CN109448034A - A kind of part pose acquisition methods based on geometric primitive - Google Patents
A kind of part pose acquisition methods based on geometric primitive Download PDFInfo
- Publication number
- CN109448034A CN109448034A CN201811242566.XA CN201811242566A CN109448034A CN 109448034 A CN109448034 A CN 109448034A CN 201811242566 A CN201811242566 A CN 201811242566A CN 109448034 A CN109448034 A CN 109448034A
- Authority
- CN
- China
- Prior art keywords
- local coordinate
- coordinate system
- point cloud
- primitive
- feature
- 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
Links
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
- G06T7/70—Determining position or orientation of objects or cameras
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Length Measuring Devices With Unspecified Measuring Means (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of part pose acquisition methods based on geometric primitive, belong to vision positioning field;It is usually to be obtained by point cloud registration method that part posture information, which obtains, it needs to carry out the problem of essence registration can be only achieved required precision by ICP after existing Feature Descriptor method for registering existing characteristics point extraction time length, rough registration, and the present invention proposes the method for utilizing geometric primitive to obtain part pose.The method of the present invention is by extracting the geometric primitive feature (plane in part point cloud, cylindrical surface), and according to the positional relationship building local coordinate system and local coordinate feature between primitive, the local coordinate system transformational relation that point cloud and template is directly calculated by local coordinate characteristic matching, to obtain part point Yun Weizi;The present invention can be obtained accurate pose accuracy without carrying out essence registration, and computational efficiency is high.
Description
Technical field
The present invention relates to the pose acquisition methods with geometric primitives characteristic parts such as plane cylindrical surfaces, belong to vision positioning
Technical field, in particular to a kind of part pose acquisition methods based on geometric primitive.
Background technique
During Automated assembly, robot is firstly the need of the autonomous crawl for completing Assembly part, the three-dimensional position of part
Appearance information is the necessary condition that robot completes crawl.Three-dimensional pose is between part to be grabbed current pose and template pose
Relationship.Pass through part point cloud and template point cloud on the basis of obtaining part point cloud data to be grabbed using three-dimensional scanning device
Registration obtains rotation between the two, translation matrix (R, T), and the RT acquired is the pose of part.
Existing point cloud registration method can be divided into rough registration and essence two processes of registration.Rough registration method includes: extraction point
Cloud characteristic point, the Feature Descriptor for calculating characteristic point, Feature Descriptor are matched and are carried out according to the point-to-point cloud of matched feature
Rigid body translation.Essence registration generally advanced optimizes initial registration result using ICP algorithm, to obtain final pose.Through
The Feature Descriptor of allusion quotation includes: FPFH, and global characteristics description such as the local feature descriptions such as SHOT and VFH, CVFH is used
These description carry out point cloud registering result and stablize, but calculating process is complicated, and efficiency is lower.Dirk Buchholz et al. is proposed
Point swears construction feature vector to feature (PPF), using the distance and its method of two o'clock, and this method utilizes voting mechanism progress
Match, method is simple but computationally intensive, low efficiency.
In conclusion most of point cloud registration method requires to calculate point Yun Tezheng, rough registration is then carried out, after rough registration
It usually requires to carry out ICP iteration optimization realization essence registration, but this period for calculating invocation point cloud pose lengthens significantly, low efficiency
Under, it is unable to satisfy industrial application and increasingly higher demands is automatically grabbed to 3 d part, therefore pose computational efficiency needs into one
Step improves.
Summary of the invention
It is an object of the invention to overcome the deficiency of the prior art, proposes that a kind of part pose based on geometric primitive obtains
Method, to solve the problems, such as that existing three-dimensional pose calculation method is computationally intensive and low efficiency.
The technical solution adopted by the present invention to solve the technical problems is:
A kind of part pose acquisition methods based on geometric primitive, comprising:
Part point cloud data to be grabbed is obtained using three-dimensional scanning device;
The geometric primitive feature (plane and cylindrical surface etc.) in part point cloud is extracted, and according to the positional relationship structure between primitive
Build local coordinate system and local coordinate feature;
Directly calculated by local coordinate characteristic matching the local coordinate system transformational relation of part point cloud and template point cloud with
Obtain part point Yun Weizi.
Preferably, include the following steps:
The part point cloud that A1, reading are obtained by spatial digitizer seeks a cloud method with PCA algorithm after point cloud pretreatment
Arrow;
The geometric primitives such as A2, plane and the cylindrical surface being fitted using RANSAC algorithm in part point cloud, according to fitting result
In primitive parameter, extract plane and three cylindrical surfaces (axis is parallel with the method for plane arrow), and be set as effective primitive;
A3, one group is selected from primitive extraction result, according to the friendship of primitive parameter Calculation Plane and three face of cylinder axis
Point, according to the triangle of 3 points of compositions in plane and planar process arrow building local coordinate system and local translation specifications;
A4, equally calculated in CAD model it is all be likely to occur primitives combination, and calculate all local coordinate systems and its
Corresponding local coordinate feature obtains pair of the local coordinate system of corresponding model and part point cloud with local coordinate characteristic matching
It should be related to, calculate the transformational relation of local coordinate system.
Preferably, the step of effective primitive of geometric primitive extraction is fitted in step A2 are as follows:
A21, using in RANSAC algorithm match point cloud plane and cylindrical surface, primitive fitting result be expressed as the primitive
Spatial parameter.Wherein there are four parameters for the fitting result of plane: A, B, C, D have respectively corresponded four of plane parameter equation
Parameter.The parametric equation for the plane being fitted are as follows: Ax+By+Cz+D=0.The result of cylindrical surface fitting has seven parameters: a, b, c,
x0,y0,z0, this seven parameters of r are to constitute the necessary condition on space cylindrical surface, and wherein the first six parameter constitutes the bus on cylindrical surface,
Its parametric equation are as follows:The radius on cylindrical surface is r.
There are some invalid primitives in the result that A22, primitive extract, and are screened using the constraint of the positional relationship of geometric primitive
It can be used for the primitive of construction feature.Constraint 1: planar process arrow is parallel with face of cylinder axis;Constraint 2: effective cylinder on part
The radius in face;Constraint 3: the distance between the center of gravity of cylinder and the intersection point of plane and planar point cloud.It is expressed with following three formulas:Wherein θ is the threshold value of planar process arrow and cylinder bus angle;abs
(r-r0)<rmin, wherein r0It is the radius of effective cylinder in model, rminIt is the threshold value of cylinder fitting error;dis(m,p)<rpart,
Wherein m is the center of gravity of planar point cloud, and p is the intersection point of cylinder bus and plane, rpartFor the radius of circular flat in part.
Preferably, the step of local coordinate system and local translation specifications are established in step A3 are as follows:
A31, according to the parametric solution plane for the plane and cylindrical surface being fitted in A2 and the intersection point P of face of cylinder axis1,
P2,P3, then calculate the distance between 3 points d12,d23,d13, work as d12,d23,d13When not congruent, it is assumed that d12>d23>d13, then with P2
For origin, longest edgeFor X-axis, n is sweared with the method for planePLocal coordinate system is constructed for Z axis, wherein Z=nP,Y
=Z × X, this completes the foundation of local coordinate system;
A32, for ordinary circumstance, according to P1,P2,P3The distance between d12,d23,d13, it is assumed that d12≥d23≥d13, according to
Distance value, which sorts from large to small, can construct a three-dimensional feature vector t=(d12,d23,d13), this three-dimensional feature can be used
The matching of vector completion local coordinate system.For special circumstances, such as the case where 3 points of composition isosceles triangles or on part
When having symmetrical structure, there are different local coordinate systems to correspond to identical feature.In this case, due to shared with X-axis
There are two types of situation, α>90 ° or α<90 ° for the adjacent side of origin and the angle α of Y-axis, and cosine value of angle is made according to this rule
Carry out supplemental characteristic matching for a feature, to keep matching result unique, the new feature vector of building is t=(d12,d23,
d13, cos α), calculated when in order to match it is more convenient, simplify feature vector, when α>90 °, α<0 cos enables t [4]=0.When α <
At 90 °, α > 0 cos enables t [4]=1, constructs and the one-to-one local coordinate feature of local coordinate system.
Preferably, the step of model coordinate systems and feature are established, match and solve transformational relation in step A4 are as follows:
A41, according to satisfactory local coordinate systems all in the calculation of design parameters model of CAD model, and calculating office
The local coordinate feature of portion's coordinate system saves local coordinate system and corresponding local coordinate feature.
A42, the matching that feature vector is carried out according to the local coordinate feature for the local coordinate system established in part, obtain
The matching of corresponding local coordinate system is matched to obtain corresponding template characteristic vector with k-nearest neighbor.Most common matching is quasi-
It is then euclidean metric i.e. Euclidean distance, since there is a certain error for primitive fitting, and there are some similar on part
Feature, matching result the case where there are error hidings.In order to avoid such situation, it is when calculating feature vector Euclidean distance
Feature vector adds a weight per one-dimensional.Due to feature vector it is last it is one-dimensional made binary conversion treatment, so reliability with
Discrimination is relatively high, so the weight of fourth dimension should be higher than that preceding three-dimensional.Thus add weight distance computer formula: the feature vector of part
Are as follows: Tc=[Tc1,Tc2,Tc3,Tc4], the feature vector of template are as follows: Tm=[Tm1,Tm2,Tm3,Tm4].Characteristic distance are as follows:
The corresponding feature vector of characteristic distance minimum value is matching result, the i.e. corresponding local coordinate system of two local coordinate features
Match.
After A43, local coordinate system matching, under the same world coordinate system, part point cloud local coordinate origin is Oc,
The unit direction vector of three axis is xc,yc,zc.The origin of corresponding model local coordinate system is Om, the unit of three axis
Direction vector is xm,ym,zm.Transformational relation between local coordinate system can be indicated by spin matrix R and translation matrix t.It is logical
The corresponding relationship for crossing local coordinate system axis and origin can be in the hope of transformational relation.Point PcIt is one in part point cloud coordinate system
Point, the corresponding point in model local coordinate system are Pm.The relationship of two o'clock is indicated with formula: Pc=R*Pm+t;It rotates simultaneously
Matrix meets: [xc,yc,zc]=R* [xm,ym,zm], it is possible thereby to acquire R=[xc,yc,zc]*([xm,ym,zm])-1;By Oc, Om
Substitute into Shi Ke get: Oc=R*Om+t;Thus the transformational relation of part point cloud and model: t=O is acquiredc-R*Om, i.e., part is with respect to mould
The three-dimensional pose of plate.
The present invention has the advantage that compared with existing the relevant technologies
The present invention can be used for the primitive of construction feature using the constraint screening of the positional relationship of geometric primitive, quick, quasi-
Really complete the extraction of primitive;The new feature vector of building has simultaneously carried out the simplification of feature vector, facilitates local coordinate system
Matching;It proposes that weighted euclidean distance matches local coordinate feature, avoids since there is a certain error for primitive fitting, and zero
There are some similar features on part, cause matching there are the case where.It is compared with the traditional method, this method does not need to carry out ICP
Essence registration, the precision and efficiency that pose calculates all are significantly improved.
Invention is further described in detail with reference to the accompanying drawings and embodiments;But one kind of the invention is based on geometry base
The part pose acquisition methods of member are not limited to the embodiment.
Detailed description of the invention
Fig. 1 is the flow chart of the embodiment of the present invention;
Fig. 2 is the CAD model of the embodiment of the present invention;
Fig. 3 is the part point cloud of the embodiment of the present invention;
Fig. 4 is the plane and cylindrical surface that the embodiment of the present invention uses RANSAC algorithm to be fitted;
Fig. 5 be the embodiment of the present invention according to constraint condition to geometric primitive the selection result;
Fig. 6 is local coordinate system construction method of the embodiment of the present invention;
Fig. 7 is some special circumstances of building local coordinate system of the embodiment of the present invention;Wherein, Fig. 7 (a) indicates 3 points of compositions
Isosceles triangle;Fig. 7 (b) indicates there is symmetrical structure on part;Fig. 7 (c) indicates the case where α > 90 °;Fig. 7 (d) indicates α < 90 °
The case where;
Fig. 8 is that pose of the embodiment of the present invention calculates registration result;Wherein, Fig. 8 (a) indicates single part registration result;Fig. 8
(b) multiple part registration results are indicated.
Specific embodiment
A kind of part pose acquisition methods based on geometric primitive of the present invention, comprising: obtained by three-dimensional point cloud scanning device
The point cloud data for taking part, using in RANSAC algorithm match point cloud plane and cylindrical surface;Extract the plane and axis in part
Line swears parallel cylindrical surface with the method for plane, and according to this position constraint condition, constructs local coordinate system;According between primitive
Positional relationship can establish local coordinate feature, by local coordinate characteristic matching, construct on model corresponding with part point cloud
Local coordinate system, can be obtained the pose of part point cloud by the transformational relation both solved between local coordinate system.
It is illustrated so that shower part pose is sought as an example as follows, overall process flow chart is shown in Figure 1, specific steps
It is as follows:
The CAD model that step 1, reading are scanned by spatial digitizer, the CAD model is shown in Figure 2, and the zero of acquisition
Part point cloud is shown in Figure 3, after point cloud pretreatment, seeks point Yun Fashi with PCA algorithm;
Step 2 is fitted the plane and two kinds of cylindrical surface geometric primitive in part point cloud using RANSAC algorithm, referring to fig. 4
It is shown, according to the primitive parameter in fitting result, plane and axis three cylindrical surfaces parallel with the method for plane arrow are extracted, and set
For effective primitive, as shown in Figure 5;
The effective primitive of fitting geometric primitive extraction are as follows:
(21) plane primitive is carried out respectively to cloud, cylindrical surface primitive extracts, and the result of primitive fitting is expressed as the primitive
Spatial parameter.Wherein there are four parameters for the fitting result of plane: A, B, C, D have respectively corresponded four of plane parameter equation
Parameter.The parametric equation for the plane being fitted are as follows: Ax+By+Cz+D=0.The result of cylindrical surface fitting has seven parameters: a, b, c,
x0,y0,z0, this seven parameters of r are to constitute the necessary condition on space cylindrical surface, and wherein the first six parameter constitutes the bus on cylindrical surface,
Its parametric equation are as follows:The radius on cylindrical surface is r;
(22) there are some invalid primitives in the result that primitive extracts, we utilize the constraint of the positional relationship of geometric primitive
Screening can be used for the primitive of construction feature.Constraint 1: planar process arrow is parallel with face of cylinder axis;Constraint 2: effective on part
The radius on cylindrical surface;Constraint 3: the distance between the center of gravity of cylinder and the intersection point of plane and planar point cloud.With following three formulas
Expression:Wherein θ=5 ° are planar process arrow and cylinder bus angle
Threshold value;abs(r-r0)<rmin, wherein r0=3.5 be the radius of effective cylinder in model, rmin=0.2 is cylinder fitting error
Threshold value;dis(m,p)<rpart, wherein m is the center of gravity of planar point cloud, and p is the intersection point of cylinder bus and plane, rpart=4.3 be zero
The radius of circular flat in part.
Step 3 is extracted in result from primitive and selects one group, according to primitive parameter Calculation Plane and three face of cylinder axis
Intersection point, according to the triangle of 3 points of compositions in plane and planar process arrow building local coordinate system and local translation specifications;
The method for establishing local coordinate system and local translation specifications are as follows:
(31) according to the intersection point P of the parametric solution plane for the plane and cylindrical surface being fitted in A2 and face of cylinder axis1,
P2,P3, then calculate the distance between 3 points d12,d23,d13, work as d12,d23,d13When not congruent, it is assumed that d12>d23>d13, then with P2
For origin, longest edgeFor X-axis, n is sweared with the method for planePLocal coordinate system is constructed for Z axis, wherein Z=nP,Y
=Z × X, as this completes the foundation of local coordinate system by Fig. 6;
(32) for ordinary circumstance, according to P1,P2,P3The distance between d12,d23,d13, it is assumed that d12>d23>d13, according to away from
A three-dimensional feature vector t=(d can be constructed by sorting from large to small from value12,d23,d13), can with this three-dimensional feature to
Amount completes the matching of local coordinate system, for special circumstances such as Fig. 7, such as the case where 3 points of composition isosceles triangles, Huo Zheling
When having symmetrical structure on part, the case where corresponding to identical feature there are different local coordinate systems, for this case, according to
The adjacent side of origin and the angle α of Y-axis are shared there are two types of situation, α>90 ° or α<90 ° with X-axis, and according to this rule, we can be with
Carry out supplemental characteristic matching using the cosine value of angle as a feature, to keep matching result unique, the new feature of building to
Amount is t=(d12,d23,d13, cos α), more convenient, simplified feature vector is calculated when in order to match, when α>90 °, α<0 cos,
Enable t [4]=0.When α<90 °, α>0 cos enables t [4]=1, and this completes locally sit correspondingly with local coordinate system
Mark the building of feature.
Step 4, calculated in CAD model it is all be likely to occur primitives combination, and calculate all local coordinate systems and its
Corresponding local coordinate feature obtains the local coordinate system of corresponding model and part point cloud with local coordinate characteristic matching, meter
Calculate the transformational relation of local coordinate system.
The model coordinate systems and feature are established and match solution transformational relation are as follows:
(41) according to satisfactory local coordinate systems all in the calculation of design parameters model of CAD model, and calculating office
The local coordinate feature of portion's coordinate system saves local coordinate system and corresponding local coordinate feature.
(42) matching that feature vector is carried out according to the local coordinate feature for the local coordinate system established in part, obtains
The matching of corresponding local coordinate system is matched to obtain corresponding template characteristic vector with k-nearest neighbor.Most common matching is quasi-
It is then euclidean metric i.e. Euclidean distance, since there is a certain error for primitive fitting, and there are some similar on part
Feature, matching result the case where there are error hidings.In order to avoid such situation, it is when calculating feature vector Euclidean distance
Feature vector adds a weight per one-dimensional.Due to feature vector it is last it is one-dimensional made binary conversion treatment, so reliability with
Discrimination is relatively high, so the weight of fourth dimension should be higher than that preceding three-dimensional.Accordingly, add weight distance computer formula is devised: part
Feature vector are as follows: Tc=[Tc1,Tc2,Tc3,Tc4], the feature vector of template are as follows: Tm=[Tm1,Tm2,Tm3,Tm4].Characteristic distance are as follows:
The corresponding feature vector of characteristic distance minimum value is matching result, the i.e. corresponding local coordinate system of two local coordinate features
Match.
(3) after local coordinate system matching, under the same world coordinate system, part point cloud local coordinate origin is Oc,
The unit direction vector of three axis is xc,yc,zc.The origin of corresponding model local coordinate system is Om, the unit of three axis
Direction vector is xm,ym,zm.Transformational relation between local coordinate system can be indicated by spin matrix R and translation matrix t.It is logical
The corresponding relationship for crossing local coordinate system axis and origin can be in the hope of transformational relation.Point PcIt is one in part point cloud coordinate system
Point, the corresponding point in model local coordinate system are Pm.The relationship of two o'clock is indicated with formula: Pc=R*Pm+t;It rotates simultaneously
Matrix meets: [xc,yc,zc]=R* [xm,ym,zm], it is possible thereby to acquire R=[xc,yc,zc]*([xm,ym,zm])-1;By Oc, Om
Substitute into Shi Ke get: Oc=R*Om+t;Thus it acquires: t=Oc-R*Om.So far the transformational relation of part point cloud and model has been acquired,
That is the posture information of part.
In order to verify the correctness that pose seeks result, the rotational translation matrix acquired is applied to model, viewing transformation
The registration of model and part point cloud afterwards.Shown in Figure 8, the method for the present invention can effectively seek the pose of part, and
The pose of multiple parts can be sought simultaneously.Compared to traditional pose calculation, since method of the invention does not need to carry out
ICP essence registration, therefore be all significantly improved in pose computational accuracy and computational efficiency.
The above is only a preferable embodiments in present example.But the present invention is not limited to above-mentioned embodiment party
Case, it is all by the present invention any equivalent change and modification done, generated function without departing from this programme range when,
It belongs to the scope of protection of the present invention.
Claims (7)
1. a kind of part pose acquisition methods based on geometric primitive characterized by comprising
Part point cloud data to be grabbed is obtained by three-dimensional point cloud scanning device;
The geometric primitive feature in part point cloud is extracted using RANSAC algorithm, and part is constructed according to the positional relationship between primitive
Coordinate system and local coordinate feature;The geometric primitive feature includes plane and cylindrical surface;
The local coordinate system transformational relation of part point cloud and template point cloud is calculated by local coordinate characteristic matching to obtain part
Point Yun Weizi.
2. the part pose acquisition methods according to claim 1 based on geometric primitive, which is characterized in that described to pass through three
Dimension point cloud scanning device obtains part point cloud data to be grabbed, and specifically includes:
The part point cloud that A1, reading are obtained by spatial digitizer seeks point Yun Fashi with PCA algorithm after point cloud pretreatment.
3. the part pose acquisition methods according to claim 1 based on geometric primitive, which is characterized in that the use
RANSAC algorithm extract part point cloud in geometric primitive feature, and according between primitive positional relationship building local coordinate system with
And local coordinate feature;The geometric primitive feature includes plane and cylindrical surface, is specifically included:
The geometric primitives such as A2, plane and the cylindrical surface being fitted using RANSAC algorithm in part point cloud, according in fitting result
Primitive parameter extracts plane and three cylindrical surfaces (axis is parallel with the method for plane arrow), and is set as effective primitive;
A3, one group is selected from primitive extraction result, according to the intersection point of primitive parameter Calculation Plane and three face of cylinder axis, root
According to the triangle and planar process arrow building local coordinate system of 3 points of compositions in plane and local translation specifications.
4. the part pose acquisition methods according to claim 3 based on geometric primitive, which is characterized in that intend in step A2
The step of closing geometric primitive extraction effective primitive, specifically includes:
A21, using in RANSAC algorithm match point cloud plane and cylindrical surface, the result of primitive fitting be expressed as the sky of the primitive
Between parameter;Wherein the fitting result of plane has tetra- parameters of A, B, C and D, has respectively corresponded four parameters of plane parameter equation,
The parametric equation for the plane being fitted is Ax+By+Cz+D=0;The result of cylindrical surface fitting has a, b, c, x0、y0、z0With r seven
Parameter, this seven parameters are to constitute the necessary condition on space cylindrical surface, and wherein the first six parameter constitutes the bus on cylindrical surface, ginseng
Counting equation isThe radius on cylindrical surface is r;
There are some invalid primitives in the result that A22, primitive extract, and are screened using the constraint of the positional relationship of geometric primitive with structure
Build the primitive of feature;Constraint includes: constraint 1, and planar process arrow is parallel with face of cylinder axis;2 are constrained, effective cylindrical surface on part
Radius;The intersection point of constraint 3, cylinder and plane and the distance between the center of gravity of planar point cloud;Especially by following three formulas
Expression:
Wherein, θ is the threshold value of planar process arrow and cylinder bus angle;
abs(r-r0)<rmin
Wherein, r0It is the radius of effective cylinder in model, rminIt is the threshold value of cylinder fitting error;
dis(m,p)<rpart
Wherein, m is the center of gravity of planar point cloud, and p is the intersection point of cylinder bus and plane, rpartFor the radius of circular flat in part.
5. the part pose acquisition methods according to claim 3 based on geometric primitive, which is characterized in that built in step A3
It the step of vertical local coordinate system and local translation specifications, specifically includes:
A31, according to the parametric solution plane for the plane and cylindrical surface being fitted in A2 and the intersection point P of face of cylinder axis1、P2With
P3, then calculate the distance between 3 points d12、d23And d13, work as d12、d23And d13When not congruent, it is assumed that d12>d23>d13, then with P2
For origin, longest edgeFor X-axis, n is sweared with the method for planePLocal coordinate system is constructed for Z axis, wherein Z=nP,Y
=Z × X, this completes the foundation of local coordinate system;
A32, for ordinary circumstance, it is assumed that d12≥d23≥d13, sorted from large to small according to distance value can construct one it is three-dimensional
Feature vector t=(d12,d23,d13), the matching of local coordinate system is completed with this three-dimensional feature vector;
Special circumstances when having a symmetrical structure on A33, the case where constituting isosceles triangle for 3 points or part, exist not
Same local coordinate system corresponds to identical feature;In these cases, due to sharing the adjacent side of origin and the angle α of Y-axis with X-axis
There are two types of situation, α>90 ° or α<90 °, according to this rule using the cosine value of angle as a feature come supplemental characteristic
Match, to keep matching result unique, the new feature vector of building is t=(d12,d23,d13, cos α), simplify feature vector, when
When α>90 °, α<0 cos enables t [4]=0;When α<90 °, α>0 cos enables t [4]=1, constructs a pair of with local coordinate system one
The local coordinate feature answered.
6. the part pose acquisition methods according to claim 1 based on geometric primitive, which is characterized in that described to pass through office
The local coordinate system transformational relation of portion's translation specifications matching primitives part point cloud and template point cloud is to obtain part point Yun Weizi, tool
Body includes:
A4, all primitive combinations being likely to occur equally are calculated in CAD model, and calculate all local coordinate systems and its correspondence
Local coordinate feature, obtain the corresponding of the local coordinate system of corresponding model and part point cloud with local coordinate characteristic matching and close
System, calculates the transformational relation of local coordinate system.
7. the part pose acquisition methods according to claim 6 based on geometric primitive, which is characterized in that step A4 is specific
Include:
A41, according to satisfactory local coordinate systems all in the calculation of design parameters model of CAD model, and calculate local seat
The local coordinate feature for marking system saves local coordinate system and corresponding local coordinate feature;
A42, the matching that feature vector is carried out according to the local coordinate feature for the local coordinate system established in part, are corresponded to
Local coordinate system matching, obtain corresponding template characteristic vector with Euclidean distance matching criterior;Calculating feature vector Europe
Family name apart from when for feature vector add a weight per one-dimensional;
It is as follows thus to obtain add weight distance computer formula:
Wherein, Tc=[Tc1,Tc2,Tc3,Tc4] indicate part feature vector, Tm=[Tm1,Tm2,Tm3,Tm4] indicate template spy
Levy vector;
The corresponding feature vector of characteristic distance minimum value is matching result, the i.e. corresponding local coordinate system of two local coordinate features
Matching;
After A43, local coordinate system matching, under the same world coordinate system, part point cloud local coordinate origin is Oc, three
The unit direction vector of axis is xc、ycAnd zc;The origin of corresponding model local coordinate system is Om, the unit side of three axis
It is x to vectorm、ymAnd zm;Transformational relation between local coordinate system can be indicated by spin matrix R and translation matrix t;It is logical
The corresponding relationship for crossing local coordinate system axis and origin can be in the hope of transformational relation;Point PcIt is one in part point cloud coordinate system
Point, the corresponding point in model local coordinate system are Pm;The relationship of two o'clock formula Pc=R*Pm+ t is indicated;It rotates simultaneously
Matrix meets [xc,yc,zc]=R* [xm,ym,zm], it is possible thereby to acquire R=[xc,yc,zc]*([xm,ym,zm])-1;By OcAnd Om
Substitution formula can obtain Oc=R*Om+t;Thus the transformational relation t=O of part point cloud and model is acquiredc-R*Om, i.e. part opposite formwork
Three-dimensional pose.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811242566.XA CN109448034B (en) | 2018-10-24 | 2018-10-24 | Part pose acquisition method based on geometric elements |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811242566.XA CN109448034B (en) | 2018-10-24 | 2018-10-24 | Part pose acquisition method based on geometric elements |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109448034A true CN109448034A (en) | 2019-03-08 |
CN109448034B CN109448034B (en) | 2021-10-01 |
Family
ID=65547284
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811242566.XA Active CN109448034B (en) | 2018-10-24 | 2018-10-24 | Part pose acquisition method based on geometric elements |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109448034B (en) |
Cited By (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110111421A (en) * | 2019-05-10 | 2019-08-09 | 武汉海达数云技术有限公司 | A kind of method and device of mobile mapping point cloud |
CN110648361A (en) * | 2019-09-06 | 2020-01-03 | 深圳市华汉伟业科技有限公司 | Real-time pose estimation method and positioning and grabbing system of three-dimensional target object |
CN110936077A (en) * | 2019-12-31 | 2020-03-31 | 南京衍构科技有限公司 | Method for generating surfacing path of membrane type water-cooled wall |
CN111156948A (en) * | 2019-12-29 | 2020-05-15 | 苏州赛腾精密电子股份有限公司 | Three-dimensional data coordinate correction method and device for 3C glass panel detection |
CN111805543A (en) * | 2020-07-10 | 2020-10-23 | 佛山科学技术学院 | Infrared imaging target operation track detection system and coordinate conversion method thereof |
CN112102397A (en) * | 2020-09-10 | 2020-12-18 | 敬科(深圳)机器人科技有限公司 | Method, equipment and system for positioning multilayer part and readable storage medium |
CN113554614A (en) * | 2021-07-21 | 2021-10-26 | 中国人民解放军陆军工程大学 | Pipeline measurement system pose calibration method for point cloud splicing |
CN113609985A (en) * | 2021-08-05 | 2021-11-05 | 诺亚机器人科技(上海)有限公司 | Object pose detection method, detection device, robot and storage medium |
CN113643270A (en) * | 2021-08-24 | 2021-11-12 | 凌云光技术股份有限公司 | Image registration method and device based on point cloud data |
CN113706454A (en) * | 2021-07-13 | 2021-11-26 | 广东泽亨智能科技有限公司 | Workpiece offset detection method based on registration and spraying device |
CN114529704A (en) * | 2022-04-21 | 2022-05-24 | 广州中望龙腾软件股份有限公司 | Assembly mirror image method based on mass center main inertial coordinate system, terminal and storage medium |
CN114676528A (en) * | 2022-04-07 | 2022-06-28 | 西北工业大学 | Non-ideal model assembly deviation algorithm based on combined primitive method |
CN115049730A (en) * | 2022-05-31 | 2022-09-13 | 北京有竹居网络技术有限公司 | Part assembling method, part assembling device, electronic device and storage medium |
CN117523111A (en) * | 2024-01-04 | 2024-02-06 | 山东省国土测绘院 | Method and system for generating three-dimensional scenic spot cloud model |
CN117523206A (en) * | 2024-01-04 | 2024-02-06 | 南京航空航天大学 | Automatic assembly method based on cross-source point cloud and multi-mode information |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103247041A (en) * | 2013-05-16 | 2013-08-14 | 北京建筑工程学院 | Local sampling-based multi-geometrical characteristic point cloud data splitting method |
CN106228539A (en) * | 2016-07-12 | 2016-12-14 | 北京工业大学 | Multiple geometric primitive automatic identifying method in a kind of three-dimensional point cloud |
US20170191826A1 (en) * | 2016-01-05 | 2017-07-06 | Texas Instruments Incorporated | Ground Plane Estimation in a Computer Vision System |
CN107564059A (en) * | 2017-07-11 | 2018-01-09 | 北京联合大学 | Object positioning method, device and NI Vision Builder for Automated Inspection based on RGB D information |
WO2018007518A1 (en) * | 2016-07-06 | 2018-01-11 | Respiratory Innovations Pty Ltd | A realtime radiotherapy markerless calibration and measurement system |
CN108171748A (en) * | 2018-01-23 | 2018-06-15 | 哈工大机器人(合肥)国际创新研究院 | A kind of visual identity of object manipulator intelligent grabbing application and localization method |
-
2018
- 2018-10-24 CN CN201811242566.XA patent/CN109448034B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103247041A (en) * | 2013-05-16 | 2013-08-14 | 北京建筑工程学院 | Local sampling-based multi-geometrical characteristic point cloud data splitting method |
US20170191826A1 (en) * | 2016-01-05 | 2017-07-06 | Texas Instruments Incorporated | Ground Plane Estimation in a Computer Vision System |
WO2018007518A1 (en) * | 2016-07-06 | 2018-01-11 | Respiratory Innovations Pty Ltd | A realtime radiotherapy markerless calibration and measurement system |
CN106228539A (en) * | 2016-07-12 | 2016-12-14 | 北京工业大学 | Multiple geometric primitive automatic identifying method in a kind of three-dimensional point cloud |
CN107564059A (en) * | 2017-07-11 | 2018-01-09 | 北京联合大学 | Object positioning method, device and NI Vision Builder for Automated Inspection based on RGB D information |
CN108171748A (en) * | 2018-01-23 | 2018-06-15 | 哈工大机器人(合肥)国际创新研究院 | A kind of visual identity of object manipulator intelligent grabbing application and localization method |
Non-Patent Citations (2)
Title |
---|
RUI FIGUEIREDO 等: "Robust cylinder detection and pose estimation using 3D point cloud information", 《2017 IEEE INTERNATIONAL CONFERENCE ON AUTONOMOUS ROBOT SYSTEMS AND COMPETITIONS (ICARSC)》 * |
张天 等: "基于形状特征的管路接头测量和三维重建方法", 《计算机集成制造系统》 * |
Cited By (26)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110111421A (en) * | 2019-05-10 | 2019-08-09 | 武汉海达数云技术有限公司 | A kind of method and device of mobile mapping point cloud |
CN110648361A (en) * | 2019-09-06 | 2020-01-03 | 深圳市华汉伟业科技有限公司 | Real-time pose estimation method and positioning and grabbing system of three-dimensional target object |
CN110648361B (en) * | 2019-09-06 | 2022-01-11 | 深圳市华汉伟业科技有限公司 | Real-time pose estimation method and positioning and grabbing system of three-dimensional target object |
CN111156948A (en) * | 2019-12-29 | 2020-05-15 | 苏州赛腾精密电子股份有限公司 | Three-dimensional data coordinate correction method and device for 3C glass panel detection |
CN110936077B (en) * | 2019-12-31 | 2021-11-26 | 南京衍构科技有限公司 | Method for generating surfacing path of membrane type water-cooled wall |
CN110936077A (en) * | 2019-12-31 | 2020-03-31 | 南京衍构科技有限公司 | Method for generating surfacing path of membrane type water-cooled wall |
CN111805543A (en) * | 2020-07-10 | 2020-10-23 | 佛山科学技术学院 | Infrared imaging target operation track detection system and coordinate conversion method thereof |
CN111805543B (en) * | 2020-07-10 | 2022-04-26 | 佛山科学技术学院 | Infrared imaging target operation track detection system and coordinate conversion method thereof |
CN112102397B (en) * | 2020-09-10 | 2021-05-11 | 敬科(深圳)机器人科技有限公司 | Method, equipment and system for positioning multilayer part and readable storage medium |
CN112102397A (en) * | 2020-09-10 | 2020-12-18 | 敬科(深圳)机器人科技有限公司 | Method, equipment and system for positioning multilayer part and readable storage medium |
CN113706454A (en) * | 2021-07-13 | 2021-11-26 | 广东泽亨智能科技有限公司 | Workpiece offset detection method based on registration and spraying device |
CN113706454B (en) * | 2021-07-13 | 2022-05-03 | 广东泽亨智能科技有限公司 | Workpiece offset detection method based on registration and spraying device |
CN113554614A (en) * | 2021-07-21 | 2021-10-26 | 中国人民解放军陆军工程大学 | Pipeline measurement system pose calibration method for point cloud splicing |
CN113609985B (en) * | 2021-08-05 | 2024-02-23 | 诺亚机器人科技(上海)有限公司 | Object pose detection method, detection device, robot and storable medium |
CN113609985A (en) * | 2021-08-05 | 2021-11-05 | 诺亚机器人科技(上海)有限公司 | Object pose detection method, detection device, robot and storage medium |
CN113643270A (en) * | 2021-08-24 | 2021-11-12 | 凌云光技术股份有限公司 | Image registration method and device based on point cloud data |
CN113643270B (en) * | 2021-08-24 | 2024-04-26 | 凌云光技术股份有限公司 | Image registration method and device based on point cloud data |
CN114676528A (en) * | 2022-04-07 | 2022-06-28 | 西北工业大学 | Non-ideal model assembly deviation algorithm based on combined primitive method |
CN114676528B (en) * | 2022-04-07 | 2024-02-23 | 西北工业大学 | Non-ideal model assembly deviation algorithm based on combined primitive method |
CN114529704A (en) * | 2022-04-21 | 2022-05-24 | 广州中望龙腾软件股份有限公司 | Assembly mirror image method based on mass center main inertial coordinate system, terminal and storage medium |
CN115049730A (en) * | 2022-05-31 | 2022-09-13 | 北京有竹居网络技术有限公司 | Part assembling method, part assembling device, electronic device and storage medium |
CN115049730B (en) * | 2022-05-31 | 2024-04-26 | 北京有竹居网络技术有限公司 | Component mounting method, component mounting device, electronic apparatus, and storage medium |
CN117523111A (en) * | 2024-01-04 | 2024-02-06 | 山东省国土测绘院 | Method and system for generating three-dimensional scenic spot cloud model |
CN117523206A (en) * | 2024-01-04 | 2024-02-06 | 南京航空航天大学 | Automatic assembly method based on cross-source point cloud and multi-mode information |
CN117523111B (en) * | 2024-01-04 | 2024-03-22 | 山东省国土测绘院 | Method and system for generating three-dimensional scenic spot cloud model |
CN117523206B (en) * | 2024-01-04 | 2024-03-29 | 南京航空航天大学 | Automatic assembly method based on cross-source point cloud and multi-mode information |
Also Published As
Publication number | Publication date |
---|---|
CN109448034B (en) | 2021-10-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109448034A (en) | A kind of part pose acquisition methods based on geometric primitive | |
JP4785880B2 (en) | System and method for 3D object recognition | |
CN105551039B (en) | The scaling method and device of structural light three-dimensional scanning system | |
Sharp et al. | ICP registration using invariant features | |
He et al. | Sparse template-based 6-D pose estimation of metal parts using a monocular camera | |
Zhang et al. | A robust, real-time ellipse detector | |
Du et al. | An extension of the ICP algorithm considering scale factor | |
JP6483832B2 (en) | Method and system for scanning an object using an RGB-D sensor | |
CN109559340A (en) | A kind of parallel three dimensional point cloud automation method for registering | |
CN101311963A (en) | Round mark point center picture projection point position acquiring method for positioning video camera | |
CN112101073B (en) | Face image processing method, device, equipment and computer storage medium | |
CN109636852A (en) | A kind of monocular SLAM initial method | |
Yun et al. | Registration of multiview point clouds for application to ship fabrication | |
Cheung et al. | Measurement and characterization of ultra-precision freeform surfaces using an intrinsic surface feature-based method | |
CN109766903A (en) | A kind of point cloud model SURFACES MATCHING method based on curved surface features | |
Sethi et al. | Curve and surface duals and the recognition of curved 3D objects from their silhouettes | |
CN109345571B (en) | Point cloud registration method based on extended Gaussian image | |
CN109829459A (en) | Based on the vision positioning method for improving RANSAC | |
Benedens | Robust watermarking and affine registration of 3d meshes | |
Zha et al. | Registration of range images with different scanning resolutions | |
CN111275747B (en) | Virtual assembly method, device, equipment and medium | |
CN106408654B (en) | A kind of creation method and system of three-dimensional map | |
Azad et al. | Accurate shape-based 6-dof pose estimation of single-colored objects | |
Barros et al. | Real-time human pose estimation from body-scanned point clouds | |
Du et al. | ICP with bounded scale for registration of mD point sets |
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 |