CN110211145A - A kind of framework extraction method based on the careless model of reversed burning - Google Patents
A kind of framework extraction method based on the careless model of reversed burning Download PDFInfo
- Publication number
- CN110211145A CN110211145A CN201910456241.XA CN201910456241A CN110211145A CN 110211145 A CN110211145 A CN 110211145A CN 201910456241 A CN201910456241 A CN 201910456241A CN 110211145 A CN110211145 A CN 110211145A
- Authority
- CN
- China
- Prior art keywords
- point
- model
- axial plane
- reversed
- skeleton
- 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
- G06T7/10—Segmentation; Edge detection
- G06T7/149—Segmentation; Edge detection involving deformable models, e.g. active contour models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30172—Centreline of tubular or elongated structure
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention relates to a kind of based on the reversed framework extraction method for burning careless model, step are as follows: (1) generate the middle axial plane of threedimensional model；The middle axial plane refers to the centre of sphere set of all maximum inscribed spheres inside threedimensional model；(2) careless model centering axial plane each point progress centrality measurement is burnt based on reversed；(3) final skeleton curve is generated with minimal set cover algorithm.The present invention generates skeleton curve to extract the centrality feature of threedimensional model, has the advantages that easy, efficient, stability is good.
Description
Technical field
The invention belongs to the fields such as computer graphics and computer vision, and in particular to one kind burns careless model based on reversed
Framework extraction method.
Background technique
In the application of computer animation, three-dimensional modeling, three-dimensional search and human body gesture and action recognition etc., often make
The brief expression of threedimensional model is carried out, with skeleton to promote treatment effeciency.Threedimensional model skeletal extraction is computer vision and figure
The Basic Problems of shape field classics, are widely used.
The concept of skeleton proposed that he proposed skeleton in initiative paper [1] in 1967 by the Blum of MIT earliest
A kind of careless model definition of burning: hypothesized model is made of bundle of hay, is lighted a fire simultaneously on all boundary points of model, flame can be even
Fast ground to model internal communication, then the flame from boundary different piece can in a model between position meet and extinguish, own
Extinguish point and constitutes the skeleton of model.Later, the concept of skeleton was expanded to three-dimensional space by two-dimensional surface naturally by people, was mentioned
Many threedimensional model framework extraction methods are gone out.The thinking that most three-dimensional framework extracting methods are in accordance with the careless model of burning is set
The propagation frame of meter one from outside to inside, to extract the centrality feature of model, such as topological thinning method [2] [3] [4], iteration
Burning grass simulation [8] on shrinkage method [5] [6] [7], middle axial plane and the method [9] [10] based on mass transfer model.However,
These methods are not required to complicated global calculation, and exactly to the processing of each point, there are relation of interdependence, thus can not height
Parallelization handles each point, so their computational efficiency is not high.
In many practical applications of skeleton, time efficiency be influence Application effect an important factor for, it is even decisive
Factor.Therefore, the computational efficiency for improving skeletal extraction proposes quick skeletal extraction technology, and there is very important reality to answer
With value.
[1]H.Blum.A transformation for extracting new descriptors of
shape.Models for the Perception of Speech and Visual Form,MIT Press,1967.
[2]K.Palagyi and A.Kuba.A 3d 6-subiteration thinning algorithmfor
extracting medial lines.Pattern Recogn.Lett,9(7):613–627,1998.
[3]L.Liu,E.W.Chambers,D.Letscher,and T.Ju.A simple and robust
thinning algorithm on cell complexes.Computer Graphics Forum,29(7):2253–2260,
2010.
[4]M.Couprie and G.Bertrand.Asymmetric parallel 3d thinning scheme
and algorithms based on isthmuses.Pattern Recognition Letters,76:22–31,2016
[5]O.K.-C.Au,C.-L.Tai,H.-K.Chu,D.Cohen-Or,and T.-Y.Lee.Skeleton
extraction by mesh contraction.ACM Transactions on Graphics,27(3):44:1–44:10,
2008.
[6]A.Tagliasacchi,I.Alhashim,M.Olson,and H.Zhang.Mean curvature
skeletons.Computer Graphics Forum,31(5):1735–1744,2012.
[7]H.Huang,S.Wu,D.Cohen-Or,M.Gong,H.Zhang,G.Li,and B.Chen.L1-medial
skeleton of point cloud.ACM Transactions on Graphics,32(4):65:1–65:8,2013.
[8]Y.Yan,K.Sykes,E.Chambers,D.Letscher,and T.Ju.Erosion thickness on
medial axes of 3D shapes.ACM Transactions on Graphics,35(4):38:1–38:12,2016.
[9]A.C.Jalba,A.Sobiecki,and A.C.Telea.An unified multiscale framework
for planar,surface,and curve skeletonization.IEEE Transactions on Pattern
Analysis and Machine Intelligence,38(1):30–45,2016.
[10]D.Reniers,J.van Wijk,and A.Telea.Computing multiscale curve and
surface skeletons of genus 0shapes using a global importance measure.IEEE
Transactions on Visualization and Computer Graphics,14(2):355–368,2008.
Summary of the invention
The technology of the present invention solves the problems, such as: to overcome the framework extraction method computational efficiency based on the careless model of burning is not high to ask
Topic provides a kind of framework extraction method based on the careless model of reversed burning, to extract the centrality feature of threedimensional model, and then generates
Skeleton curve has the advantages that easy, efficient, stability is good.
Technical scheme is as follows: a kind of based on the reversed framework extraction method for burning careless model, step includes:
(1) the middle axial plane of threedimensional model is generated；The middle axial plane refers to the ball of all maximum inscribed spheres inside threedimensional model
Heart set；
(2) careless model centering axial plane each point progress centrality measurement is burnt based on reversed, the skeleton for obtaining model center is candidate
Point；
(3) minimal set cover algorithm is used, filters out expression model topological structure and main body from skeleton candidate point concentration
The skeleton vertex of content, generates final skeleton curve.
The step (1) specifically:
(11) threedimensional model interior point is calculated to the minimum distance on boundary, i.e. distance field；
(12) the ridge point for detecting the distance field forms axial plane in initial；
(13) from each point of initial middle axial plane, path tracing is carried out along distance field gradient direction, and will be on path
Point axial plane in initial is added, form the middle axial plane being completely connected to.
In the step (2), based on the reversed centrality for burning each point of axial plane in careless model realization propagated from inside to outside
Measurement, specific steps are as follows:
(21) for each measurement point, the process that the flame since the point is successively propagated outward is simulated, and records fire
Frontal line；
(22) when fiery frontal line is broken into multiple disconnected segmentations, stop measurement, at this time the burning time of fire experience
Centrality measured value as fire source point.Here, fiery frontal line disconnection need to be recurred repeatedly, just stop measurement, to protect
Demonstrate,prove the stability of measurement.
It states in step (3), with minimal set cover (Minimum Set Cover) algorithm, fully automatically deleting madel
Geometric detail and the corresponding central point of noise, while fully automatically choosing the bone for capableing of expression model topological structure and body matter
Frame vertex, thus generates skeleton curve, the specific steps are as follows:
(31) according in axial plane each point centrality measured value, obtain skeleton candidate's point set at its center, and with each time
The maximum inscribed sphere (maximal inscribed sphere) of model is generated centered on reconnaissance；
(32) minimal set cover algorithm is used, minimal number of inscribed sphere covering skeleton candidate point set is chosen；
(33) according to the neighbouring relations between the inscribed sphere of selection, it is connected to their central point, obtains skeleton；
(34) for the skeleton line that is located on middle axial plane difference manifold block still may disconnected situation, based on these streams
Intersection between shape block connects them, effectively to reflect the topological structure of model.
The advantages of the present invention over the prior art are that: the present invention is based on the careless model of the reversed burning propagated from inside to outside is real
The centrality measurement of axial plane each point, the measurement of each point are independent of each other in existing, therefore can be completely independent, handle to highly-parallel respectively
Point.Simultaneously as whether middle axial plane is blown can be determined by the connectivity of fiery frontal line (fire front), i.e., middle axial plane quilt
Fire frontal line also will disconnect into multiple segmentations when blowing, therefore the measurement of each point can be realized by terms of localization approach completely, be kept away
The global calculation that existing method relies on is exempted from.So the present invention can be obviously improved the calculating of skeletal extraction compared with the prior art
Efficiency promotes amplitude up to as many as 3 orders of magnitude.
Detailed description of the invention
Fig. 1 is flow chart of the method for the present invention；Wherein (a) is that axis face generates schematic diagram, is (b) the measurement signal of center property
Figure (c) generates schematic diagram for skeleton curve；
Fig. 2 is in the present invention based on the reversed middle axial plane centrality instrumentation plan for burning careless model；
Fig. 3 is the simulation algorithm schematic diagram that Flame of the present invention is propagated from inside to outside；
Fig. 4 is the processing method schematic diagram of axial plane in non-manifold in the present invention；Wherein (a) indicates the fiery frontal line disconnected point
Section is not belonging to the case where manifold dough sheet where fire source point completely, (b) indicates another situation opposite with (a)；
Fig. 5 is the connection algorithm schematic diagram of the ridge point in the present invention on middle axial plane difference manifold block；
Fig. 6 is the contrast and experiment figure of method and ET method of the invention；
Fig. 7 is that method of the invention is the skeleton result figure that many different types of threedimensional models extract.
Specific embodiment
As shown in Figure 1, the present invention has 3 steps:
(1) generate the middle axial plane of threedimensional model: middle axial plane (medial surface) refers to that threedimensional model inside is all most
The centre of sphere set of big inscribed sphere, if generate that a minimum distance by each point to boundary forms inside threedimensional model away from
It leaves the theatre, then middle axial plane can also be defined as the set of all ridge points (ridges) of the distance field.Therefore, by calculating three-dimensional mould
Distance field inside type generates wherein axial plane.
(2) it burns careless model centering axial plane each point based on reversed and carries out centrality measurement: the point at arbitrary point inside the middle axial plane
Fire, flame will burn from inside to outside, burning time needed for blowing middle axial plane, that is, blow the time, can be used as in the point
Disposition measured value.This is because on middle axial plane center line (ideal skeleton curve) each point blow the time can be less than they week
That encloses the point of disalignment blows the time.
(3) final skeleton curve is generated with minimal set cover algorithm: the centrality measured value meeting of middle axial plane each point
A scalar field is formed, is taken referred to herein as SYMMETRICAL FIELD in order to allow the point closer to center that there is higher centrality measured value
Each point blows value of the negative as SYMMETRICAL FIELD of time.Obviously, the ridge point (ridges) of SYMMETRICAL FIELD constitutes in middle axial plane
Heart line, therefore, using them as the candidate point set of skeleton.Then, it is calculated with minimal set cover (Minimum Set Cover)
Method concentrates automatically deleting madel geometric detail and the corresponding central point of noise from candidate point, while automatic choose can express
The skeleton vertex of model topology structure and body matter, to generate to the very stable skeleton curve of noise.
Each step of the invention is introduced respectively below.
1. generating the middle axial plane of threedimensional model
(1) threedimensional model interior point is calculated to the minimum distance on boundary, i.e. distance field.
(2) Local modulus maxima of detecting distance field forms axial plane in initial.
(3) from each point of initial middle axial plane, path tracing is carried out along distance field gradient direction, and will be on path
Axial plane in initial is added in point, forms the middle axial plane being completely connected to.
2. burning careless model centering axial plane each point progress centrality measurement based on reversed
Middle axial plane is divided into two kinds of situations of manifold and non-manifold, introduces the processing to both of these case separately below.
Handle manifold:
As shown in Fig. 2, an if point p on middle axial plane center line (ideal skeleton curve)1Igniting, then flame will
Middle axial plane blows the required time, that is, blows the time, be less than the point p of disalignment around it2Blow the time.Therefore,
The time needed for axial plane in blowing after lighting a fire at each point of middle axial plane is calculated, and is surveyed in this, as the centrality of middle axial plane each point
Amount.
In order to which calculate each point blows the time, an algorithm is designed to simulate the communication process of flame from inside to outside.Algorithm
Basic ideas it is as shown in Figure 3, it is assumed that minimum distance between consecutive points is unit 1, while flame is propagated between consecutive points
The minimum burning time needed is also unit 1, if P (t) indicates the point set that fire covers within the burning time of t unit,
Indicate all and P (t) adjacent external point set, the i.e. outer boundary of P (t).It updates from P (t) to P (t+1), only needs to detectIn point, the time that its moderate heat is reached less than or equal to t+1 point be added P (t).This is because fire reaches
The time at midpoint is greater than t, so that fire reachesThe time of external point is greater than t+1 certainly.Algorithm can be searched by breadth First
Rope (BFS) frame conveniently realizes, only need to iteratively update P (t) andUntilUntil not being connected to.Each iteration,
It willIn partial dot be added P (t) with update arrive P (t+1), this partial dot being newly added be flame traveled at the t+1 moment
The point reached.Meanwhile the external consecutive points of these new addition points being addedIt is arrived with updatingIn addition, entire
In iterative process, record fire reaches the propagation path of each point, i.e. shortest path from source point in real time.
Handle non-manifold:
For the middle axial plane of non-manifold, centrality measurement should execute on the manifold block where fire source point.In non-manifold
The centrality Measurement Algorithm of axial plane is as shown in figure 4, when the frontal line of fire isolates disconnected segmentation, if the segmentation is complete
The manifold block being not belonging to where fire source point, then ignore it, fire allowed to burn away；Otherwise, it means that manifold block where source point is
It is blown, it is corresponding to blow the centrality measured value that the time is exactly source point.Here, whether the segmentation that frontal line is isolated belongs to fire
Manifold block where source point is determined by the combustion path (most short propagation path) of each point on source point to frontal line.Such as Fig. 4
In shown in (a), if the combustion path of each point and the burning that certain is put on source point to other segmentations on fire source point p to the segmentation
Intersect (intersection point q), then it is assumed that the segmentation isolated is not belonging to the manifold block where fire source point in path.(b) is illustrated in Fig. 4
Another opposite situation.
3. generating final skeleton curve with minimal set cover algorithm
After the centrality measurement for completing centering axial plane each point, all measured values (blowing the time) can form one on middle axial plane
A scalar field (referred to as SYMMETRICAL FIELD) takes each point to blow to allow the point closer to center to have higher centrality measured value
Value of the negative of time as SYMMETRICAL FIELD.It is then detected that the ridge point (ridges) of SYMMETRICAL FIELD is used as skeleton candidate point.Ridge point
Detection be divided into two steps: if (1) certain point a neighborhood in, the SYMMETRICAL FIELD value of the point than in its neighborhood 90% other
The SYMMETRICAL FIELD value of point will be big, then the point is labeled as candidate ridge point；(2) from all candidate ridge points, along symmetrical
The gradient direction of field does path tracing, to be connected to them.
After handling in this way, the ridge point on middle axial plane difference manifold block still may not be connected to.In this regard, using one
Simple algorithm is connected to them.The basic ideas of algorithm are as shown in figure 5, axis surface current shape the block A and B connected for two, first
Find the intersection point p of manifold block intersection and the ridge line of A, it is clear that p is also on manifold block B.Then, find on the ridge line of B with p away from
From nearest point q.Finally, connecting the ridge point of manifold block A and B by the shortest path from p to q.
After all ridge points are obtained as the candidate point set of skeleton, the maximum inscribe of model is generated at each candidate point
Ball, then with the minimal number of inscribed sphere of minimal set cover algorithm picks to cover entire candidate point set.
In order to which the inscribed sphere quantity for covering set is as few as possible, algorithm can be automatically big at preferential Selection Model trunk
Ball, and give up geometric detail and the corresponding bead of noise, as shown in figure 1 shown in (c).Finally, according to the adjacent of the inscribed sphere of selection
Their central point is together in series by relationship, to generate to the good skeleton curve of the stability of noise.
Here is some experimental datas of the invention:
Experiment carried out on a PC, the microcomputer be equipped with Intel i7-2600 (3.2GHz) CPU, 16G RAM and
One NVIDIA GeForce GTX 1080GPU.It tests model used and is all from the common Princeton Shape of educational circles
Benchmark and Aim@Shape model library.
First with careless simulation (referred to as " ET is burnt on the middle axial plane of representative framework extraction method-that is recently proposed in the world
Method ", referring to document [8]) experiment is compared, threedimensional model used is as shown in Figure 6.Two methods extraction has been counted in following table
The runing time of each model skeleton, experiment show as many as fast 3 orders of magnitude of method ratio ET method of the invention.Fig. 6 is illustrated
Skeleton obtained by two methods needs to manually adjust threshold value to bone as a result, the skeleton that ET method generates is likely to occur extra branch
Frame branch is trimmed, and method of the invention can automatically generate succinct, high quality skeleton without manually adjusting threshold value.
| Input threedimensional model | The time-consuming of ET method extraction skeleton | The time-consuming of the method for the present invention extraction skeleton | Speed-up ratio |
| Bear | 154 seconds | 0.038 second | 4052 |
| Ant | 129 seconds | 0.045 second | 2866 |
| Horse | 140 seconds | 0.066 second | 2121 |
| Vase | 230 seconds | 0.054 second | 4259 |
In addition, it is the skeleton that extracts of more different types of threedimensional models as a result, either that Fig. 7, which illustrates the method for the present invention,
The very complicated model of topological structure, or the model with many noises are inputted, the method for the present invention can generate clean succinct
High quality skeleton.As it can be seen that the technology of the present invention has excellent universality and stability.
Non-elaborated part of the present invention belongs to techniques well known.
The above, part specific embodiment only of the present invention, but scope of protection of the present invention is not limited thereto, example
Method as described herein can be used for extracting the skeleton of two dimensional model.Any those skilled in the art the invention discloses
Technical scope in, any changes or substitutions that can be easily thought of, should be covered by the protection scope of the present invention.
Claims (6)
1. a kind of based on the reversed framework extraction method for burning careless model, which is characterized in that step includes:
(1) the middle axial plane of threedimensional model is generated；The middle axial plane refers to the centre of sphere collection of all maximum inscribed spheres inside threedimensional model
It closes；
(2) careless model centering axial plane each point progress centrality measurement is burnt based on reversed, obtains the skeleton candidate point of model center；
(3) minimal set cover algorithm is used, filters out expression model topological structure and body matter from skeleton candidate point concentration
Skeleton vertex, generate final skeleton curve.
2. according to claim 1 based on the reversed framework extraction method for burning careless model, it is characterised in that: the step
(1) specifically:
(11) threedimensional model interior point is calculated to the minimum distance on boundary, i.e. distance field；
(12) the ridge point for detecting the distance field forms axial plane in initial；
(13) each point of axial plane from initial carries out path tracing along distance field gradient direction, and by the point on path
Axial plane in initial is added, forms the middle axial plane being completely connected to.
3. according to claim 1 based on the reversed framework extraction method for burning careless model, it is characterised in that: the step
(2) in, based on the reversed centrality measurement for burning each point of axial plane in careless model realization propagated from inside to outside, specific steps are as follows:
(21) for each measurement point, the process that the flame since the point is successively propagated outward is simulated, and records the forward position of fire
Line；
(22) when fiery frontal line is broken into disconnected segmentation, stop measurement, the burning time of fire experience is as fire source at this time
The centrality measured value of point.
4. according to claim 1 based on the reversed framework extraction method for burning careless model, it is characterised in that: the step
(2) in, when based on the reversed centrality measurement for burning each point of axial plane in careless model realization propagated from inside to outside, fiery frontal line
When continuous several times being needed to be broken as disconnected segmentation, just stop measurement, to guarantee the stability of centrality measurement.
5. according to claim 1 based on the reversed framework extraction method for burning careless model, it is characterised in that: the step
(2) in, the centrality measured value based on middle axial plane each point obtains the skeleton candidate point of model center, specific steps are as follows:
(23) according in axial plane each point centrality measured value, obtain the centerline of each dough sheet of middle axial plane；
(24) dough sheet for detecting all intersections is connected them with shortest path if their centerline is not connected to.
6. according to claim 1 based on the reversed framework extraction method for burning careless model, it is characterised in that: the step
(3) in, with minimal set cover (Minimum Set Cover) algorithm, fully automatically deleting madel geometric detail and noise
Corresponding central point, while the skeleton vertex for capableing of expression model topological structure and body matter is fully automatically chosen, thus give birth to
At skeleton curve, the specific steps are as follows:
(31) for obtained skeleton candidate's point set, and centered on each candidate point generate model maximum inscribed sphere
(maximal inscribed sphere)；
(32) minimal set cover algorithm is used, minimal number of inscribed sphere covering skeleton candidate point set is chosen；
(33) according to the neighbouring relations between the inscribed sphere of selection, their central point of connecting obtains skeleton.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910456241.XA CN110211145B (en) | 2019-05-29 | 2019-05-29 | Framework extraction method based on reverse burning grass model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910456241.XA CN110211145B (en) | 2019-05-29 | 2019-05-29 | Framework extraction method based on reverse burning grass model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110211145A true CN110211145A (en) | 2019-09-06 |
| CN110211145B CN110211145B (en) | 2021-04-02 |
Family
ID=67789454
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910456241.XA Active CN110211145B (en) | 2019-05-29 | 2019-05-29 | Framework extraction method based on reverse burning grass model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110211145B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110929608A (en) * | 2019-11-12 | 2020-03-27 | 国网天津市电力公司 | Safety helmet wearing detection method and system |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101763652A (en) * | 2009-06-03 | 2010-06-30 | 中国科学院自动化研究所 | Three-dimensional framework fast extraction method based on branch feathers |
| CN102402796A (en) * | 2011-10-26 | 2012-04-04 | 重庆大学 | Three-dimensional structural description method of hepatic vascular system |
| CN107164318A (en) * | 2017-07-19 | 2017-09-15 | 叶川 | Method and hollow form vascularised heart based on 3D biometric print technique construction hollow form vascularised hearts |
| CN107507171A (en) * | 2017-08-08 | 2017-12-22 | 东北大学 | A kind of lung CT image air flue three-dimensional framework tree extraction and labeling method |
| CN107730521A (en) * | 2017-04-29 | 2018-02-23 | 安徽慧视金瞳科技有限公司 | The quick determination method of roof edge in a kind of image |
| CN109215128A (en) * | 2018-08-09 | 2019-01-15 | 北京华捷艾米科技有限公司 | The synthetic method and system of object motion attitude image |
| US20190014982A1 (en) * | 2017-07-12 | 2019-01-17 | iHealthScreen Inc. | Automated blood vessel feature detection and quantification for retinal image grading and disease screening |
-
2019
- 2019-05-29 CN CN201910456241.XA patent/CN110211145B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101763652A (en) * | 2009-06-03 | 2010-06-30 | 中国科学院自动化研究所 | Three-dimensional framework fast extraction method based on branch feathers |
| CN102402796A (en) * | 2011-10-26 | 2012-04-04 | 重庆大学 | Three-dimensional structural description method of hepatic vascular system |
| CN107730521A (en) * | 2017-04-29 | 2018-02-23 | 安徽慧视金瞳科技有限公司 | The quick determination method of roof edge in a kind of image |
| US20190014982A1 (en) * | 2017-07-12 | 2019-01-17 | iHealthScreen Inc. | Automated blood vessel feature detection and quantification for retinal image grading and disease screening |
| CN107164318A (en) * | 2017-07-19 | 2017-09-15 | 叶川 | Method and hollow form vascularised heart based on 3D biometric print technique construction hollow form vascularised hearts |
| CN107507171A (en) * | 2017-08-08 | 2017-12-22 | 东北大学 | A kind of lung CT image air flue three-dimensional framework tree extraction and labeling method |
| CN109215128A (en) * | 2018-08-09 | 2019-01-15 | 北京华捷艾米科技有限公司 | The synthetic method and system of object motion attitude image |
Non-Patent Citations (9)
| Title |
|---|
| BITTER I,AT EL.: "" Penalized distance volumetric"", 《 IEEE TRANSACTIONS ON VISUALIZATION AND》 * |
| BLUM: "《Models for the Perception of Speech and Visual form》", 31 December 1967 * |
| CHVATAL V: "" A greedy heuristic for the set-covering problem"", 《MATHEMATICS OF OPERATIONS RESEARCH》 * |
| JING LI,AT EL.: ""Fast and robust GPU-based point-in-polyhedron determination"", 《COMPUTER-AIDED DISIGN 》 * |
| MAURER C R,AT EL.: "" A linear time algorithm for computing exact Euclidean distance transforms of binary images in arbitrary dimensions"", 《IEEE》 * |
| YUANXUE CAI,AT EL.: ""Skeleton extraction based on the topology and Snake model"", 《RESULT IN PHYSICS》 * |
| 曹嘉: ""三维点云模型骨架提取算法的研究与实现"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 李雷等: ""构建中值图以快速生成高质量的三维模型骨架"", 《计算机辅助设计与图形学学报》 * |
| 黄慷明: ""最优质量传输理论在点云骨架提取中的应用研究"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110929608A (en) * | 2019-11-12 | 2020-03-27 | 国网天津市电力公司 | Safety helmet wearing detection method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110211145B (en) | 2021-04-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107767453B (en) | Building LIDAR point cloud reconstruction optimization method based on rule constraint | |
| CN108319957A (en) | A kind of large-scale point cloud semantic segmentation method based on overtrick figure | |
| CN102853803B (en) | Testing method of damaged area of cultural relic | |
| CN102945569B (en) | Three-dimensional model symmetry analysis method based on heat kernel signal | |
| CN105718528A (en) | Academic map display method based on reference relationship among thesises | |
| TWI425442B (en) | Method of Reconstructing Three - Dimensional Housing Model on Aeronautical Mapping System | |
| CN103236077B (en) | The three-dimensional model line method for drafting that a kind of feature strengthens | |
| CN109492796A (en) | A kind of Urban Spatial Morphology automatic Mesh Partition Method and system | |
| CN104077806A (en) | Automatic separate extraction method based on city building three-dimensional model | |
| CN115018982B (en) | Tree digital twin method based on foundation laser radar point cloud | |
| CN103559689A (en) | Removal method for point cloud noise points | |
| CN107610228A (en) | Curved surface increment topology rebuilding method based on massive point cloud | |
| CN112288859B (en) | Three-dimensional face modeling method based on convolutional neural network | |
| CN108875819A (en) | A kind of object and component associated detecting method based on shot and long term memory network | |
| CN114119922A (en) | Complex contour defect checking method based on point cloud three-dimensional reconstruction | |
| CN110211145A (en) | A kind of framework extraction method based on the careless model of reversed burning | |
| CN103824325A (en) | Fruit tree branch interactive three-dimensional reconstruction method and system | |
| CN117011465A (en) | Tree three-dimensional reconstruction method and device, electronic equipment and storage medium | |
| CN112966390B (en) | Method and apparatus for garment processing based on dual three-dimensional distance fields | |
| CN107643820A (en) | The passive humanoid robots of VR and its implementation method | |
| CN107644686A (en) | Medical data acquisition system and method based on virtual reality | |
| CN106023161A (en) | Automatic association method for broken ridge lines | |
| CN110322415A (en) | High-precision surface three-dimensional reconstruction method based on point cloud | |
| CN105139452A (en) | Geological curved surface reconstruction method based on image segmentation | |
| CN108256578A (en) | A kind of recognition methods of gray level image, device, equipment and readable storage medium storing program for executing |
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 |