CN105513136B - A kind of 3D actor model framework extraction methods based on level set central cluster - Google Patents
A kind of 3D actor model framework extraction methods based on level set central cluster Download PDFInfo
- Publication number
- CN105513136B CN105513136B CN201510868466.8A CN201510868466A CN105513136B CN 105513136 B CN105513136 B CN 105513136B CN 201510868466 A CN201510868466 A CN 201510868466A CN 105513136 B CN105513136 B CN 105513136B
- Authority
- CN
- China
- Prior art keywords
- level set
- mrow
- msub
- point
- plane
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
- G06T17/30—Polynomial surface description
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2200/00—Indexing scheme for image data processing or generation, in general
- G06T2200/04—Indexing scheme for image data processing or generation, in general involving 3D image data
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Physics (AREA)
- Mathematical Analysis (AREA)
- Algebra (AREA)
- Computer Graphics (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Processing Or Creating Images (AREA)
Abstract
The present invention provides a kind of 3D actor model framework extraction methods based on level set central cluster, comprises the following steps：Using height function cutting process, the intersection point set after being cut are carried out as Morse function pair 3D actor models；Intersection point set after cutting is clustered according to the connectedness between intersection point, is clustered into some sub- level sets；The level set that area is more than level set area threshold and is more than seemingly circle property threshold value like circle property screens；Level set after screening is clustered using the center of level set, so as to be both arms, both legs and the trunk of 3D actor models by level-set segmentation；Obtained skeleton line is embedded into 3D actor models.The present invention can be by being automatically the extraction of three-dimensional animation model progress skeleton, level set clustering method used by skeletal extraction, compared to traditional range conversion method, by using Slicing Model for Foreign, partitioning model position, the methods of calculating each several part joint respectively, its average consumed time is less, error smaller.
Description
Technical field
The invention belongs to animation design technique field, more particularly to a kind of 3D role based on level set central cluster
Model framework extraction method.
Background technology
With the development of virtual reality technology and digital media technology, making of the people for three-dimensional character model animation
Demand it is also more and more, as the important component of multimedia technology, the technology of three-dimensional animation production obtains in recent years
Very big development, but mainly adopted still based on making by hand among the three-dimensional character manufacturing process of current mainstream
Be designed and made with the 3DMax either business softwares such as Maya, but animation creation industry at this stage be still faced with it is scarce
The problems such as outstanding resources of human talents of few suitable computer animation related fields knowledge and experience, therefore current three
Tie up in the manufacturing process of animation there is still a need for certain labour and cost is substantial amounts of mental.People are now using relatively broad
Computer Animated Graph have key frame and dynamic method etc..But these methods have a common defect：Make
Take time and effort, while need producer to possess the manufacture skill and professional knowledge of higher degree.
As there is demand, it is necessary to which a kind of easy mode is three-dimensional dynamic to carry out for three-dimensional character cartoon making in more people
The making of picture, to allow more people for lacking three dimensional character cartoon making experience to participate in this work, such as user oneself
Preference pattern, action etc., afterwards by the suitable animation of programming automatic generation.Therefore it is automatic to carry out three dimensional character animation design technique
Researcher or the research hotspot of engineers are become.It is so not only time saving and energy saving, moreover it is possible to allow more people to participate in dynamic
Draw in making.The skeletal extraction of model in motion retargeting technology is simplified and automated in this context, is become
Unusual the key link in role animation field.
The content of the invention
It is an object of the invention to provide a kind of 3D actor model framework extraction methods based on level set central cluster.
The technical scheme is that：
A kind of 3D actor model framework extraction methods based on level set central cluster, comprise the following steps：
Step 1：Cutting process is carried out as Morse function pair 3D actor models using height function, after being cut
Intersection point set；
Step 2：Intersection point set after cutting is clustered according to the connectedness between intersection point, is clustered into some sub- water
Flat collection：If what certain two intersection point was crossed to form for elevation plane and a triangular topological relations, by the two intersection points preserve into
One line segment, is clustered using the connectivity between line segment, obtains some sub- level sets；
Step 3：Screening conditions are used as using the area of level set and like circle property, by area more than level set area threshold and seemingly
The level set that circle property is more than seemingly circle property threshold value screens；
Step 4：Level set after screening is clustered using the center of level set, so as to be 3D by level-set segmentation
Both arms, both legs and the trunk of actor model；
Step 5：Obtained skeleton line is embedded into 3D actor models.
The step 1 comprises the following steps that：
Step 1-1：Height function is chosen 3D actor models are handled to obtain 3D actor model height hmodel；
Step 1-2：3D actor model cutting times n are set, cutting is proceeded by from the top of 3D actor models, every
hmodel/ n distances take a plane i.e. elevation plane；
Step 1-3：The triangular topological relations in FBX files are read, by maximum of each triangular topological relations in y-axis direction
ymaxWith minimum value yminIt is saved in array, to maximum y of each triangular topological relations in y-axis directionmaxIt is ranked up；
Step 1-4：Each elevation plane of 3D actor models and the intersection point of triangular topological relations are calculated, each height is obtained and puts down
The triangular topological relations list that face can intersect；
Step 1-5：The triangular topological relations list that can be intersected according to each elevation plane, determines that each elevation plane can with it
Intersection point between intersecting triangular topological relations, that is, the intersection point set after cutting.
The step 1-4's comprises the following steps that：
Step 1-4-1：Travel through each triangular topological relations in array, if current triangular topological relations in y-axis direction
Maximum ymaxMore than present level plane Pcurrent, then step 1-4-2 is performed, otherwise after current triangular topological relations
All triangular topological relations all no longer with present level plane PcurrentIt is intersecting, stop traversal, obtained triangular topological relations list is
Can be with present level plane PcurrentIntersecting triangular topological relations list；
Step 1-4-2：Judge the minimum value y in y-axis direction of current triangular topological relationsminWhether put down more than present level
Face Pcurrent：It is that then current triangular topological relations are in present level plane PcurrentOn, next triangular topological relations are traveled through, are returned
Step 1-4-1；Otherwise current triangular topological relations and present level plane PcurrentIt can intersect, preserve current triangular topological relations,
Travel through next triangular topological relations, return to step 1-4-1.
The step 3 comprises the following steps that：
Step 3-1：Set screening conditions, including level set area threshold and like circle property threshold value；
Step 3-2：Calculate the area of each level set and like circle property；
Step 3-3：The level set that area is more than level set area threshold and is more than seemingly circle property threshold value like circle property filters out
Come.
The step 4 comprises the following steps that：
Step 4-1：The central point of each level set is calculated, forms center point set；
Step 4-2：To all central points according to the another two direction in three dimensions apart from height axial length value into
Row sequence, the both arms direction of extension of direction, that is, 3D actor models of axial length value maximum；
Step 4-3：Calculate sequence after center point set preceding 5 central points adjacent center point between distance average value,
Afterwards between the adjacent center point of 5 central points distance average value, respectively using 2 times of the two average values as two clusters half
Footpath；
Step 4-4：First central point of the center point set after sequence is chosen respectively, last central point is used as starting
Point is clustered, and obtains the level set of the both arms of 3D actor models；
Step 4-5：All central points are ranked up by y direction is ascending, find the endpoint value of both legs, are carried out
The cluster identical with step 4-4, obtains the level set of the both legs of 3D actor models；
Step 4-6：Central point is concentrated into the central point not being clustered as in the level set of the trunk of 3D actor models
Heart point；
Step 4-7：Calculate the fitting a straight line of central point of the level set of the both arms of 3D actor models, the level set of both legs
The fitting a straight line of central point, the level set of trunk central point fitting a straight line respectively as the skeleton line of both arms, the bone of both legs
The skeleton line of stringing, trunk.
The step 5 comprises the following steps that：
Step 5-1：Calculate the position of crucial artis；
Crucial artis includes：Head, neck, shoulder joint, elbow joint, wrist joint, hand, hip joint, big leg joint, ankle
Joint and foot；
Step 5-2：Matched using between the position of crucial artis and standard skeleton structure, by between artis
Position proportional obtain the position of remaining all artis；
Step 5-3：Connect all semantically related artis and obtain final skeleton, be embedded into 3D actor models
In.
The step 5-1's comprises the following steps that：
Step 5-1-1：Head, hand, hip are tried to achieve by the end of the skeleton line of the 3D actor model each section after cutting
The position in joint, big leg joint and foot's artis：Two ends of leg skeleton line are big leg joint and foot；Arm skeleton
Line lower end is hand；Two ends of trunk skeleton line are head and hip joint；
Step 5-1-2：The level set obtained using the triangular topological relations of the both arms part of horizontal plane cutting 3D actor models
Girth, determine carpal position；
Step 5-1-3：Calculate the position of P point：Utilize trunk more than the method processing arm of step 5-1-2
The level set of the grid model of position, the position using the midpoint of the level set of girth minimum as P point；
Step 5-1-4：Calculate the position of shoulder joint：Intersect with plane where shoulder joint in direction using the skeleton line of both arms
An initial shoulder joint position is obtained to be modified again；
Step 5-1-5：Calculate the position of ankle-joint：The corresponding water in leg in the level set that selecting step 3 filters out first
Flat collection, then calculates angle between the midpoint connecting line of selected each adjacent level collection, not less than the angular position of given threshold
That is ankle-joint location.
The step 5-1-2's comprises the following steps that：
Step 5-1-2-1：Go out a spin moment using the direction of skeleton line and the angle calcu-lation of longitudinal axis positive direction of both arms
Battle array；
Spin matrix represents as follows：
Wherein, θ is the angle of direction and the longitudinal axis positive direction of the skeleton line of both arms, (nx,ny,nz) it is rotation axis, using double
The skeleton line of arm is tried to achieve with longitudinal axis positive direction multiplication cross；
Step 5-1-2-2：Both arms grid model is multiplied by the spin matrix, both arms grid model is upright；
Step 5-1-2-3：Cross-cutting operations are carried out to both arms grid model using horizontal plane, are obtained using the method for step 2
A series of level sets；
Step 5-1-2-4：If there are multiple level sets under some elevation plane, the vertex of all level sets is considered as
One point set, the convex closure of the point set is calculated using Graham algorithms, by the use of the convex closure as the elevation plane under unique level set；
Step 5-1-2-5：The girth of each level set is calculated, is closed the midpoint of the level set of girth minimum as wrist
The position of node.
The step 5-1-4's comprises the following steps that：
Step 5-1-4-1：The plane and P point determined using the point of the skeleton line longitudinal axis value maximum of both arms is indulged
The mid-plane for the plane that axis value determines, obtains plane residing for shoulder joint；
Step 5-1-4-2：Intersect with the plane where shoulder joint to obtain an intersection point using the direction of the skeleton line of both arms,
The intersection point is the position of shoulder joint；
Step 5-1-4-3：Using the position of shoulder joint and the position of wrist joint point, pass through elbow in standard skeleton structure
Joint obtains the initial position of elbow joint with shoulder joint, the position proportional of wrist joint point；
Step 5-1-4-4：Using the point more than the plane that elbow joint longitudinal axis value determines as upper arm part, by upper arm parts
The central point divided re-starts fitting a straight line, makes the straight line intersect to obtain modified shoulder joint position with plane where shoulder joint,
5-1-4-3 is repeated until the position of shoulder joint no longer changes, the shoulder joint node obtained at this time is final shoulder joint
Position.
Beneficial effect：
The present invention can be by being automatically the extraction of three-dimensional animation model progress skeleton, level set used by skeletal extraction
Clustering method, compared to traditional range conversion method, by using Slicing Model for Foreign, partitioning model position, calculates each several part respectively
The methods of joint, its average consumed time is less, error smaller.Calculated by this method instead of traditional-handwork,
It can make the reduction of user's threshold, it is easy to use, improve the precision and efficiency of skeletal extraction.
Brief description of the drawings
Fig. 1 is the schematic diagram of the method clustered to level set central point of the specific embodiment of the invention；
Fig. 2 is gathered the intersection point set after cutting according to the connectedness between intersection point for the specific embodiment of the invention
The flow chart of class；
Fig. 3 is the flow chart that the specific embodiment of the invention clusters level set central point；
Fig. 4 (a)~(d) is respectively the distance transformation method to this method and the prior art in the specific embodiment of the invention
Four models used by contrast；
Fig. 5 (a)~(h) is that the contrast of the distance transformation method of this method and the prior art is real in the specific embodiment of the invention
Test result figure；
Fig. 6 is 3D actor model framework extraction method flow of the specific embodiment of the invention based on level set central cluster
Figure；
Fig. 7 is the particular flow sheet of specific embodiment of the invention step 1；
Fig. 8 is the particular flow sheet of specific embodiment of the invention step 1-4；
Fig. 9 is the particular flow sheet of specific embodiment of the invention step 4；
Figure 10 is the particular flow sheet of specific embodiment of the invention step 5；
Figure 11 is the particular flow sheet of specific embodiment of the invention step 5-1-2.
Embodiment
Elaborate below in conjunction with the accompanying drawings to the embodiment of the present invention.
A kind of 3D actor model framework extraction methods based on level set central cluster, as shown in fig. 6, including following step
Suddenly：
Step 1：Cutting process is carried out as Morse function pair 3D actor models using height function, after being cut
Intersection point set；
As shown in fig. 7, comprise the following steps that：
Step 1-1：Height function is chosen 3D actor models are handled to obtain 3D actor model height hmodel；
Step 1-2：Set 3D actor model cutting times n=80, since the top (crown) of 3D actor models into
Row cutting, every hmodel/ n distances take a plane i.e. elevation plane；
Step 1-3：In order to improve efficiency, when reading the triangular topological relations in FBX files, each triangular topological relations is existed
The maximum y in y-axis directionmaxWith minimum value yminIt is saved in array, to maximum y of each triangular topological relations in y-axis directionmax
It is ranked up；
Step 1-4：Each elevation plane of 3D actor models and the intersection point of triangular topological relations are calculated, each height is obtained and puts down
The triangular topological relations list that face can intersect；
As shown in figure 8, comprise the following steps that：
Step 1-4-1：Travel through each triangular topological relations in array, if current triangular topological relations in y-axis direction
Maximum ymaxMore than present level plane Pcurrent, then step 1-4-2 is performed, otherwise after current triangular topological relations
All triangular topological relations all no longer with present level plane PcurrentIt is intersecting, stop traversal, obtained triangular topological relations list is
Can be with present level plane PcurrentIntersecting triangular topological relations list, performs step 1-5；
Step 1-4-2：Judge the minimum value y in y-axis direction of current triangular topological relationsminWhether put down more than present level
Face Pcurrent：It is that then current triangular topological relations are in present level plane PcurrentOn, next triangular topological relations are traveled through, are returned
Step 1-4-1；Otherwise current triangular topological relations and present level plane PcurrentIt can intersect, preserve current triangular topological relations,
Travel through next triangular topological relations, return to step 1-4-1.
Elevation plane has two intersection points with triangular topological relations intersection point, and it is as follows to calculate formula of intersection point：
X=x1+(Pcurrent-y1)*(x2-x1)÷(y2-y1)
Z=z1+(Pcurrent-y1)*(z2-z1)÷(y2-y1)
Wherein, x and z represents a line and present level plane P of triangular topological relations respectivelycurrentIntersection point x and z-axis
Value, x1With x2The x-axis value of two endpoints on this side, z are represented respectively1With z2Represent the z-axis value of two endpoints on this this side,
The y values of intersection point are present level plane PcurrentValue.
Step 1-5：The triangular topological relations list that can be intersected according to each elevation plane, determines that each elevation plane can with it
Intersection point between intersecting triangular topological relations, that is, the intersection point set after cutting.
Step 2：Intersection point set after cutting is clustered according to the connectedness between intersection point, is clustered into some sub- water
Flat collection：If what certain two intersection point was crossed to form for elevation plane and a triangular topological relations, by the two intersection points preserve into
One line segment, is clustered using the connectivity between line segment, obtains some sub- level sets；
Described the intersection point set after cutting in Fig. 2 according to the connectedness between intersection point the flow chart that is clustered,
Specific method flow is as follows：
Step 2-1：Record present level plane PcurrentLine segment aggregate L { the l being crossed to form with triangular topological relations1,
l2,…,ln}；
Step 2-2：Arbitrarily take lk, k ∈ [1, n], record lkAn endpoint be vstart, it is rising for whole ergodic process
Point, its another endpoint are vcurrent, be current search point, remember that current cluster point set is combined into Ei, i=1, by vstartAdd
It is added among current collection；
Step 2-3：Remaining all line segment in line segment aggregate L are traveled through, ifliWherein some vertex value
Equal to vcurrent, then l is takeniAnother endpoint be vcurrent, while delete l in line segment aggregate Li, while in EiMiddle addition
vcurrent, step 2-4 is carried out if L at this time is sky, if vcurrent=vstart, then step 2-2 is jumped to, and i is added 1,
If traversal terminates not form closed curve yet, v is setcurrentFor new vstart, step 2-3 is repeated, traversal is to the end
The head and the tail of junction curve form a closed curve, and the curve is preserved, and jump to step 2-4 if L is empty jumps, otherwise
Go to step 2-2；
Step 2-4：The corresponding intersection point set of each closed curve is obtained as each sub- level set.
Step 3：Screening conditions are used as using the area of level set and like circle property, by area more than level set area threshold and seemingly
The level set that circle property is more than seemingly circle property threshold value screens；
Step 3-1：Set screening conditions, including level set area threshold and like circle property threshold value；
The calculation formula of level set area threshold T is as follows：
Wherein, hmodelFor the height of 3D actor models, hstandardFor master pattern set in advance highly, in this implementation
It is the level set area threshold obtained in master pattern that 160, Round () function is taken as in mode as the function that rounds up, ST,
ST=0.2 is taken in present embodiment.
Step 3-2：Calculate the area of each level set and like circle property；
" by 2 lines are formed successively on any point and polygon DELTA vectors area and tried to achieve " using theorem,
The area of each level set is calculated, is calculated using equation below：
Wherein, S (P) represents the area of the polygon of sub- level set P compositions, and M is any point,For point M and Zi Shui
The vector to be formed is connected in flat collection P at i-th point,For point M points shape is connected with the i+1 intersection point in sub- level set P
Into vector.
Level set like circle property calculation formula be：
Wherein, C (P) represents the polygon for sub- level set P compositions like circle property, S (P) of the polygon of sub- level set P compositions
The area of shape, P (P) represent the girth of the polygon of sub- level set P compositions.
Step 3-3：The level set that area is more than level set area threshold and is more than seemingly circle property threshold value like circle property filters out
Come.Seemingly the threshold value of circle property rule of thumb, is taken as 0.4 in present embodiment.
Step 4：Level set after screening is clustered using the center of level set, so as to be 3D by level-set segmentation
Both arms, both legs and the trunk of actor model；
As shown in figure 9, comprise the following steps that：
Step 4-1：The central point of each level set is calculated, forms center point set；
Wherein C is the central point of level set P.
Step 4-2：To all central points according to the another two direction in three dimensions apart from height axial length value into
Row sequence, the both arms direction of extension of direction, that is, 3D actor models of axial length value maximum；
Step 4-3：Calculate the center point set C after sequenceallPreceding 5 central points adjacent center point between distance be averaged
Value, rear 5 central points adjacent center point between distance average value, respectively using 2 times of the two average values as two cluster
Radius；
Step 4-4：The center point set C after sequence is chosen respectivelyallFirst central point, last central point conduct
Starting point is clustered, and obtains the level set of the both arms of 3D actor models；
Fig. 1 depicts the schematic diagram of the method clustered to level set central point, wherein, starting point T1 is added poly-
In class set, judge whether T2 is being given within cluster radius with the element distance in cluster set, if there are at least one
Meet the situation of condition, then add T2 in cluster set, ensuing element and so on.
Fig. 3 is the particular flow sheet clustered to level set central point, and specific execution step is as follows：
Step 4-4-1：Queue Q is established, starting point Cstart is added among queue Q, and by Cstart labeled as
Through processing；
Step 4-4-2：The header element of queue Q is taken as the point Ccurrent currently judged, and Ccurrent is added to
In cluster result set Cresult；
Step 4-4-3：Traversal center point set CallIf the distance for the central point distance Ccurrent being currently traversed to
Less than cluster radius R and not processed mistake, then the central point is added among queue Q, and by the central point labeled as
Processing；
Step 4-4-4：Judge whether queue Q is empty, and step 4-4-2 is performed if not for sky；Otherwise, step 4- is performed
4-5；
Step 4-4-5：Cresult is the level set of final cluster result, the i.e. both arms of 3D actor models；
Step 4-5：All central points are ranked up by y direction is ascending, find the endpoint value of both legs, are carried out
The cluster identical with step 4-4, obtains the level set of the both legs of 3D actor models；
Step 4-6：Central point is concentrated into the central point not being clustered as in the level set of the trunk of 3D actor models
Heart point；
Step 4-7：Calculate the fitting a straight line of central point of the level set of the both arms of 3D actor models, the level set of both legs
The fitting a straight line of central point, the level set of trunk central point fitting a straight line respectively as the skeleton line of both arms, the bone of both legs
The skeleton line of stringing, trunk.
Step 5：Obtained skeleton line is embedded into 3D actor models.
As shown in Figure 10, comprise the following steps that：
Step 5-1：Calculate the position of crucial artis；
Crucial artis includes：Head, neck, shoulder joint, elbow joint, wrist joint, hand, hip joint, big leg joint, ankle
Joint and foot；
Step 5-1-1：Head, hand, hip are tried to achieve by the end of the skeleton line of the 3D actor model each section after cutting
The position in joint, big leg joint and foot's artis：Two ends of leg skeleton line are big leg joint and foot；Arm skeleton
Line lower end is hand；Two ends of trunk skeleton line are head and hip joint；
Step 5-1-2：The level set obtained using the triangular topological relations of the both arms part of horizontal plane cutting 3D actor models
Girth, determine carpal position；
As shown in figure 11, comprise the following steps that：
Step 5-1-2-1：Go out a spin moment using the direction of skeleton line and the angle calcu-lation of longitudinal axis positive direction of both arms
Battle array；
Spin matrix represents as follows：
Wherein, θ is the angle of direction and the longitudinal axis positive direction of the skeleton line of both arms, (nx,ny,nz) it is rotation axis, using double
The skeleton line of arm is tried to achieve with longitudinal axis positive direction multiplication cross；
Step 5-1-2-2：Both arms grid model is multiplied by the spin matrix, both arms grid model is upright；
Step 5-1-2-3：Cross-cutting operations are carried out to both arms grid model using horizontal plane, are obtained using the method for step 2
A series of level sets；
Step 5-1-2-4：If there are multiple level sets under some elevation plane, the vertex of all level sets is considered as
One point set, the convex closure of the point set is calculated using Graham algorithms, by the use of the convex closure as the elevation plane under unique level set；
Step 5-1-2-5：The girth of each level set is calculated, is closed the midpoint of the level set of girth minimum as wrist
The position of node.
Step 5-1-3：Calculate the position of P point：Utilize trunk more than the method processing arm of step 5-1-2
The level set of the grid model of position, the position using the midpoint of the level set of girth minimum as P point；
Step 5-1-4：Calculate the position of shoulder joint：Intersect with plane where shoulder joint in direction using the skeleton line of both arms
An initial shoulder joint position is obtained to be modified again；
Step 5-1-4-1：The plane and P point determined using the point of the skeleton line longitudinal axis value maximum of both arms is indulged
The mid-plane for the plane that axis value determines, obtains plane residing for shoulder joint；
Step 5-1-4-2：Intersect with the plane where shoulder joint to obtain an intersection point using the direction of the skeleton line of both arms,
The intersection point is the position of shoulder joint；
Step 5-1-4-3：Using the position of shoulder joint and the position of wrist joint point, pass through elbow in standard skeleton structure
Joint obtains the initial position of elbow joint with shoulder joint, the position proportional of wrist joint point；
Step 5-1-4-4：Using the point more than the plane that elbow joint longitudinal axis value determines as upper arm part, by upper arm parts
The central point divided re-starts fitting a straight line, makes the straight line intersect to obtain modified shoulder joint position with plane where shoulder joint,
5-1-4-3 is repeated until the position of shoulder joint no longer changes, the shoulder joint node obtained at this time is final shoulder joint
Position.
Step 5-1-5：Calculate the position of ankle-joint：The corresponding water in leg in the level set that selecting step 3 filters out first
Flat collection, then calculates angle between the midpoint connecting line of selected each adjacent level collection, angle of 70 degree not less than given threshold
Position, that is, ankle-joint location.
Step 5-2：Matched using between the position of crucial artis and standard skeleton structure, by between artis
Position proportional obtain the position of remaining all artis；
Step 5-3：Connect all semantically related artis and obtain final skeleton, be embedded into 3D actor models
In.
Contrasted using the distance transformation method of this method and the prior art, four be directed in Fig. 4 (a)~(d)
Model is compared, and the skeleton that two methods obtain is as shown in Figure 5.It is distance transformation method in wherein Fig. 5 (a) (c) (e) (g)
The skeleton of acquisition, the skeleton obtained for this method in Fig. 5 (b) (d) (f) (h).It is respectively compared the time efficiency of meter both approaches
And two kinds of parameters of accuracy, the time that two methods are illustrated in table 1 compare, illustrating two methods in table 2 is directed to arm
The comparison of long error, the application condition that two methods are directed to leg length is illustrated in table 3.
The time of 1 two methods of table compares
2 two methods of table are directed to the comparison of the error of brachium
3 two methods of table are directed to the application condition of leg length
By contrast as can be seen that the method for the present invention is directed to the method for range conversion in time efficiency and accuracy
There is certain advantage.
Claims (9)
1. a kind of 3D actor model framework extraction methods based on level set central cluster, it is characterised in that comprise the following steps：
Step 1：Using height function cutting process, the intersection point after being cut are carried out as Morse function pair 3D actor models
Set；
Step 2：Intersection point set after cutting is clustered according to the connectedness between intersection point, it is horizontal to be clustered into some sons
Collection：If certain two intersection point is crossed to form for elevation plane with a triangular topological relations, the two intersection points are preserved into one
A line segment, is clustered using the connectivity between line segment, obtains some sub- level sets；
Step 3：Screening conditions are used as using the area of level set and like circle property, area is more than level set area threshold and like round property
Screened more than the level set like circle property threshold value；
Step 4：Level set after screening is clustered using the center of level set, obtains the water of the both arms of 3D actor models
Flat collection, the level set of both legs, trunk level set central point, the central point of the level set of the both arms of calculating 3D actor models
Fitting a straight line, the fitting a straight line of central point of level set of both legs, the fitting a straight line of central point of level set of trunk are made respectively
Skeleton line, the skeleton line of trunk of skeleton line, both legs for both arms；
Step 5：Obtained skeleton line is embedded into 3D actor models.
2. the 3D actor model framework extraction methods according to claim 1 based on level set central cluster, its feature exist
In the step 1 comprises the following steps that：
Step 1-1：Height function is chosen 3D actor models are handled to obtain 3D actor model height hmodel；
Step 1-2：3D actor model cutting times n are set, cutting is proceeded by from the top of 3D actor models, every
hmodel/ n distances take a plane i.e. elevation plane；
Step 1-3：The triangular topological relations in FBX files are read, by maximum y of each triangular topological relations in y-axis directionmax
With minimum value yminIt is saved in array, to maximum y of each triangular topological relations in y-axis directionmaxIt is ranked up；
Step 1-4：Each elevation plane of 3D actor models and the intersection point of triangular topological relations are calculated, obtains each elevation plane energy
Enough intersecting triangular topological relations lists；
Step 1-5：The triangular topological relations list that can be intersected according to each elevation plane, determines that each elevation plane can intersect with it
Triangular topological relations between intersection point, that is, cut after intersection point set.
3. the 3D actor model framework extraction methods according to claim 2 based on level set central cluster, its feature exist
In the step 1-4's comprises the following steps that：
Step 1-4-1：Travel through each triangular topological relations in array, if current triangular topological relations in y-axis direction most
Big value ymaxMore than present level plane Pcurrent, then step 1-4-2 is performed, otherwise from all after current triangular topological relations
Triangular topological relations all no longer with present level plane PcurrentIt is intersecting, stop traversal, obtained triangular topological relations list is can
With present level plane PcurrentIntersecting triangular topological relations list；
Step 1-4-2：Judge the minimum value y in y-axis direction of current triangular topological relationsminWhether present level plane is more than
Pcurrent：It is that then current triangular topological relations are in present level plane PcurrentOn, next triangular topological relations are traveled through, return to step
Rapid 1-4-1；Otherwise current triangular topological relations and present level plane PcurrentIt can intersect, preserve current triangular topological relations, time
Go through next triangular topological relations, return to step 1-4-1.
4. the 3D actor model framework extraction methods according to claim 1 based on level set central cluster, its feature exist
In the step 3 comprises the following steps that：
Step 3-1：Set screening conditions, including level set area threshold and like circle property threshold value；
Step 3-2：Calculate the area of each level set and like circle property；
Step 3-3：The level set that area is more than level set area threshold and is more than seemingly circle property threshold value like circle property screens.
5. the 3D actor model framework extraction methods according to claim 1 based on level set central cluster, its feature exist
In the step 4 comprises the following steps that：
Step 4-1：The central point of each level set is calculated, forms center point set；
Step 4-2：All central points are arranged according to the axial length value in the another two direction in three dimensions apart from height
Sequence, the both arms direction of extension of direction, that is, 3D actor models of axial length value maximum；
Step 4-3：Calculate the average value, 5 latter of distance between the adjacent center point of preceding 5 central points of the center point set after sequence
The average value of distance between the adjacent center point of central point, respectively using 2 times of the two average values as two cluster radius；
Step 4-4：First central point of the center point set after sequence is chosen respectively, last central point is clicked through as starting
Row cluster, obtains the level set of the both arms of 3D actor models；
Step 4-5：All central points are ranked up by y direction is ascending, find the endpoint value of both legs, carries out and walks
Rapid cluster identical 4-4, obtains the level set of the both legs of 3D actor models；
Step 4-6：Central point using the central point that central point concentration is not clustered as the level set of the trunk of 3D actor models；
Step 4-7：The center of the fitting a straight line of the central point of the level set of the both arms of calculating 3D actor models, the level set of both legs
The fitting a straight line of point, the level set of trunk central point fitting a straight line respectively as the skeleton lines of both arms, both legs skeleton line,
The skeleton line of trunk.
6. the 3D actor model framework extraction methods according to claim 1 based on level set central cluster, its feature exist
In the step 5 comprises the following steps that：
Step 5-1：Calculate the position of crucial artis；
Crucial artis includes：Head, neck, shoulder joint, elbow joint, wrist joint, hand, hip joint, big leg joint, ankle-joint
And foot；
Step 5-2：Matched using between the position of crucial artis and standard skeleton structure, pass through the position between artis
The ratio of putting obtains the position of remaining all artis；
Step 5-3：Connect all semantically related artis and obtain final skeleton, be embedded into 3D actor models.
7. the 3D actor model framework extraction methods according to claim 6 based on level set central cluster, its feature exist
In the step 5-1's comprises the following steps that：
Step 5-1-1：By the end of the skeleton line of the 3D actor model each section after cutting try to achieve head, hand, hip joint,
The position of big leg joint and foot's artis：Two ends of leg skeleton line are big leg joint and foot；Under arm skeleton line
Portion end is hand；Two ends of trunk skeleton line are head and hip joint；
Step 5-1-2：The week of the level set obtained using the triangular topological relations of the both arms part of horizontal plane cutting 3D actor models
It is long, determine carpal position；
Step 5-1-3：Calculate the position of P point：Utilize trunk position more than the method processing arm of step 5-1-2
Grid model level set, the position using the midpoint of the level set of girth minimum as P point；
Step 5-1-4：Calculate the position of shoulder joint：Intersect to obtain with plane where shoulder joint in direction using the skeleton line of both arms
One initial shoulder joint position is modified again；
Step 5-1-5：Calculate the position of ankle-joint：The corresponding level set in leg in the level set that selecting step 3 filters out first,
Then angle between the midpoint connecting line of each adjacent level collection selected by calculating, angular position, that is, ankle not less than given threshold close
Save position.
8. the 3D actor model framework extraction methods according to claim 7 based on level set central cluster, its feature exist
In the step 5-1-2's comprises the following steps that：
Step 5-1-2-1：Go out a spin matrix using the direction of skeleton line and the angle calcu-lation of longitudinal axis positive direction of both arms；
Spin matrix represents as follows：
<mrow>
<mi>M</mi>
<mo>=</mo>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mrow>
<msubsup>
<mi>n</mi>
<mi>x</mi>
<mn>2</mn>
</msubsup>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mi>cos</mi>
<mi>&theta;</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mi>cos</mi>
<mi>&theta;</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<msub>
<mi>n</mi>
<mi>x</mi>
</msub>
<msub>
<mi>n</mi>
<mi>y</mi>
</msub>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mi>cos</mi>
<mi>&theta;</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msub>
<mi>n</mi>
<mi>z</mi>
</msub>
<mi>sin</mi>
<mi>&theta;</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<msub>
<mi>n</mi>
<mi>x</mi>
</msub>
<msub>
<mi>n</mi>
<mi>z</mi>
</msub>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mi>cos</mi>
<mi>&theta;</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<msub>
<mi>n</mi>
<mi>y</mi>
</msub>
<mi>sin</mi>
<mi>&theta;</mi>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>n</mi>
<mi>x</mi>
</msub>
<msub>
<mi>n</mi>
<mi>y</mi>
</msub>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mi>cos</mi>
<mi>&theta;</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<msub>
<mi>n</mi>
<mi>z</mi>
</msub>
<mi>sin</mi>
<mi>&theta;</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<msubsup>
<mi>n</mi>
<mi>y</mi>
<mn>2</mn>
</msubsup>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mi>cos</mi>
<mi>&theta;</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mi>cos</mi>
<mi>&theta;</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<msub>
<mi>n</mi>
<mi>y</mi>
</msub>
<msub>
<mi>n</mi>
<mi>z</mi>
</msub>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mi>cos</mi>
<mi>&theta;</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msub>
<mi>n</mi>
<mi>x</mi>
</msub>
<mi>sin</mi>
<mi>&theta;</mi>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>n</mi>
<mi>x</mi>
</msub>
<msub>
<mi>n</mi>
<mi>z</mi>
</msub>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mi>cos</mi>
<mi>&theta;</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msub>
<mi>n</mi>
<mi>y</mi>
</msub>
<mi>sin</mi>
<mi>&theta;</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<msub>
<mi>n</mi>
<mi>y</mi>
</msub>
<msub>
<mi>n</mi>
<mi>z</mi>
</msub>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mi>cos</mi>
<mi>&theta;</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<msub>
<mi>n</mi>
<mi>x</mi>
</msub>
<mi>sin</mi>
<mi>&theta;</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<msubsup>
<mi>n</mi>
<mi>z</mi>
<mn>2</mn>
</msubsup>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mi>cos</mi>
<mi>&theta;</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mi>cos</mi>
<mi>&theta;</mi>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
Wherein, θ is the angle of direction and the longitudinal axis positive direction of the skeleton line of both arms, (nx, ny, nz) it is rotation axis, utilize both arms
Skeleton line is tried to achieve with longitudinal axis positive direction multiplication cross；
Step 5-1-2-2：Both arms grid model is multiplied by the spin matrix, both arms grid model is upright；
Step 5-1-2-3：Cross-cutting operations are carried out to both arms grid model using horizontal plane, a system is obtained using the method for step 2
Row level set；
Step 5-1-2-4：If there are multiple level sets under some elevation plane, the vertex of all level sets is considered as one
Point set, the convex closure of the point set is calculated using Graham algorithms, by the use of the convex closure as the elevation plane under unique level set；
Step 5-1-2-5：The girth of each level set is calculated, using the midpoint of the level set of girth minimum as wrist joint point
Position.
9. the 3D actor model framework extraction methods according to claim 7 based on level set central cluster, its feature exist
In the step 5-1-4's comprises the following steps that：
Step 5-1-4-1：The plane and P point longitudinal axis value determined using the point of the skeleton line longitudinal axis value maximum of both arms
The mid-plane of definite plane, obtains plane residing for shoulder joint；
Step 5-1-4-2：Intersect with the plane where shoulder joint to obtain an intersection point, the friendship using the direction of the skeleton line of both arms
Point is the position of shoulder joint；
Step 5-1-4-3：Using the position of shoulder joint and the position of wrist joint point, pass through elbow joint in standard skeleton structure
The initial position of elbow joint is obtained with the position proportional of shoulder joint, wrist joint point；
Step 5-1-4-4：Using the point more than the plane that elbow joint longitudinal axis value determines as upper arm part, by upper arm part
Central point re-starts fitting a straight line, makes the straight line intersect to obtain modified shoulder joint position with plane where shoulder joint, repeats
5-1-4-3 is performed until the position of shoulder joint no longer changes, the shoulder joint node obtained at this time is final shoulder joint position.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510868466.8A CN105513136B (en) | 2015-11-30 | 2015-11-30 | A kind of 3D actor model framework extraction methods based on level set central cluster |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510868466.8A CN105513136B (en) | 2015-11-30 | 2015-11-30 | A kind of 3D actor model framework extraction methods based on level set central cluster |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105513136A CN105513136A (en) | 2016-04-20 |
| CN105513136B true CN105513136B (en) | 2018-05-04 |
Family
ID=55721088
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510868466.8A Active CN105513136B (en) | 2015-11-30 | 2015-11-30 | A kind of 3D actor model framework extraction methods based on level set central cluster |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105513136B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107219925B (en) * | 2017-05-27 | 2021-02-26 | 成都通甲优博科技有限责任公司 | Posture detection method and device and server |
| CN110135397B (en) * | 2019-05-27 | 2022-07-08 | 上海埃蒙迪材料科技股份有限公司 | Method for identifying dental crown part in three-dimensional tooth model |
| CN111583232B (en) * | 2020-05-09 | 2024-04-26 | 北京天智航医疗科技股份有限公司 | Femoral head center determining method, device, computer equipment and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101271589A (en) * | 2007-03-22 | 2008-09-24 | 中国科学院计算技术研究所 | Three-dimensional mannequin joint center extraction method |
| CN102254343A (en) * | 2011-07-01 | 2011-11-23 | 浙江理工大学 | Convex hull and OBB (Oriented Bounding Box)-based three-dimensional grid model framework extracting method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101815975B1 (en) * | 2011-07-27 | 2018-01-09 | 삼성전자주식회사 | Apparatus and Method for Detecting Object Pose |
-
2015
- 2015-11-30 CN CN201510868466.8A patent/CN105513136B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101271589A (en) * | 2007-03-22 | 2008-09-24 | 中国科学院计算技术研究所 | Three-dimensional mannequin joint center extraction method |
| CN102254343A (en) * | 2011-07-01 | 2011-11-23 | 浙江理工大学 | Convex hull and OBB (Oriented Bounding Box)-based three-dimensional grid model framework extracting method |
Non-Patent Citations (2)
| Title |
|---|
| Automatic Joints Extraction of Scanned Human Body;Yong Yu等;《International Conference on Digital Human Modeling》;20071231;第4561卷;第286-293页 * |
| 一种姿态无关的人体模型骨骼提取方法;于勇 等;《计算机研究与发展》;20080731;第45卷(第7期);第1249-1258页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105513136A (en) | 2016-04-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110287904A (en) | A kind of lane line extracting method, device and storage medium based on crowdsourcing data | |
| CN105513136B (en) | A kind of 3D actor model framework extraction methods based on level set central cluster | |
| CN102508973B (en) | Rapid intersection method for STL (stereo lithography) models of products | |
| CN106446910B (en) | Complex geological curved surface feature extraction and reconstruction method | |
| CN103699654B (en) | A kind of across engineer's scale map vector network of rivers data target matching method of the same name | |
| CN105354881B (en) | Distortion of the mesh optimized algorithm based on Category Attributes data | |
| CN104268138A (en) | Method for capturing human motion by aid of fused depth images and three-dimensional models | |
| CN104794221B (en) | A kind of multi-Dimensional Data Analytical System based on business object | |
| CN103559705A (en) | Computer method for comparing similarity of different plant forms | |
| CN105354335B (en) | For the time-varying tree graph layout method of time-varying hierarchical data and application | |
| CN105069239A (en) | Individualized clothing template implementation method and system | |
| CN108765180A (en) | The overlapping community discovery method extended with seed based on influence power | |
| CN101783016A (en) | Crown appearance extract method based on shape analysis | |
| CN106418850A (en) | Method for manufacturing two-piece sleeve basic pattern of women's suit | |
| CN104504731A (en) | Human motion synthesis method based on motion diagram | |
| CN102013114A (en) | Microstation v8i-based city rapid-modeling method | |
| CN107908899A (en) | A kind of wire selection system and selection method for Construction of Wind Power construction road | |
| CN108595631A (en) | Three-dimensional CAD model bilayer search method based on Graph Spectral Theory | |
| CN102393825A (en) | Non-coplanar elimination based detecting method for continuous collision of flexible scene | |
| CN112116709B (en) | Terrain characteristic line processing method for improving terrain expression precision | |
| CN109033746A (en) | A kind of protein complex recognizing method based on knot vector | |
| Pantuwong et al. | Skeleton-growing: a vector-field-based 3D curve-skeleton extraction algorithm | |
| CN104318035A (en) | General-to-part type triangular network multi-threading parallel generation method for massive terrain big data cloud | |
| Xiong et al. | Automated structured all-quadrilateral and hexahedral meshing of tubular surfaces | |
| CN114463494A (en) | Automatic topographic feature line extracting algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |