704.
Step 709:, go to step 711 for ci generates a new combination gk.
K is a positive integer, the sequence number of expression combination.
Step 710: use gj ' to upgrade gj.
Step 711: judge whether that all clusters all were combined, if this flow process finishes; Otherwise, return step 703.
A moving target is promptly represented in each combination.
Having mentioned in above-mentioned steps 709 is the processing that a cluster ci generates a combination gk, and Fig. 8 has provided the process flow diagram that one or more cluster is generated combination that the embodiment of the invention provides, and as shown in Figure 8, its concrete steps are as follows:
Step 801: extract major axis in all clusters that will make up out, defining this major axis is v1, and v1 according to being rotated counterclockwise 90 degree, 180 degree, 270 degree, is obtained v2, v3, v4 respectively.
The processing of extracting an axle from cluster out can be by existing techniques in realizing.
Generally speaking, when the face profile of moving target is n limit shape, then with v1 according to clockwise or order rotation counterclockwise (
) degree, wherein, s is a positive integer, and gets n-1 successively from 1, a directed edge v (s+1) of the corresponding n of each s limit shape, and wherein, v1 is the basic limit of n limit shape, has also claimed initial line.
Step 802: the rectangular coordinate that all laser spots in all clusters is projected to v1, v2 formation is fastened, and establishing the projected length of laser spots on v1 is len1, and the projected length on v2 is len2, determines the central point p of all laser spots.
The coordinate of all laser spots on v1, v2 averaged respectively, can obtain central point p.
Step 803: each direction vector um in all clusters (m is a positive integer) is mated with v1, v2, v3, v4 respectively, obtain the support vector of v1, v2, v3, v4.
If um and vn (n=1,2,3,4) coupling is arranged, claims that then um is the support vector of vn, puts V_valn=true.
Step 804: if a pair of adjacent edge vx, vy are arranged, all there is support vector on each limit, then put C_val (x, y)=true, and with this angle point to adjacent edge give c (x, y).Wherein, x, y=1,2,3,4.
So far, combination gk generates, and the four edges of gk is respectively v1, v2, v3, v4, wherein, len1 is the length of side of v1, v3, len2 is the length of side of v2, v4, and p is the central point of gk, and, the vn of V_valn=true has support vector, C_val (x, y)=adjacent edge vx, the vy of True have angle point c (x, y).
Flow process shown in Figure 8 is not only applicable to a cluster is generated the situation of a combination, is applicable to the situation that two above clusters is generated a combination yet.
Mentioned in the step 705 of flow process shown in Figure 7 cluster ci has been added to the processing that generates a new combination gj ' among the combination gj, Fig. 9 has provided the flow process that a cluster is added to a new combination of generation in the existing combination that the embodiment of the invention provides, as shown in Figure 9, its concrete steps are as follows:
Step 901: from the cluster ci that will be added to combination gj, select an axle um successively.
M is the sequence number of the axle among the cluster ci.
Step 902: v1, v2, v3, the v4 of um and gj are compared respectively.
Step 903: the differential seat angle that judges whether vn (n=1,2,3 or 4) and um in predetermined threshold value, if, execution in step 904; Otherwise, execution in step 914.
Step 904: with the support vector of um as vn, V_valn=True.
Vn in this step promptly, with the vn of differential seat angle in predetermined threshold value of um.If the differential seat angle that two above vn (n=1,2,3 or 4) and um are arranged in predetermined threshold value, is then selected and the vn of the differential seat angle minimum of um, um is the support vector of selected vn.
Step 905: judge whether all selected mistake of all axles of cluster ci, if, execution in step 906; Otherwise, return step 901.
Step 906: in all um of the support vector on the limit that becomes gj, select a umax the longest.
Step 907: whether the length of judging umax greater than the maximal side of gj, if, execution in step 908; Otherwise, execution in step 912.
Step 908: umax as v1 ', according to being rotated counterclockwise 90 degree, 180 degree, 270 degree, is obtained v2 ', v3 ', v4 ' with v1 ' respectively.
Step 909: the rectangular coordinate that all laser spots in all clusters (comprising the cluster among cluster ci and the gj) is projected to v1 ', v2 ' formation is fastened, if the projected length of laser spots on v1 ' is len1 ', projected length on v2 ' is len2 ', determines the central point p ' of all laser spots.
Step 910: each direction vector uq in all clusters (comprising the cluster among cluster ci and the gj) (q is a positive integer) is mated with v1 ', v2 ', v3 ', v4 ' respectively.
If uq and vn ' (n=1,2,3 or 4) coupling is arranged, claims that then uq is the support vector of vn ', puts V_valn '=true.
Step 911: if a pair of adjacent edge vx ', vy ' are arranged, all there is support vector on each limit, then puts C_val (x ', y ')=True, and gives c (x ', y ') with this angle point to adjacent edge, and this flow process finishes.
So far newly made up gj ', the four edges of gj ' is respectively v1 ', v2 ', v3 ', v4 ', wherein, len1 ' is the length of side of v1 ', v3 ', and len2 ' is the length of side of v2 ', v4 ', p ' is the central point of gj ', and the vn ' of V_valn '=true has support vector, C_val (x ', y ')=adjacent edge vx ', the vy ' of True have angle point c (x ', y ').
Step 912: whether have the situation of support vector according to the adjacent edge of vn, upgrade angle point c and C_val that vn is adjacent the limit.
Step 913: the rectangular coordinate that all laser spots among cluster ci and the combination gj is projected to v1, v2 formation is fastened, and obtains the new length of side len1 ' of v1, the new length of side len2 ' of v2, and the new central point p ' of definite all laser spots, and this flow process finishes.
So far, combination gj ' produces, and gj ' is that with the different of gi len1 ' is the length of side of v1, v3, and len2 ' is the length of side of v2, v4, and p ' is the central point of gj ', simultaneously, increases new support vector um, V_valn=True according to step 906:vn; According to step 912, vn is adjacent the limit may have angle point.
Step 914: determine that ci can't be added among the combination gj.
After executing this step 914, return step 708.
Mentioned in the step 706 of flow process shown in Figure 7 and judged that whether believable processing of newly-generated combination gj ', Figure 10 have provided whether believable flow process of definite newly-generated combination that the embodiment of the invention provides, as shown in figure 10, its concrete steps are as follows:
Step 1001: whether each the length (len1 or len1, len2) of judging newly-generated combination gj ' all less than preset length, if, execution in step 1002; Otherwise, execution in step 1006.
For example: if scanning circumstance is traffic intersection, then the length of len1, len2 is inevitable wide less than single track, if having among len1, the len2 one wide greater than single track, can determine that then newly-generated combination is insincere.
Step 1002: for each the laser spots ak in all clusters among the gj ', determine that the axle um corresponding limit among gj ' in the cluster at ak place is vn ', calculating ak project on the vn ' apart from d (k).
Wherein, k is the sequence number of the laser spots in all clusters among the gj '.
Step 1003: the variance Vard that calculates all { d (k) }.
Step 1004: whether judge Vard less than default variance, if, execution in step 1005; Otherwise, execution in step 1006.
Default variance rule of thumb or by great amount of samples study obtains usually, and for example: default variance can be 0.25.
Step 1005: determine that this newly-generated combination gj ' is credible, this flow process finishes.
Step 1006: determine that this newly-generated combination gj ' is insincere.
The composition diagram that Figure 11 provides for the embodiment of the invention based on the data fusion system of multi-laser scanner, as shown in figure 11, it mainly comprises:
Cluster module 111: for every frame scan data, the laser spots in these frame data is carried out cluster, cluster result is outputed to data fusion module 112.
Data fusion module 112: receive the cluster result that cluster module 111 is sent, in all clusters that different laser scanner obtained, for the every group cluster of phase mutual edge distance in preset value, if all direction vectors in this group cluster and the face profile Model Matching of predefined moving target then should be organized cluster and be combined as a moving target.
Matching module: in all clusters, select a cluster that does not add any combination, in the distance with this cluster is combination in the preset value, select a combination, each the bar axle that to from this cluster, extract out respectively with each directed edge coupling of selected combination, if every axle extracting out all mates with a directed edge respectively, then this cluster sign and combination sign are sent to Fusion Module from this cluster; If there is any axle of from cluster, extracting out not mate with any directed edge, then this cluster sign is sent to Fusion Module, be that this cluster itself generates a combination with the indication Fusion Module.
Fusion Module: receive cluster sign and combination sign that matching module is sent, the corresponding cluster of this cluster sign is joined in the corresponding combination of this combination sign; Receive the cluster sign that matching module is sent, the cluster that this cluster sign is corresponding generates a combination.
In actual applications, Fusion Module can comprise: limit determination module, the length of side and center determination module and angle point determination module, wherein:
The limit determination module: receive the cluster sign that matching module is sent, the major axis that will extract out from the corresponding cluster of this cluster sign should rotate according to preset direction on the base limit, respectively each directed edge of other that is newly made up as the basic limit of new combination.
The length of side and center determination module: when receiving cluster sign that matching module sends and combination sign, all laser spots that this cluster is identified in corresponding cluster and the corresponding combination of this combination sign project on each directed edge of new combination, the length of side of each directed edge that is newly made up, calculate the center of all laser spots in described cluster and the described combination, this center is the center of new combination; When receiving the cluster sign that matching module sends, all laser spots that this cluster is identified in the corresponding cluster project on each directed edge that will generate combination, obtain to generate the length of side of each directed edge of combination, calculate the center of all laser spots in the described cluster, this center is the center that will generate combination.
Angle point determination module: when receiving cluster sign that matching module sends and combination sign, in this cluster identifies all direction vectors in the cluster in the corresponding combination of corresponding cluster and this combination sign, the support vector of each directed edge of the new combination of search, if there is a pair of adjacent directed edge that support vector is all arranged, determine that then this adjacent directed edge has angle point; When receiving the cluster sign that matching module sends, in this cluster identifies all direction vectors in the corresponding cluster, search will generate the support vector of each directed edge of combination, if there is a pair of adjacent directed edge that support vector is all arranged, determines that then this adjacent directed edge has angle point.
The above only is process of the present invention and method embodiment, in order to restriction the present invention, all any modifications of being made within the spirit and principles in the present invention, is not equal to replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (15)
1. the data fusion method based on multi-laser scanner is characterized in that, the face profile model of pre-defined moving target, and this method comprises:
For every frame scan data, the laser spots in these frame data is carried out cluster;
In all clusters that different laser scanner obtained, for the every group cluster of phase mutual edge distance in preset value, if all direction vectors in this group cluster and the face profile Model Matching of moving target then should be organized cluster and be combined as a moving target;
The face profile model of described moving target is an oriented polygon, polygonal each bar limit is a directed edge, the direction vector on each bar limit constitutes counterclockwise or a turn clockwise closed hoop, and, if axle of extracting out from cluster and oriented polygonal directed edge coupling are then with this support vector as this directed edge.
2. the method for claim 1, it is characterized in that, described for the every group cluster of phase mutual edge distance in preset value, if this group in cluster all direction vectors and the face profile Model Matching of moving target, then should organize cluster and be combined as a moving target and comprise:
In all clusters that different laser scanner obtained, select a cluster that does not add any combination, in the distance with this cluster is combination in the preset value, select a combination, each the bar axle that to from this cluster, extract out respectively with each directed edge coupling of selected combination, if every axle extracting out from this cluster all mates with a directed edge respectively, then this cluster is joined in this combination.
3. method as claimed in claim 2, it is characterized in that, described each the bar axle that will extract out from this cluster with each directed edge coupling of selected combination is respectively: for every axle extracting out from this cluster, calculate this and the differential seat angle of each bar directed edge of selected combination respectively, if with the differential seat angle of a directed edge less than preset value, think that then this axle and this directed edge mate on.
4. method as claimed in claim 2, it is characterized in that, when every axle from this cluster, extracting out all respectively with directed edge coupling, and the major axis of extracting out from cluster is during greater than the length of side of the longest directed edge of combination, described this cluster is joined in this combination comprises:
The major axis that to extract out from described cluster should rotate according to preset direction on the base limit, respectively each directed edge of other that is newly made up as the basic limit of new combination;
All laser spots in described cluster and the described combination are projected on each directed edge of new combination, the length of side of each directed edge that is newly made up, the center of calculating all laser spots in described cluster and the described combination, this center is the center of new combination;
In all direction vectors in the cluster in described cluster and described combination, the support vector of each directed edge of the new combination of search if new combination has a pair of adjacent directed edge that support vector is all arranged, determines that then this adjacent directed edge has angle point;
Upgrade described combination with new combination.
5. method as claimed in claim 2, it is characterized in that, when every axle from this cluster, extracting out all respectively with directed edge coupling, and the length of every axle extracting out from cluster is when all being not more than the length of side of the longest directed edge of combination, described this cluster is joined in this combination comprises:
All laser spots in described cluster and the described combination are projected on each directed edge of described combination, obtain the new length of side of each directed edge of described combination, calculate the center of all laser spots in described cluster and the described combination, this center is the new center of described combination;
If find the axial projection from described cluster, extract out to directed edge be adjacent the limit and become and all have support vector by not all having support vector, then determine and should produce angle point adjacent directed edge.
6. method as claimed in claim 2 is characterized in that, when any one axle of extracting out from described cluster was not gone up with any directed edge coupling, described method further comprised: for this cluster itself generates a combination.
7. method as claimed in claim 6 is characterized in that, described is that combination of this cluster generation itself comprises:
The major axis that to extract out from described cluster should rotate according to preset direction on the base limit, respectively each directed edge of other that is newly made up as the basic limit of the combination that will generate;
All laser spots in the described cluster are projected on each directed edge that will generate combination, obtain to generate the length of side of each directed edge of combination; Calculate the center of all laser spots in the described cluster, this center is the center that will generate combination;
In all direction vectors in described cluster, search will generate the support vector of each directed edge of combination, if there is a pair of adjacent directed edge that support vector is all arranged, determines that then this adjacent directed edge has angle point.
8. the method for claim 1 is characterized in that, describedly further comprises after this cluster is joined in this combination: judge whether described cluster is joined the new combination that is generated in the described combination credible, if replace described combination with this new combination; Otherwise, discarded should newly the combination.
9. method as claimed in claim 8 is characterized in that, the face profile model of described moving target is an oriented polygon,
Described judgement joins the new combination that generated in the described combination credible comprising whether with described cluster: for each laser spots in all clusters in the new combination, determine the corresponding sides of axle in new combination in the cluster at this laser spots place, calculate the projector distance of this laser spots to these corresponding sides, calculate the variance of the distance of all laser spots correspondences, if it is insincere that this variance yields less than default variance, is then determined newly to make up; Otherwise, determine newly to make up credible.
10. the data fusion system based on multi-laser scanner is characterized in that, this system comprises:
The cluster module for every frame scan data, is carried out cluster to the laser spots in these frame data, and cluster result is outputed to data fusion module;
Data fusion module, in all clusters that different laser scanner obtained, for the every group cluster of phase mutual edge distance in preset value, if all direction vectors in this group cluster and the face profile Model Matching of predefined moving target, then should organize cluster and be combined as a moving target, the face profile model of described moving target is an oriented polygon, polygonal each bar limit is a directed edge, the direction vector on each bar limit constitutes counterclockwise or a turn clockwise closed hoop, and, if axle of extracting out from cluster and oriented polygonal directed edge coupling are then with this support vector as this directed edge.
11. system as claimed in claim 10 is characterized in that, described data fusion module comprises:
Matching module, in all clusters, select a cluster that does not add any combination, in the distance with this cluster is combination in the preset value, select a combination, each the bar axle that to from this cluster, extract out respectively with each directed edge coupling of selected combination, if every axle extracting out all mates with a directed edge respectively, then this cluster sign and combination sign are sent to Fusion Module from this cluster;
Fusion Module receives cluster sign and combination sign that matching module is sent, and the corresponding cluster of this cluster sign is joined in the corresponding combination of this combination sign.
12. system as claimed in claim 11 is characterized in that, described Fusion Module comprises:
The limit determination module, the major axis that will extract out from described cluster should rotate according to preset direction on the base limit, respectively each directed edge of other that is newly made up as the basic limit of new combination;
The length of side and center determination module, all laser spots in described cluster and the described combination are projected on each directed edge of new combination, the length of side of each directed edge that is newly made up, the center of calculating all laser spots in described cluster and the described combination, this center is the center of new combination;
The angle point determination module, in all direction vectors in the cluster in described cluster and described combination, the support vector of each directed edge of the new combination of search if there is a pair of adjacent directed edge that support vector is all arranged, determines that then this adjacent directed edge has angle point.
13. system as claimed in claim 11, it is characterized in that, described matching module further comprises a module, this module is used for after finding have any axle of extracting out from cluster not mate with any directed edge, this cluster sign is sent to Fusion Module, is that this cluster itself generates a combination with the indication Fusion Module.
14. system as claimed in claim 13 is characterized in that, described Fusion Module comprises:
The limit determination module, the major axis that will extract out from described cluster should rotate according to preset direction on the base limit, respectively each directed edge of other that is newly made up as the basic limit of the combination that will generate;
The length of side and center determination module, all laser spots in the described cluster are projected on each directed edge that will generate combination, obtain generating the length of side of each directed edge of combination, the center of calculating all laser spots in the described cluster, this center is the center that will generate combination;
The angle point determination module, in all direction vectors in described cluster, search will generate the support vector of each directed edge of combination, if there is a pair of adjacent directed edge that support vector is all arranged, determines that then this adjacent directed edge has angle point.
15. system as claimed in claim 10, it is characterized in that described data fusion module further comprises a module, this module is used for judging whether described cluster is joined the new combination that described combination generates credible, if replace described combination with this new combination; Otherwise, discarded should newly the combination.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2008101713176A CN101387700B (en) | 2008-10-12 | 2008-10-12 | Data fusing method and system based on multi-laser scanner |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2008101713176A CN101387700B (en) | 2008-10-12 | 2008-10-12 | Data fusing method and system based on multi-laser scanner |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101387700A CN101387700A (en) | 2009-03-18 |
| CN101387700B true CN101387700B (en) | 2011-09-21 |
Family
ID=40477228
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2008101713176A Expired - Fee Related CN101387700B (en) | 2008-10-12 | 2008-10-12 | Data fusing method and system based on multi-laser scanner |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101387700B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101853575A (en) * | 2010-05-31 | 2010-10-06 | 北京航空航天大学 | Road junction traffic conflict detection and safety evaluation method based on two-dimensional laser scanners |
| CN102980512A (en) * | 2012-08-29 | 2013-03-20 | 武汉武大卓越科技有限责任公司 | Fixed type automatic volume measurement system and measuring method thereof |
| CN109753981B (en) * | 2017-11-06 | 2020-12-22 | 彼乐智慧科技（北京）有限公司 | Image recognition method and device |
| CN108596117B (en) * | 2018-04-28 | 2021-03-02 | 河北工业大学 | Scene monitoring method based on two-dimensional laser range finder array |
| JP7205180B2 (en) * | 2018-11-19 | 2023-01-17 | 株式会社Ihi | Monitoring system and processing equipment |
| JP7205179B2 (en) * | 2018-11-19 | 2023-01-17 | 株式会社Ihi | Monitoring system |
| JP7205181B2 (en) * | 2018-11-19 | 2023-01-17 | 株式会社Ihi | Monitoring system and processing equipment |
| CN109613553B (en) * | 2018-12-18 | 2021-05-18 | 歌尔光学科技有限公司 | Method, device and system for determining number of objects in scene based on laser radar |
| CN113534095B (en) * | 2021-06-18 | 2024-05-07 | 北京电子工程总体研究所 | Laser radar map construction method and robot autonomous navigation method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5000564A (en) * | 1990-03-09 | 1991-03-19 | Spectra-Physics, Inc. | Laser beam measurement system |
| CN1825138A (en) * | 2006-04-07 | 2006-08-30 | 哈尔滨工业大学 | Multi-frequency synchronous modified large range high precision fast laser ranging method and apparatus |
| CN1851499A (en) * | 2006-05-19 | 2006-10-25 | 武汉大学 | Data acquisition device for laser distance measurement, and its collecting flowchart |
-
2008
- 2008-10-12 CN CN2008101713176A patent/CN101387700B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5000564A (en) * | 1990-03-09 | 1991-03-19 | Spectra-Physics, Inc. | Laser beam measurement system |
| CN1825138A (en) * | 2006-04-07 | 2006-08-30 | 哈尔滨工业大学 | Multi-frequency synchronous modified large range high precision fast laser ranging method and apparatus |
| CN1851499A (en) * | 2006-05-19 | 2006-10-25 | 武汉大学 | Data acquisition device for laser distance measurement, and its collecting flowchart |
Non-Patent Citations (2)
| Title |
|---|
| 刘志强 等.车辆防撞检测技术研究.《公路交通科技应用技术版》.2008,33-35. * |
| 金峰 等.机器人IMU与激光扫描测距传感器数据融合.《机器人》.2000,第22卷(第6期),470-473. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101387700A (en) | 2009-03-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101387700B (en) | Data fusing method and system based on multi-laser scanner | |
| CN101393264B (en) | Moving target tracking method and system based on multi-laser scanner | |
| JP7465312B2 (en) | High resolution lidar using high frequency pulse emission. | |
| US10495757B2 (en) | Intelligent ladar system with low latency motion planning updates | |
| Lambert et al. | Performance analysis of 10 models of 3D LiDARs for automated driving | |
| US11187790B2 (en) | Laser scanning system, laser scanning method, movable laser scanning system, and program | |
| JP4669661B2 (en) | How to correct data for multiple optoelectronic sensors | |
| Liang et al. | The use of a mobile laser scanning system for mapping large forest plots | |
| CN108875804B (en) | Data processing method based on laser point cloud data and related device | |
| CN111781608B (en) | Moving target detection method and system based on FMCW laser radar | |
| JP2020021326A (en) | Information processing method, information processing apparatus and program | |
| CN111045000A (en) | Monitoring system and method | |
| KR102151815B1 (en) | Method and Apparatus for Vehicle Detection Using Lidar Sensor and Camera Convergence | |
| CN111856507B (en) | Environment sensing implementation method, intelligent mobile device and storage medium | |
| Rosenberger et al. | Analysis of real world sensor behavior for rising fidelity of physically based lidar sensor models | |
| CN115661269A (en) | External parameter calibration method and device for camera and laser radar and storage medium | |
| RU2679923C1 (en) | Method for obtaining spatial model of environment in real time on basis of laser location and device for implementation thereof | |
| US11879979B2 (en) | Method and apparatus for dynamic reconfiguration of region of interest in integrated FMCW LIDAR | |
| Tu et al. | Geometrical model of laser triangulation system based on synchronized scanners | |
| Kloeker et al. | Real-time point cloud fusion of multi-lidar infrastructure sensor setups with unknown spatial location and orientation | |
| KR102050756B1 (en) | Micro target detection device | |
| Zhang et al. | An algorithm for motion estimation based on the interframe difference detection function model | |
| KR20220032792A (en) | Edge computing-based headlight inspection device and method | |
| Wang et al. | An Advanced System‐Level Testing for Roadside Multimodal Sensing and Processing in IoV | |
| Sobottka et al. | Vision-based driver assistance using range imagery |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20110921 Termination date: 20181012 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |