CN109508141A - Redundant point detection method for point cloud data fitting - Google Patents
Redundant point detection method for point cloud data fitting Download PDFInfo
- Publication number
- CN109508141A CN109508141A CN201810347829.7A CN201810347829A CN109508141A CN 109508141 A CN109508141 A CN 109508141A CN 201810347829 A CN201810347829 A CN 201810347829A CN 109508141 A CN109508141 A CN 109508141A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- cloud data
- data
- point
- simplified
- 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
- 238000001514 detection method Methods 0.000 title abstract description 11
- 239000011159 matrix material Substances 0.000 claims abstract description 49
- 238000006243 chemical reaction Methods 0.000 claims abstract description 21
- 238000005070 sampling Methods 0.000 claims abstract description 14
- 230000007704 transition Effects 0.000 claims description 37
- 238000005457 optimization Methods 0.000 claims description 24
- 238000000034 method Methods 0.000 claims description 17
- 230000005540 biological transmission Effects 0.000 abstract description 5
- 238000013500 data storage Methods 0.000 abstract description 5
- 238000004364 calculation method Methods 0.000 abstract 1
- 238000005520 cutting process Methods 0.000 description 10
- 238000006073 displacement reaction Methods 0.000 description 10
- 241001269238 Data Species 0.000 description 8
- 238000010586 diagram Methods 0.000 description 8
- 210000000214 mouth Anatomy 0.000 description 3
- 238000010276 construction Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0602—Interfaces specially adapted for storage systems specifically adapted to achieve a particular effect
- G06F3/0626—Reducing size or complexity of storage systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0602—Interfaces specially adapted for storage systems specifically adapted to achieve a particular effect
- G06F3/061—Improving I/O performance
- G06F3/0613—Improving I/O performance in relation to throughput
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0628—Interfaces specially adapted for storage systems making use of a particular technique
- G06F3/0638—Organizing or formatting or addressing of data
- G06F3/064—Management of blocks
- G06F3/0641—De-duplication techniques
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Image Processing (AREA)
- Image Generation (AREA)
Abstract
The invention provides a redundant point detection method for point cloud data fitting, which is executed in a data processing device and comprises the following steps: receiving first and second point cloud data having a continuous relationship from a scanner; carrying out reduction sampling on the first point cloud data and the second point cloud data to respectively become first simplified point cloud data and second simplified point cloud data; establishing a three-dimensional point cloud index structure of the first simplified point cloud data for searching each data point in the second simplified point cloud data to correspond to each data point of the first simplified point cloud data; establishing an optimized conversion matrix, and enabling the tangent planes of two data points corresponding to the first simplified point cloud data and the second simplified point cloud data to have a minimum distance after the second simplified point cloud data is converted by the optimized conversion matrix; and when the minimum distance is smaller than the judgment value, the two data points are redundant data points. The technical scheme reduces the throughput of data, so that the cost of data storage can be effectively reduced, and the transmission efficiency and the calculation efficiency of the data are improved.
Description
Technical field
The present invention is to particularly relate to have point for two in relation to a kind of redundancy point detecting method for point cloud data fitting
When cloud (point cloud) data are bonded, the method that detects the repeated and redundant point of two point cloud data.
Background technique
Known point cloud data presentation system is to carry out continuous scanning to an object using scanner in movement, wherein
A point cloud data can be obtained in every run-down, which includes multiple data points, and each data point includes three-dimensional seat
Mark marks information for the image of the construction object.So can produce more in a short time when carrying out quickly scanning capture
Point cloud data, the three-dimensional coordinate information of each point cloud data are to enter scanning system with tens of speed to hundreds of per second.
For example, Fig. 8 A is please referred to, scanner successively generates one first point cloud data 41 and one second point cloud data 42,
Please refer to Fig. 8 B, scanning system can according to the repeat region 40 of two point cloud data 41,42 by two point cloud data 41,42 each other
Fitting, wherein the boundary 410 of first point cloud data 41 is fallen in the range of the second point cloud data 42, opposite, the second point
The boundary 420 of cloud data 42 is fallen in the range of the first point cloud data 41, is the repeat region between two boundary 410,420
40.Please refer to Fig. 8 C and Fig. 8 D, scanner then generates a third point cloud data 43, scanning system be after being bonded first,
Second point cloud data 41,42 is bonded the third point cloud data 43 again.The rest may be inferred, please refers to Fig. 8 E, and scanning system can receive again
One the 4th point cloud data 44 or again more point cloud datas of received in sequence, are bonded with further progress.
As previously mentioned, the scanning system is according to the repeat region of two adjacent point cloud datas and by the two o'clock cloud serial data
Fitting is connect, on the whole, after completing to scan the scanning of the object, between different point cloud datas, repeat region number
The repeatability at strong point is high, not only promotes the cost of data storage, also influences the efficiency of transmission and computational efficiency of point cloud data.
Summary of the invention
In view of this, the main object of the present invention is to provide a kind of redundancy point detecting method for point cloud data fitting,
To detect the repeated and redundant point of the point cloud data of fitting, when excluding duplicate redundant points, that is, data storage can effectively reduce
Cost also promotes the efficiency of transmission and computational efficiency of data.
The present invention is used for the redundancy point detecting method of point cloud data fitting, executes in a data processing equipment, at the data
Device line scanner is managed to receive point cloud data, this method includes:
Receive one first point cloud data and one second point cloud data with serial relation；
Reduced sampling is carried out to first point cloud data and second point cloud data, simplifies point cloud to respectively become one first
Data simplify point cloud data with one second；
The three-dimensional point cloud index structure for establishing the first simplified point cloud data, for searching in the second simplified point cloud data
Each data point is with each data point corresponding to the first simplified point cloud data；
An optimization transition matrix is established, the second simplified point cloud data is made to pass through the conversion of the optimization transition matrix
Afterwards, this first simplifies point cloud data and this second simplifies between the tangent plane at point cloud data corresponding two stroke count strong point and have one most
Small distance；
Judge the minimum range whether less than a judgment value；If so, two stroke count strong point is the data point of redundancy；If
It is no, the three-dimensional point cloud index structure for establishing the first simplified point cloud data is returned to, for searching each number in second point cloud data
Strong point with correspond to this first simplify point cloud data each data point the step of, with carry out the optimization transition matrix iteration fortune
It calculates.
Detection method according to the present invention is to detect duplicate redundant digit strong point in two simplified point cloud datas
Come, the redundant digit strong point represents the data point for the overlapping region that the two simplified point cloud data scans, therefore, upper in application
Duplicate data point between different point cloud datas will not be repeated to deposit at the redundant digit strong point that can exclude any simplified point cloud data
Storage and operation, reduce the handling capacity of data, thus can be effectively reduced data storage cost, also promoted data efficiency of transmission and
Computational efficiency.
Detailed description of the invention
Fig. 1: implement the block schematic diagram of the embodiment of the system of detection method.
Fig. 2A~Fig. 2 C: the schematic diagram that the first point cloud data is bonded by the embodiment of the present invention each other with the second point cloud data.
Fig. 3: the flow diagram of the embodiment of detection method.
Fig. 4: the first simplifies the schematic diagram of point cloud data.
Fig. 5: the second simplifies the schematic diagram of point cloud data.
Fig. 6 A~Fig. 6 F: the schematic diagram of present invention generation three-dimensional point cloud index structure.
Fig. 7: the schematic diagram that Fig. 4 is bonded each other with the simplification point cloud data of Fig. 5.
Fig. 8 A~Fig. 8 E: the known schematic diagram by more point cloud data concatenation fittings.
Specific embodiment
Referring to FIG. 1, the system for executing detection method includes scanner 10 and a data processing equipment 20.
The scanner 10 can be used structure light (structured light) scanning technique and be scanned an object to obtain one
Point cloud (point cloud) data, the point cloud data include multiple data points, and the data points such as this may make up the image of object,
Wherein each data point may include luminance information, depth information (range image) and three-dimensional space parameter, three-dimensional space
Parameter may include x-axis rotation index value (yaw) of each data point, the angle (α) of x coordinate and x-axis, y-axis rotation index value
(pitch), y-coordinate and the angle (β) of y-axis, z-axis rotation index value (roll), z coordinate and the angle (γ) of z-axis etc..The number
It can be computer according to processing unit 20, but not limited to this, 20 line of data processing equipment scanner 10, to receive the scanning
Point cloud data captured by device 10.
It should be noted that the embodiment of detection method be illustrate by taking scanning system in mouth as an example, but not as
Limit, detection method are equally applicable for other types of three-dimensional reconstruction system, for example, present invention can apply to each productions
Three dimensional detection, Quick-forming and the three-dimensional in industry field are printd.
The overall architecture process of detection method is briefly explained first, firstly, the scanner 10 is moved along a track
Dynamic, while mobile, the scanner 10 is to an object continuous scanning, wherein every run-down can produce a point cloud number
According to, therefore continuous scanning can produce more point cloud datas, and the point cloud datas such as this are received by the data processing equipment 20.Please refer to figure
2A considers the first point cloud data P and the second point cloud data Q successively scanned, since scanner 10 is continuously swept between traveling
It retouches, so the content of two point cloud data P, Q not fit like a glove, but has duplicate part each other, the data processing equipment 20
The orientation of the second point cloud data Q be can adjust for correctly fitting in the first point cloud data P.In order to adjust the second point cloud data Q's
Orientation, the present invention establish the three-dimensional point cloud index structure of the first point cloud data P, according to the three-dimensional point cloud index structure, for first
Step obtains the corresponding relationship of each data point between two point cloud data P, Q.Then, the present invention generates after interative computation for several times
One optimizes transition matrix, for the second point cloud data Q by the conversion of the optimization transition matrix to fit in first cloud
Data P.By fitting after, two point cloud data P, Q are substantially overlapped duplicate part each other, thus can two point cloud data P,
The overlapping of Q detects duplicate redundant digit strong point.
Technological means of the invention described further below, please with reference to Figure 3, the embodiment of detection method includes
Following steps:
Step S1: scanning object
The scanner 10 can be hand-scanner or fixed scanner, in a handheld for scanner, dentist
Scanner 10 is protruded into the oral cavity of patient to the shooting operation for carrying out tooth, when the scanner 10 is moved along patient's Tooth surface
When moving and scanning, the several tooth body point cloud datas with serial relation can be obtained.In order to illustrate, please refer to Fig. 2A, the present invention with
The scanner 10 is continuously shot for obtaining the first point cloud data P and the second point cloud data Q between traveling, because of the two o'clock cloud
Data P, Q are therefore the first point cloud data P and the second point cloud number in the mobile in the state of institute continuous scanning of the scanner 10
Different mouth intracavitary locations are corresponded respectively to according to Q.
Step S2: reduced sampling (Down-sampling)
The data processing equipment 20 receives the first point cloud data P and second point cloud data Q, wherein the data processing
Uniform sampling can be used in device 20, is captured according to a sampling period and stores the first point cloud data P and second point cloud data Q
Data point.For example, when the first point cloud data P has N stroke count strong point, which can take for every n stroke count strong point
One stroke count strong point, then first point cloud data P only includes N/n stroke count strong point after reduced sampling, therefore the data point after sampling
Quantity is effectively preliminary whereby to reduce volume of transmitted data and operand lower than the number of data points before sampling.
In order to illustrate, the point cloud data after reduced sampling is known as simplifying point cloud data, so, please refer to Fig. 2 B, Fig. 4
Become first after reduced sampling with Fig. 5, first point cloud data P and simplify point cloud data P', second point cloud data Q is through reducing
Become second after sampling and simplifies point cloud data Q'.
Step S3: three-dimensional point cloud index structure is established
In order to promote the searching efficiency of data point, the present invention is to establish the three-dimensional point cloud rope of the first simplified point cloud data P'
Guiding structure, for searching each data point in the second simplified point cloud data Q' to correspond to each of the first simplified point cloud data P'
Data point, the present invention establish the purpose of the three-dimensional point cloud index structure, are quickly to search in the first simplified point cloud data P'
It seeks to a data point pi, make data point piSimplify the data point q in point cloud data Q' with secondjClosest, the present invention is implemented
Example is to establish three-dimensional point cloud index structure with k-d tree (k-dimensional tree), only establishes point cloud data by k-d tree
Three-dimensional point cloud index structure is the usual knowledge of technical field, is only simply described as follows.
First to illustrate in the case where one-dimensional, it is assumed that first simplifies the dimension of point cloud data P' with the second simplification point cloud data Q'
Degree is a dimension, by binary search method (binary search), all data points that point cloud data P' can be simplified to first
piIt is ranked up according to its coordinate value to establish one-dimensional point cloud index structure.Because being the first letter by taking a dimension as an example
Data point p after changing point cloud data P' sequenceiSeveral lines expansion after, allow second simplify point cloud data Q' data point qjOne by one with
It sorts in the data point p of middle position (middle element)iCompare, makes second to simplify each of point cloud data Q' data point
qjIt can search and simplify immediate data point p in point cloud data P' with firsti。
If the first dimension for simplifying the simplified point cloud data Q' of point cloud data P' and second is three dimensions, by k- above-mentioned
D tree, which is extended to, establishes three-dimensional point cloud index structure of the invention.Firstly, simplifying all data point p of point cloud data P' to firstiRoot
It is ranked up according to its coordinate value to establish three-dimensional point cloud index structure, this structure includes root node (root as shown in Figure 6A
Node) RN, root node RN include the first all data point p for simplifying point cloud data P'i, please refer to Fig. 6 B, Bounding Box
(bounding box) BB stores all data point pi, Bounding Box BB is by data point piCoordinate x, y and z three dimensionality most
Big value determines with minimum value, that is, first simplifies point cloud data P' in the data area of triaxial coordinate system, root node RN it
Under each layer of node all represent the data area that the node covers a cloud, its three-dimensional Bounding Box BB, left and right is recorded in node
The index of node, the cutting dimension (cutdimension) as judgment basis and cut value (cut value), and comprising
Data point if finding the median m1 of the corresponding three-dimensional point cloud Z coordinate of root node RN along Z axis, please refers to figure without loss of generality
The space Bounding Box BB is then cut into two parts of skies with the plane Plane1 (that is: the plane of z=m1) that this median m1 is established by 6D
Between, which respectively corresponds two node SN shown in Fig. 6 C；Then the step of repeating the above cutting room, that is, it is right
The node of the bottom at present, such as two node SN of Fig. 6 C bottom, find median m2, m3 each along Y-axis, such as Fig. 6 F institute
Show, (that is: y=m3's is flat by each own cutting planes Plane2 (that is: the plane of y=m2) of median m2, m3 and Plane3
Face), two spaces that the two cutting planes Plane2, Plane3 again separates plane Plane1 are respectively cut into two pieces, if at this time
Space after the cutting of plane Plane2, Plane3 only remains a three-dimensional point, then stops cutting, and the space after each cutting at this time
Corresponding node is leaf node (leaf node) LN shown in Fig. 6 E, that is, leaf node LN only includes a data point, and is
The terminal of search.If not only one three-dimensional point in space, continues on X-axis cutting, such cutting step will after cutting
Until being performed repeatedly until in space an only surplus three-dimensional point.
Step S4: initial conversion matrix is established
Under rectangular coordinate system, the present invention utilize initial x-axis angle α ', initial y-axis angle β ', initial z-axis angle
γ ', initial x-axis displacement tx ', initial y-axis displacement ty ' and initial z-axis displacement tz ' establish the initial conversion matrix of a 4x4
Minitial, it is expressed as follows:
Wherein RiniFor initial rotation vector, tiniFor initial displacement matrix, respectively indicate as follows:
In the embodiment of the present invention, initial x-axis angle α ', initial y-axis angle β ', initial z-axis angle γ ', the displacement of initial x-axis
Tx ', initial y-axis displacement ty ' and initial z-axis displacement tz ' may respectively be 0, and but not limited to this.
Step S5: corresponding data point is searched
The purpose of this step is all data point q for simplifying point cloud data Q' to secondj, simplify point cloud data P' first
In find the data point p being closer toi, to establish two data point pi、qjCorresponding relationship.It is indexed according to the three-dimensional point cloud of step S3
Structure, comparison query data point q when searchjWill be toward the child node movement on which side with the judgement of place range of nodes, and persistently chase after
Track k-d tree is until the leaf node of the bottom.
After completing this step for the first time, first simplifies each data point p of point cloud data P'iSecond can tentatively be corresponded to and simplify point
Each data point q of cloud data Q'j.Referring to FIG. 3, first illustrating the step of step S5~S7 is interative computation herein, change each time
It can produce an optimization transition matrix for operation, consideration is at present kth time interative computation, and kth -1 time changes to be previous
For operation, and when returning step S5 by -1 interative computation of kth, first simplifies each data point p of point cloud data P'iIt can be right
It should arrive and transition matrix conversion is optimized by kth -1Second afterwards simplifies each data point q of point cloud data Q'j.Wherein k
When=1, which is initial conversion matrix Minitial。
Step S6: it establishes and optimizes transition matrix
After abovementioned steps S5, if first simplifies point cloud data P' with a data point di, second simplifies point cloud data Q'
With corresponding data point si, that is, data point siIt is to search to optimize transition matrix by kth -1 in step S5Second after conversion simplifies the data point of point cloud data Q', in step s 6, each data point siBy a transition matrix
MkBecome data point M afterwardsk·si, data point Mk·siWith data point diEstablish a vector, wherein as transition matrix MkIt is one
Optimize transition matrixThen data point diTangent plane and data point Mk·siTangent plane between have a most narrow spacing
From the minimum range can be by the vector in data point diSurface by Tangent Plane Method vector niProjection determine, the optimization transition matrix
Mopt_kGeneral formula can be expressed as follows:
Wherein:
Mk=Tk(tx,ty,tz)·Rk(α,β,γ)；
In above formula, TkIt is transposed matrix, RkIt is spin matrix, respectively indicates as follows:
Wherein:
r11=cos γ cos β；
r12=-sin γ cos α+cos γ sin β sin α；
r13=sin γ sin α+cos γ sin β cos α；
r21=sin γ cos β；
r22=cos γ cos α+sin γ sin β sin α；
r23=-cos γ sin α+sin γ sin β cos α；
r31=-sin β；
r32=cos β sin α；
r33=cos β cos α.
Tx, ty and tz are respectively that x-axis displacement, y-axis displacement and z-axis displacement, the present invention approach means using linear solution, it is assumed that
Three axle clamp angles level off to 0, that is, x-axis angle α ≈ 0, y-axis angle β ≈ 0, z-axis angle γ ≈ 0, then sin α ≈ α, cos α ≈ 0,
Sin β ≈ β, cos β ≈ 0, sin γ ≈ γ, cos γ ≈ 0, then spin matrix RkBecomeIt is expressed as follows:
Transition matrix MkBecomeIt is expressed as follows:
This is converted and substitutes into optimization transition matrix Mopt_k, make optimization transition matrix Mopt_kBecomeTable
Show as follows:
Then wherein i-th all useable linear is unfolded, and is expressed as follows:
So all i of linear expansions can form a linear system:
Ax=b；
Wherein:
X=(α β γ tx ty tz)T；
Wherein:
ai1=nizsiy-niysiz；
ai2=nixsiz-nizsix；
ai3=niysix-nixsiy；
Then
Also that is, above formula can be made | Ax-b |2Parameter vector x with minimum value (includes α, β, γ, tx、ty、tzDeng six
Conversion parameter) it is to optimize conversion parameter, it is expressed as αopt、βopt、γopt、txopt、tyopt、tzopt, it is expressed as follows:
xopt=argminx|Ax-b|2；
xopt=(αopt,βopt,γopt,txopt,tyopt,tzopt)T；
Wherein, the equal optimization conversion parameter αopt、βopt、γopt、txopt、tyopt、tzopt(pseudo is returned using quasi-
Inverse) operation and obtain.
Finally, according to the equal optimization conversion parameter αopt、βopt、γopt、txopt、tyopt、tzoptIt is applied to rotation above-mentioned
Torque battle array RkWith transposed matrix TkTo get arrive the optimization transition matrixThis is the optimization transition matrix updated, table
Show as follows:
So of the invention second simplifies the optimization transition matrix that point cloud data Q' passes through updateConversion after
BecomeFirst can be made to simplify the second simplified point cloud data after point cloud data P' and conversionCorresponding two
There is a minimum range between the tangent plane at stroke count strong point.
Step S7: convergence assessment
Abovementioned steps S5 to this step S7 is iterative process, and whether step 7 assessment iteration meets a condition of convergence, if meeting
The condition of convergence indicates that this two simplified point cloud data P', Q' can be by the optimization transition matrixes of the updateFitting
It completes；Conversely, utilizing the optimization transition matrix updated if not meeting the condition of convergenceIt replys and executes step S5~S7,
Transition matrix is optimized for carrying out convergence assessment, until convergence to update again.
It should be noted that being to make each data point of the first simplified point cloud data P' sharp whenever turning again to step S5
With the three-dimensional point cloud index structure of step S3, hunt out through previous optimization transition matrix (that is: the optimization of kth -1
Transition matrix) conversion after second simplify point cloud data Q' corresponding data point；And in step s 6, the used conversion square
Battle array M is previous optimization transition matrix, that is, optimizes transition matrix using kth -1 and generate a new update again
Transition matrix (that is: kth optimize transition matrix) is optimized, then calculates this and first simplifies point cloud data and to pass through kth best
Second after changing transition matrix conversion simplifies between point cloud data, the most narrow spacing between the tangent plane at corresponding two stroke count strong point
From, and so on.
About the judgement of the condition of convergence, illustrate as after.In step S6 generation the result is that the minimum range, it is assumed that ekIt is
After k iteration, two data point di、siBetween the minimum range have a worst error value ek, it is expressed as follows:
Then the condition of convergence is to be judged with kth time with kth -1 time worst error value, and the condition of convergence of the invention is | ek-
ek-1| < te, wherein te is a preset judgment value, if iteration repeatedly still fails to restrain, it is likely that because the number of iterations is excessively voluminous
Raw operation bottleneck, so the present invention can set an iteration upper limit number, when the data processing equipment 20 judges that the number of iterations reaches
When to the iteration upper limit number, iteration will be stopped, and will have in the multiple groups estimated conversion one group of minimal error as best
Transition matrix.
Step S8: label redundant points
Redundant points are the data point that two simplified point cloud data P', the Q' repeat capture, Fig. 2 C and Fig. 7 are please referred to, after fitting
It should overlap and generate an overlapping region 30, the first boundary 31 for simplifying point cloud data P' falls into the second simplified point cloud number
According in the range of Q', and the second boundary 32 for simplifying point cloud data Q' falls into this and first simplifies in the range of point cloud data P', should
Two pattern characteristics for simplifying point cloud data P', Q' are substantially overlapped.But because Tooth surface passes through digital scan, actually two stroke counts
Identical coordinate value may not be had according to the data point of repeat region, the present invention utilizes the corresponding data point of following formula Rule of judgment
Between minimum range assess whether as redundant points:
Wherein tRFor preset judgment value, the present invention will meet the data point d of above formula conditioni、siAs the two simplified point cloud
The duplicate redundant digit strong point of data P', Q'.Wherein, since two data point d have been established in step S5i、siCorresponding relationship, therefore can
Directly to calculate two data point d according to coordinate informationi、siThe distance between, for judging the distance whether lower than the judgment value
tR。
In conclusion after the present invention detects redundant points, using can above exclude redundant points, and then number is effectively reduced
According to carrying cost, and advanced optimize data transmission efficiency and data operation efficiency.Furthermore the present invention by reduced sampling,
Data volume and improving operational speed can be also effectively reduced in the technological means such as three-dimensional index structure, linear approximation.
Claims (5)
1. a kind of redundancy point detecting method for point cloud data fitting, which is characterized in that it is executed in a data processing equipment, it should
To receive point cloud data, this method includes data processing equipment line scanner:
Receive one first point cloud data and one second point cloud data with serial relation；
Reduced sampling is carried out to first point cloud data and second point cloud data, simplifies point cloud data to respectively become one first
Simplify point cloud data with one second；
The three-dimensional point cloud index structure for establishing the first simplified point cloud data, for searching each number in the second simplified point cloud data
Strong point is with each data point corresponding to the first simplified point cloud data；
An optimization transition matrix is established, it, should after the conversion for making the second simplified point cloud data pass through the optimization transition matrix
First simplifies between point cloud data and the tangent plane at the second simplified point cloud data corresponding two stroke count strong point with a most narrow spacing
From；
Judge the minimum range whether less than a judgment value；If so, two stroke count strong point is the data point of redundancy；If it is not,
The three-dimensional point cloud index structure for establishing the first simplified point cloud data is returned to, for searching each data point in second point cloud data
With correspond to this first simplify point cloud data each data point the step of, to carry out the interative computation of the optimization transition matrix.
2. the redundancy point detecting method for point cloud data fitting as described in claim 1, which is characterized in that the optimization turns
Change matrixIt is expressed as follows:
Wherein, k: the number of iterations；Rk: spin matrix；Tk: transposed matrix；αopt、βopt、γopt、txopt、tyopt、tzopt: it optimizes
Conversion parameter.
3. the redundancy point detecting method for point cloud data fitting as claimed in claim 2, which is characterized in that further judgement
Whether the first simplified point cloud data meets one with the second worst error value for simplifying point cloud data corresponding two stroke count strong point
The condition of convergence；If it is not, the optimization transition matrix is recalculated, until meeting the condition of convergence；
The condition of convergence is expressed as follows:
|ek-ek-1| < te；
Wherein, ek: the worst error value obtained according to kth suboptimum transition matrix,diIt is
The data point of the first simplified point cloud data, siIt is the corresponding data point of the second simplification point cloud data；
ek-1: the worst error value obtained according to -1 suboptimum transition matrix of kth；
te: judgment value.
4. the redundancy point detecting method for point cloud data fitting as claimed in claim 3, which is characterized in that in judging this most
Whether small distance was less than in the step of judgment value, was that following formula judges whether the minimum range is redundant points:
Wherein, tRFor a judgment value.
5. the redundancy point detecting method for point cloud data fitting as claimed in claim 4, which is characterized in that utilize linear solution
The means of approaching establish the optimization transition matrix.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW106131674A TWI625700B (en) | 2017-09-15 | 2017-09-15 | Redundant point detection method for point cloud data bonding |
| TW106131674 | 2017-09-15 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109508141A true CN109508141A (en) | 2019-03-22 |
| CN109508141B CN109508141B (en) | 2022-02-18 |
Family
ID=63255703
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810347829.7A Active CN109508141B (en) | 2017-09-15 | 2018-04-18 | Redundant point detection method for point cloud data fitting |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN109508141B (en) |
| TW (1) | TWI625700B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112417965A (en) * | 2020-10-21 | 2021-02-26 | 湖北亿咖通科技有限公司 | Laser point cloud processing method, electronic device and storage medium |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11288493B2 (en) | 2020-02-25 | 2022-03-29 | Raytheon Company | Point cloud registration with error propagation |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102607459A (en) * | 2012-03-06 | 2012-07-25 | 中国矿业大学(北京) | Splicing method and splicing device of Lidar measurement data |
| CN105989604A (en) * | 2016-02-18 | 2016-10-05 | 合肥工业大学 | Target object three-dimensional color point cloud generation method based on KINECT |
| CN106725966A (en) * | 2015-11-20 | 2017-05-31 | 财团法人金属工业研究发展中心 | Three-dimensional dental arch fitting method |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101236660A (en) * | 2008-03-06 | 2008-08-06 | 张利群 | Three-dimensional scanners and its three-dimensional model reconfiguration method |
| CN104794687A (en) * | 2014-01-20 | 2015-07-22 | 鸿富锦精密工业（深圳）有限公司 | Point clouds simplifying system and method |
| CN105654422A (en) * | 2015-12-23 | 2016-06-08 | 北京观著信息技术有限公司 | Point cloud registration method and system |
-
2017
- 2017-09-15 TW TW106131674A patent/TWI625700B/en active
-
2018
- 2018-04-18 CN CN201810347829.7A patent/CN109508141B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102607459A (en) * | 2012-03-06 | 2012-07-25 | 中国矿业大学(北京) | Splicing method and splicing device of Lidar measurement data |
| CN106725966A (en) * | 2015-11-20 | 2017-05-31 | 财团法人金属工业研究发展中心 | Three-dimensional dental arch fitting method |
| CN105989604A (en) * | 2016-02-18 | 2016-10-05 | 合肥工业大学 | Target object three-dimensional color point cloud generation method based on KINECT |
Non-Patent Citations (1)
| Title |
|---|
| 王永波，等.: "线状特征约束下基于四元数描述的LiDAR点云配准方法", 《武汉大学学报 信息科学版》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112417965A (en) * | 2020-10-21 | 2021-02-26 | 湖北亿咖通科技有限公司 | Laser point cloud processing method, electronic device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| TW201915953A (en) | 2019-04-16 |
| TWI625700B (en) | 2018-06-01 |
| CN109508141B (en) | 2022-02-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109816703B (en) | Point cloud registration method based on camera calibration and ICP algorithm | |
| CN106845515B (en) | Robot target identification and pose reconstruction method based on virtual sample deep learning | |
| CN104050390B (en) | Mobile robot path planning method based on variable-dimension particle swarm membrane algorithm | |
| CN103196370B (en) | Measuring method and measuring device of conduit connector space pose parameters | |
| CN102682103B (en) | Three-dimensional space index method aiming at massive laser radar point cloud models | |
| CN111222418B (en) | Crowdsourcing data rapid fusion optimization method for multiple road segments of lane line | |
| Bampas et al. | Almost optimal asynchronous rendezvous in infinite multidimensional grids | |
| Fan et al. | Automated view and path planning for scalable multi-object 3D scanning | |
| CN106600661B (en) | The method for accurately generating segmental arc geologic section | |
| CN108305289B (en) | Three-dimensional model symmetry characteristic detection method and system based on least square method | |
| CN112712596B (en) | Dense matching point cloud building structured model fine reconstruction method | |
| CN109508141A (en) | Redundant point detection method for point cloud data fitting | |
| CN111192364A (en) | Low-cost mobile multi-robot vision simultaneous positioning and map creating method | |
| CN108648277A (en) | A kind of fast reconstructing method of laser radar point cloud data | |
| CN108389502A (en) | A kind of method and device for drawing traffic network | |
| JP2020101437A (en) | Three-dimensional position/attitude recognition device and method | |
| CN107123138B (en) | Based on vanilla-R point to the point cloud registration method for rejecting strategy | |
| CN110363801B (en) | Method for matching corresponding points of workpiece real object and three-dimensional CAD (computer-aided design) model of workpiece | |
| CN106803266A (en) | A kind of hull complexity outside plate point cloud costal line extracting method and device | |
| CN111858785B (en) | Map discrete element matching method, device, system and storage medium | |
| CN115861563B (en) | Three-dimensional reconstruction method for map topological rigid point cloud registration | |
| Stockman et al. | Sensing and recognition of rigid objects using structured light | |
| CN117152374A (en) | Digital twin establishing method of efficient cable dome structure based on three-dimensional laser point cloud | |
| CN116476070A (en) | Method for adjusting scanning measurement path of large-scale barrel part local characteristic robot | |
| US12045939B2 (en) | Multi-agent path planning method and device, and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |