CN104103035A - Three-dimensional model scaling method - Google Patents
Three-dimensional model scaling method Download PDFInfo
- Publication number
- CN104103035A CN104103035A CN201310130687.6A CN201310130687A CN104103035A CN 104103035 A CN104103035 A CN 104103035A CN 201310130687 A CN201310130687 A CN 201310130687A CN 104103035 A CN104103035 A CN 104103035A
- Authority
- CN
- China
- Prior art keywords
- voxel
- seam
- energy value
- optimum
- initial
- 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
Landscapes
- Image Generation (AREA)
Abstract
The invention provides a three-dimensional model scaling method. The method comprises: an initiation voxel and a termination voxel in a point cloud model are defined, wherein the initiation voxel and the termination voxel are two which are in a tight bounding box of the point cloud model and parallel with the scaling direction and whose opposite angles are parallel; and the following substeps are executed on each scaling direction: a slit set starting from each initiation voxel and arriving at each termination voxel is acquired and the optimal slit set is acquired from the slit sets, and according to the energy value of the optimal slit in the optimal slit set, a deleting and patching method of the optimal slit is adopted for scaling the point cloud model. By adopting the method, the three-dimensional point cloud model can be scaled, and the application range is wider and more universal.
Description
Technical field
The present invention relates to 3-D view treatment technology, particularly relate to a kind of Zoom method of three-dimensional model.
Background technology
The zoom technology of three-dimensional model is in the situation that the scene of establishment new model and adaptation different size is very useful.But, three-dimensional model is unified to convergent-divergent everywhere and in application, can be very restricted, because directly model is unified to some feature of convergent-divergent meeting damage model, thereby cause some visual distortions.Therefore, how can when convergent-divergent, keep as much as possible original structure and feature to be just called the large study hotspot of one in scaling of model technology.
Traditional scaling of model method comprises seam and deletes reinforcing method and nonuniformity zoom technology.Wherein, seam is deleted reinforcing method specifically: use a kind of energy function (such as gradient function, entropy function, vision highlight function etc.) to define the importance of pixel, and definition " seam " is that one group of top by image is to bottommost, or the 8 pixel paths, field by the high order end of image to low order end, make this paths meet a kind of optimization based on image energy, then towards a direction, reject or increase " seam " iteratively, can under the prerequisite that keeps image subject content, change the size of image.For image, shrink, in order to retain the body matter on image, seam is deleted reinforcing method and is rejected as much as possible some low-energy pixels and keep high-octane pixel.For image stretch, according to the ascending order of energy, insert one by one " seam ".
Nonuniformity zoom technology is to propose for complex model.It is inadequate that complex model is merely used to an overall zoom factor, because complex model usually can comprise some, through convergent-divergent, holds yielding parts or feature.Make discovery from observation, if complex model is used to unified overall zoom factor, the deformation extent between different parts is different, but this different deformation extent be not uniformity be distributed in the surface of model, some are significantly out of shape in having the surface of some special characteristic, the impact that the position of some feature is not out of shape substantially.Therefore, for complex model, convergent-divergent in various degree should be carried out in the position of different characteristic, keeps those to be subject to original feature that convergent-divergent brings the position of variable effect as far as possible, and to not being subject to convergent-divergent, brings the position of deformation effect to carry out the convergent-divergent of relative large scale.By so-called slippage analysis and normal curvature analysis, some position that can measure complex object is subject to the degree size of the deformation effect that convergent-divergent brings.
Yet existing seam is deleted reinforcing method can only be applied to two dimensional image, also seam is not deleted to reinforcing method and extended to the effective ways in field of three dimension.Although and the nonuniformity zoom technology of complex model is the zoom technology in three-dimensional picture field, it has the limitation of 2: one, the method can only operate for grid model; Its two, to the assay method of the yielding degree of model, be effective for the geometry of some systematicness substantially, that is to say that the complex model that the method is just comprised of some basic bodies for those is effective.
Summary of the invention
Based on this, be necessary for the circumscribed problem in conventional art, providing a kind of can carry out the Zoom method that convergent-divergent, range of application more extensively have more the three-dimensional model of universality to three-dimensional point cloud model.
A Zoom method for three-dimensional model, described method comprises:
Initial voxel in defining point cloud model and stop voxel, in the tight bounding box that described initial voxel and termination voxel are point cloud model parallel with zoom direction and diagonal angle is parallel two;
Direction to each convergent-divergent, carry out:
Obtain from each initial voxel and arrive the seam set that stops voxel, from described seam set, obtain optimum seam set;
Energy value according to the optimum seam in described optimum seam set, adopts the reinforcing method of deleting of optimum seam to carry out convergent-divergent to point cloud model.
Therein in an embodiment, described in obtain from each initial voxel and arrive the seam set that stops voxel, from seam set, obtain optimum seam set, comprising:
To each initial voxel, obtain the institute that stops voxel from initial voxel arrival seamed, form seam and gather, the summation that the energy value stitching in described seam set is the energy value of its voxel comprising, the seam that obtains energy value minimum in described seam set is optimum seam, forms optimum seam set.
Therein in an embodiment, described in obtain from each initial voxel and arrive the seam set that stops voxel, from seam set, obtain optimum seam set, comprising:
To each initial voxel, carry out:
The voxel that finds all and initial voxel to adjoin, obtain the voxel that adjoins accumulative total minimum energy value minimum in voxel, recursion is to stopping voxel so layer by layer, by each, stopping voxel dates back to forward initial voxel and records path, this path is an optimum seam with this termination voxel ending, and the energy value of this optimum seam stops the accumulative total minimum energy value of voxel for it;
The optimum seam that each initial voxel is obtained forms optimum seam set.
Therein in an embodiment, described in find all and initial voxel to adjoin voxel, obtain the voxel that adjoins accumulative total minimum energy value minimum in voxel, recursion, to stopping voxel, comprising so layer by layer:
From initial voxel, each voxel circulation is carried out to following process, until arrive, stop voxel:
Find 26 of voxel to adjoin voxel, calculate the accumulative total minimum energy value that each adjoins voxel, obtain the voxel that adjoins that wherein adds up minimum energy value minimum.
Therein in an embodiment, each that adopts that following formula calculates voxel adjoined the accumulative total minimum energy value of voxel:
M(i,j,k)=e(i,j,k)+min{M(i-1,j-1,k+1),M(i,j-1,k+1),M(i+1,j-1,k+1),
M(i-1,j-1,k),M(i,j-1,k),M(i+1,j-1,k),M(i-1,j-1,k-1),M(i,j-1,k-1),M(i+1,j-1,k-1),
M(i-1,j,k+1),M(i,j,k+1),M(i+1,j,k+1),M(i-1,j,k),M(i+1,j,k),M(i-1,j,k-1),
M(i,j,k-1),M(i+1,j,k-1),M(i-1,j+1,k+1),M(i,j+1,k+1),M(i+1,j+1,k+1),
M(i-1,j+1,k),M(i,j+1,k),M(i+1,j+1,k),M(i-1,j+1,k-1),M(i,j+1,k-1),M(i+1,j+1,k-1)}
Wherein, M (i, j, k) is the accumulative total minimum energy value of voxel v (i, j, k), the energy value that e (i, j, k) is voxel, and i, j, k are respectively the call number of voxel in X, Y, Z axis.
In an embodiment, described according to the energy value of the optimum seam in described optimum seam set, the reinforcing method of deleting of the optimum seam of employing is carried out convergent-divergent to point cloud model, comprising therein:
Described optimum seam set is sorted according to energy value is ascending;
If shrink in the direction of described convergent-divergent, from the set of optimum seam, select successively the optimum seam that energy value is low to delete, and optimum seam of every deletion, the voxel reach below of the optimum seam of described deletion is supplemented to the optimum seam of deleting;
If uphold in the direction of described convergent-divergent, from optimum seam set, select successively the optimum seam that energy value is low, after optimum seam that will described selection voxel below, move, the optimum that then copies described selection is sewed and mend and is filled the position of vacating.
In an embodiment, described method also comprises therein:
Adopt three directions of the tight bounding box of principle component analysis calculation level cloud model;
Take and preset unit length as standard is divided into several mikey lattices along described three directions to point cloud model, be voxel.
In an embodiment, described method also comprises therein:
In acquisition point cloud model normal information a little;
To each voxel, the energy value that obtains voxel in this voxel normal direction a little depart from the degree of average normal direction.
Therein in an embodiment, to { the p a little of the institute in each voxel
0, p
1..., p
s, these put corresponding normal information is { n
0, n
1..., n
s, the variance of normal direction is
the energy value that calculates voxel is
In an embodiment, described initial voxel and termination voxel are all effective voxel therein, in described initial voxel and termination voxel, all comprise a little.
The Zoom method of above-mentioned three-dimensional model, by the initial voxel in defining point cloud model and termination voxel, and stitch and delete reinforcing method with traditional seam and carried out combination by optimum, make seam delete reinforcing method and can expand to field of three dimension, thereby realized the convergent-divergent to three-dimensional point cloud model, range of application is wider and have more universality.
Accompanying drawing explanation
Fig. 1 is the schematic flow sheet of the Zoom method of three-dimensional model in an embodiment;
Fig. 2 is the schematic diagram of original point cloud model;
Fig. 3 is by the schematic diagram of point cloud model voxelization;
Fig. 4 is the schematic diagram of energy distribution in point cloud model;
Fig. 5 is for to be contracted to the schematic diagram of 0.7 times by point cloud model;
Fig. 6 is for to extend to the schematic diagram of 1.6 times by point cloud model;
Fig. 7 is the schematic diagram that the voxel in X-direction is arranged.
Embodiment
As shown in Figure 1, in one embodiment, a kind of Zoom method of three-dimensional model, comprising:
Step 102, the initial voxel in defining point cloud model and stop voxel, wherein, in the tight bounding box that initial voxel and termination voxel are point cloud model parallel with zoom direction and diagonal angle is parallel two.
Before step 102, can first to voxel, define.Concrete, can adopt three directions of the tight bounding box of traditional principle component analysis calculation level cloud model, and take and preset unit length as standard is divided into several mikey lattices along these three directions to point cloud model, be voxel.
So-called point cloud model, is the magnanimity point set of expressing object space distribution and target surface feature under the same space reference frame, wherein comprises three-dimensional coordinate and normal information.As shown in Figure 2, be original point cloud model, being divided into voxel need to have three local reference direction (σ
1, σ
2, σ
3), three directions of the tight bounding box of employing principle component analysis (PCA) calculation level cloud model, take unit length as standard is along (σ
1, σ
2, σ
3) this point cloud model is divided into several mikey lattices, be called " voxel " (as shown in Figure 3).In voxel, may comprise some points, also may not comprise a little, the voxel not comprising a little thinks that its energy is zero, is invalid voxel, otherwise the voxel that has comprised point is effective voxel.
Further, also need to obtain the energy value of each voxel.Concrete, in acquisition point cloud model normal information a little; To each voxel, the energy value of voxel in this voxel normal direction a little depart from the degree of average normal direction.The energy characterization of voxel voxel have in much degree can representative point cloud model feature, due to what process, be point cloud model, that therefore can utilize only has locus and two information of normal direction.Can imagine, if one section of some cloud is very smooth, this section of some cloud do not comprise the characteristic information in too much master pattern so, because smooth some cloud can be filled up to generate by interpolation completely; If the out-of-flatness of one section of some cloud, illustrates that the characteristic information that this section of some cloud comprise is very abundant so.Therefore putting the whether smooth of cloud, whether to have characteristic with it be closely-related.Therefore, the normal information of utilisation point calculates the planarization of a cloud, namely the energy of voxel.As shown in Figure 4, can utilize normal distributing homogeneity to calculate the energy of voxel, in figure, the energy of different masses is different.
Further, in one embodiment, to { the p a little of institute in each voxel
0, p
1..., p
s, these put corresponding normal information is { n
0, n
1..., n
s, the variance of normal direction is
the energy value that calculates voxel is
represented that its all normal directions depart from the degree of average normal direction.
No matter be two dimensional model or three-dimensional model, seam all needs a starting point and ending point, for three-dimensional point cloud model, initial voxel and termination voxel are parallel with zoom direction in the tight bounding box of point cloud model and diagonal angle is parallel two.For example, if in X-direction convergent-divergent, initial voxel with stop voxel should be parallel with X-direction, and parallel a plurality of voxels in diagonal angle.Preferably, initial voxel and termination voxel are all effective voxel.That is to say, it is effective voxel and two conditions of edge voxel that initial voxel and termination voxel all should meet.In another embodiment, also can, by manually specifying initial voxel and stopping voxel, make initial voxel and stop voxel to meet above two conditions.
Step 104, the direction to each convergent-divergent, carries out: obtain from each initial voxel and arrive the seam set that stops voxel, obtain optimum seam set from seam set.
So-called seam, refers to from initial voxel and arrives the path that stops voxel, because initial voxel comprises a plurality ofly, from each initial voxel, can find a lot of paths to arrive and stop voxels, so each initial voxel corresponding a seam set.From each seam set, get optimum seam, that is to say, for each initial voxel, can find a paths, meet the optimization on energy, the optimized path that these initial voxels arrival stop voxel forms optimum seam set.
In one embodiment, in step 104, to each initial voxel, obtain the institute that stops voxel from initial voxel arrival seamed, form seam set, the summation that the energy value stitching in described seam set is the energy value of its voxel comprising, the seam that obtains energy value minimum in described seam set is optimum seam, forms optimum seam set.
Concrete, the set of establishing initial voxel is S={v
s1, v
s2..., v
sm, the set that stops voxel is T={v
t1, v
t2..., v
tn.For convenience of description, by three direction (σ of the tight bounding box of point cloud model
1, σ
2, σ
3) snap in X, Y, Z axis.Because X, Y, tri-axles of Z are rotation symmetries, therefore can only consider the convergent-divergent of one of them direction, to two other direction, can adopt identical principle.
Suppose only to consider the convergent-divergent in X-direction, can adopt following formula definition seam:
And v (i (0), j (0), k (0)) ∈ S, v (i (d), j (d), k (d)) ∈ T
Wherein, t represents that this variable is 1 when initial voxel from initial voxel to the step number variable that stops voxel, stopping voxel, is d;
represent to make all should meet following condition for any t; I (t), j (t), k (t) are illustrated in the call number the X, Y, Z axis of t the voxel of this seam from initial voxel.So-called call number, refer to: three directions of the tight bounding box of point cloud model have been snapped in X, Y, Z axis, and closely a summit of bounding box has necessarily snapped to the initial point of coordinate system, from this former lighting, toward X-direction, be followed successively by each voxel and be numbered the index X-axis so, owing to being three-dimensional, also number consecutively index on Y, Z axis, so each voxel is just determined by three call numbers are unique.
If v for voxel (i, j, k) expression, its energy is e (i, j, k), and the energy definition of seam is as follows:
Further, from each initial voxel v
sp(1≤p≤m) sets out, and can find some to be sewn to and to reach certain and stop voxel v
tq(1≤q≤n), this termination voxel v of serving as reasons
tqa seam S set of ending
tq, need to stitch S set at this
tqin search out one seam s
*, make E (s
*)=min{E (s
i), s
i∈ S
tq, this seam is optimum seam.
Concrete, in step 104, to each initial voxel, carry out: the voxel that finds all and initial voxel to adjoin, obtain the voxel that adjoins accumulative total minimum energy value minimum in voxel, recursion, to stopping voxel, stops voxel by each and dates back to forward initial voxel and record path so layer by layer, this path is an optimum seam with this termination voxel ending, and the energy value of this optimum seam stops the accumulative total minimum energy value of voxel for it; The optimum seam that each initial voxel is obtained forms optimum seam set.
Due to for each voxel, it has 26 voxels that adjoin around, from it, arrive termination voxel and have 26 possible entrances, to each voxel, need to select its voxel that adjoins accumulative total minimum energy value minimum in voxel as entrance, recursion, to stopping voxel, forms a paths so layer by layer, is optimum seam.
Therefore, further, from initial voxel, each voxel circulation is carried out to following process, until arrive, stop voxel: find 26 of voxel to adjoin voxel, calculate the accumulative total minimum energy value that each adjoins voxel, obtain the voxel that adjoins that wherein adds up minimum energy value minimum.
Each that in one embodiment, can adopt that following formula calculates voxel adjoined the accumulative total minimum energy value of voxel:
M(i,j,k)=e(i,j,k)+min{M(i-1,j-1,k+1),M(i,j-1,k+1),M(i+1,j-1,k+1),
M(i-1,j-1,k),M(i,j-1,k),M(i+1,j-1,k),M(i-1,j-1,k-1),M(i,j-1,k-1),M(i+1,j-1,k-1),
M(i-1,j,k+1),M(i,j,k+1),M(i+1,j,k+1),M(i-1,j,k),M(i+1,j,k),M(i-1,j,k-1),
M(i,j,k-1),M(i+1,j,k-1),M(i-1,j+1,k+1),M(i,j+1,k+1),M(i+1,j+1,k+1),
M(i-1,j+1,k),M(i,j+1,k),M(i+1,j+1,k),M(i-1,j+1,k-1),M(i,j+1,k-1),M(i+1,j+1,k-1)}
Wherein, M (i, j, k) is the accumulative total minimum energy value of voxel v (i, j, k), the energy value that e (i, j, k) is voxel, and i, j, k are respectively the call number of voxel in X, Y, Z axis.
For each initial voxel, can find an optimized path to arrive and stop voxel, be i.e. optimum seam.。Optimum seam is to stop voxel v
tqending, stops voxel v
tqaccumulative total minimum energy value (M value) be the energy value of this optimum seam.
Step 106, the direction to each convergent-divergent, continues to carry out: the energy value according to the optimum seam in the set of optimum seam, adopts the reinforcing method of deleting of optimum seam to carry out convergent-divergent to point cloud model.
Concrete, in step S106, can first optimum seam be gathered and be sorted according to energy value is ascending.For example, for optimum seam S set
tq, after sorting according to energy value is ascending, obtain the optimum seam S set that one group of energy increases progressively successively
op={ s
1, s
2..., s
r.Further, if shrink in the direction of described convergent-divergent, from optimum seam S set
opin select successively the low optimum of energy value seam to delete, and optimum seam of every deletion, supplements by the optimum seam of deleting voxel reach below the optimum seam of deleting.As shown in Figure 5, when point cloud model is contracted to after 0.7 times, can substantially retain the feature of master mould.So-called voxel below, refers in the direction of convergent-divergent, and the voxel in the optimum of deletion seam is toward the voxel in above-mentioned local reference direction, and the voxel of the call number of these voxels in stitching than the optimum of deleting is large.As shown in Figure 7, in X-direction, there are voxel A, B, C, D, voxel A, B, C, the D call number in X-direction increases progressively successively, be for example 1,2,3,4, in the optimum seam of deleting, comprise voxel B, voxel C and D are voxel B voxel below, need be by voxel (not only comprising voxel C and the D) reach after these of voxel B, the opposite direction towards X-axis moves a unit length, supplements the voxel B deleting.
If uphold in the direction of described convergent-divergent, from the set of optimum seam, select successively the optimum seam that energy value is low, will after the optimum seam of selection voxel below, move, then the optimum of Replica Selection is sewed and mend and is filled the position of vacating.As shown in Figure 6, point cloud model is stretched to 1.6 times, also substantially retained the feature of master mould.So-called voxel below, refers in the direction of convergent-divergent, and the voxel in the optimum of selection seam is toward the voxel in above-mentioned local reference direction, and the voxel of the call number of these voxels in stitching than the optimum of selecting is large.As shown in Figure 7, in the optimum of selection seam, comprise voxel A, voxel B, C and D are voxel A voxels below, will after the voxel after these of voxel A, move, and the direction towards X-axis moves a unit length, and in the position of vacating, copies voxel A and supplement entering.
The Zoom method of above-mentioned three-dimensional model, deletes by the seam of two dimensional image the convergent-divergent that reinforcing method has expanded to three-dimensional point cloud model, has greatly widened seam and has deleted reinforcing method in the scope of application in the convergent-divergent field of content-based and feature.For this three-dimensional model form of point cloud model, utilize normal distributing homogeneity to express the feature of model, expression as voxel energy function, and the energy function proposing is also more simply used than traditional slippage analysis and normal curvature analysis, therefore, range of application is wider and have more universality.
The above embodiment has only expressed several embodiment of the present invention, and it describes comparatively concrete and detailed, but can not therefore be interpreted as the restriction to the scope of the claims of the present invention.It should be pointed out that for the person of ordinary skill of the art, without departing from the inventive concept of the premise, can also make some distortion and improvement, these all belong to protection scope of the present invention.Therefore, the protection domain of patent of the present invention should be as the criterion with claims.
Claims (10)
1. a Zoom method for three-dimensional model, described method comprises:
Initial voxel in defining point cloud model and stop voxel, in the tight bounding box that described initial voxel and termination voxel are point cloud model parallel with zoom direction and diagonal angle is parallel two;
Direction to each convergent-divergent, carry out:
Obtain from each initial voxel and arrive the seam set that stops voxel, from described seam set, obtain optimum seam set;
Energy value according to the optimum seam in described optimum seam set, adopts the reinforcing method of deleting of optimum seam to carry out convergent-divergent to point cloud model.
2. method according to claim 1, is characterized in that, described in obtain from each initial voxel and arrive the seam set that stops voxel, from seam set, obtain optimum seam set, comprising:
To each initial voxel, obtain the institute that stops voxel from initial voxel arrival seamed, form seam and gather, the summation that the energy value stitching in described seam set is the energy value of its voxel comprising, the seam that obtains energy value minimum in described seam set is optimum seam, forms optimum seam set.
3. method according to claim 2, is characterized in that, described in obtain from each initial voxel and arrive the seam set that stops voxel, from seam set, obtain optimum seam set, comprising:
To each initial voxel, carry out:
The voxel that finds all and initial voxel to adjoin, obtain the voxel that adjoins accumulative total minimum energy value minimum in voxel, recursion is to stopping voxel so layer by layer, by each, stopping voxel dates back to forward initial voxel and records path, this path is an optimum seam with this termination voxel ending, and the energy value of this optimum seam stops the accumulative total minimum energy value of voxel for it;
The optimum seam that each initial voxel is obtained forms optimum seam set.
4. method according to claim 3, is characterized in that, described in find all and initial voxel to adjoin voxel, obtain the voxel that adjoins in voxel accumulative total minimum energy value minimum, recursion, to stopping voxel, comprising so layer by layer:
From initial voxel, each voxel circulation is carried out to following process, until arrive, stop voxel:
Find 26 of voxel to adjoin voxel, calculate the accumulative total minimum energy value that each adjoins voxel, obtain the voxel that adjoins that wherein adds up minimum energy value minimum.
5. method according to claim 4, is characterized in that, each that adopts that following formula calculates voxel adjoined the accumulative total minimum energy value of voxel:
M(i,j,k)=e(i,j,k)+min{M(i-1,j-1,k+1),M(i,j-1,k+1),M(i+1,j-1,k+1),
M(i-1,j-1,k),M(i,j-1,k),M(i+1,j-1,k),M(i-1,j-1,k-1),M(i,j-1,k-1),M(i+1,j-1,k-1),
M(i-1,j,k+1),M(i,j,k+1),M(i+1,j,k+1),M(i-1,j,k),M(i+1,j,k),M(i-1,j,k-1),
M(i,j,k-1),M(i+1,j,k-1),M(i-1,j+1,k+1),M(i,j+1,k+1),M(i+1,j+1,k+1),
M(i-1,j+1,k),M(i,j+1,k),M(i+1,j+1,k),M(i-1,j+1,k-1),M(i,j+1,k-1),M(i+1,j+1,k-1)}
Wherein, M (i, j, k) is the accumulative total minimum energy value of voxel v (i, j, k), the energy value that e (i, j, k) is voxel, and i, j, k are respectively the call number of voxel in X, Y, Z axis.
6. according to the method described in any one in claim 1 to 5, it is characterized in that, described according to the energy value of the optimum seam in described optimum seam set, the reinforcing method of deleting of the optimum seam of employing is carried out convergent-divergent to point cloud model, comprising:
Described optimum seam set is sorted according to energy value is ascending;
If shrink in the direction of described convergent-divergent, from the set of optimum seam, select successively the optimum seam that energy value is low to delete, and optimum seam of every deletion, the voxel reach below of the optimum seam of described deletion is supplemented to the optimum seam of deleting;
If uphold in the direction of described convergent-divergent, from optimum seam set, select successively the optimum seam that energy value is low, after optimum seam that will described selection voxel below, move, the optimum that then copies described selection is sewed and mend and is filled the position of vacating.
7. method according to claim 6, is characterized in that, described method also comprises:
Adopt three directions of the tight bounding box of principle component analysis calculation level cloud model;
Take and preset unit length as standard is divided into several mikey lattices along described three directions to point cloud model, be voxel.
8. method according to claim 7, is characterized in that, described method also comprises:
In acquisition point cloud model normal information a little;
To each voxel, the energy value that obtains voxel in this voxel normal direction a little depart from the degree of average normal direction.
9. method according to claim 8, is characterized in that, to { the p a little of institute in each voxel
0, p
1..., p
s, these put corresponding normal information is { n
0, n
1..., n
s, the variance of normal direction is
The energy value that calculates voxel is
10. method according to claim 9, is characterized in that, described initial voxel and termination voxel are all effective voxel, in described initial voxel and termination voxel, all comprises a little.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310130687.6A CN104103035B (en) | 2013-04-15 | 2013-04-15 | The Zoom method of threedimensional model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310130687.6A CN104103035B (en) | 2013-04-15 | 2013-04-15 | The Zoom method of threedimensional model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104103035A true CN104103035A (en) | 2014-10-15 |
| CN104103035B CN104103035B (en) | 2017-12-05 |
Family
ID=51671162
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310130687.6A Active CN104103035B (en) | 2013-04-15 | 2013-04-15 | The Zoom method of threedimensional model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104103035B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017133170A1 (en) * | 2016-02-03 | 2017-08-10 | 上海源胜文化传播有限公司 | Three-dimensional human body model scaling system and method |
| CN108009987A (en) * | 2017-12-01 | 2018-05-08 | 中国科学院长春光学精密机械与物理研究所 | A kind of image scaling device and Zoom method |
| CN110163863A (en) * | 2018-11-06 | 2019-08-23 | 腾讯科技（深圳）有限公司 | Three-dimension object dividing method, equipment and medium |
| CN112800511A (en) * | 2020-12-31 | 2021-05-14 | 杭州群核信息技术有限公司 | Home decoration material adjusting method and device, computer equipment and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060170687A1 (en) * | 2003-06-16 | 2006-08-03 | Sony Corporation | Electronic device and its operation explanation display method |
| CN102945551A (en) * | 2012-10-16 | 2013-02-27 | 同济大学 | Graph theory based three-dimensional point cloud data plane extracting method |
-
2013
- 2013-04-15 CN CN201310130687.6A patent/CN104103035B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060170687A1 (en) * | 2003-06-16 | 2006-08-03 | Sony Corporation | Electronic device and its operation explanation display method |
| CN102945551A (en) * | 2012-10-16 | 2013-02-27 | 同济大学 | Graph theory based three-dimensional point cloud data plane extracting method |
Non-Patent Citations (1)
| Title |
|---|
| 项勃: "点云模型局部细节提取与融合技术研究", 《中国优秀硕士学位论文全文数据库 哲学与人文科学辑》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017133170A1 (en) * | 2016-02-03 | 2017-08-10 | 上海源胜文化传播有限公司 | Three-dimensional human body model scaling system and method |
| CN108009987A (en) * | 2017-12-01 | 2018-05-08 | 中国科学院长春光学精密机械与物理研究所 | A kind of image scaling device and Zoom method |
| CN108009987B (en) * | 2017-12-01 | 2021-08-20 | 中国科学院长春光学精密机械与物理研究所 | Image zooming device and zooming method |
| CN110163863A (en) * | 2018-11-06 | 2019-08-23 | 腾讯科技（深圳）有限公司 | Three-dimension object dividing method, equipment and medium |
| WO2020093950A1 (en) * | 2018-11-06 | 2020-05-14 | 腾讯科技（深圳）有限公司 | Three-dimensional object segmentation method and device and medium |
| US11481973B2 (en) | 2018-11-06 | 2022-10-25 | Tencent Technology (Shenzhen) Company Limited | Method, device, and storage medium for segmenting three-dimensional object |
| CN110163863B (en) * | 2018-11-06 | 2022-11-04 | 腾讯科技（深圳）有限公司 | Three-dimensional object segmentation method, apparatus, and medium |
| CN112800511A (en) * | 2020-12-31 | 2021-05-14 | 杭州群核信息技术有限公司 | Home decoration material adjusting method and device, computer equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104103035B (en) | 2017-12-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106023303B (en) | A method of Three-dimensional Gravity is improved based on profile validity and is laid foundations the dense degree of cloud | |
| CN107330885A (en) | A kind of multi-operator image reorientation method of holding important content region the ratio of width to height | |
| CN106097313B (en) | Image segmentation method and device | |
| CN104103035A (en) | Three-dimensional model scaling method | |
| CN106127791A (en) | A kind of contour of building line drawing method of aviation remote sensing image | |
| CN107103277A (en) | A kind of gait recognition method based on depth camera and 3D convolutional neural networks | |
| CN110059760A (en) | Geometric figure recognition methods based on topological structure and CNN | |
| CN104517317A (en) | Three-dimensional reconstruction method of vehicle-borne infrared images | |
| CN104299263A (en) | Method for modeling cloud scene based on single image | |
| CN108876723A (en) | A kind of construction method of the color background of gray scale target image | |
| CN104392433B (en) | Image-based digital oil painting canvas automatic generation method | |
| CN105761202A (en) | Color migration method of color images | |
| CN103198470A (en) | Image cutting method and image cutting system | |
| CN103927785B (en) | A kind of characteristic point matching method towards up short stereoscopic image data | |
| CN106548494A (en) | A kind of video image depth extraction method based on scene Sample Storehouse | |
| Cherian et al. | Accurate 3D ground plane estimation from a single image | |
| CN108958231A (en) | A kind of multiresolution region partitioning method of mobile robot map | |
| CN104992403A (en) | Hybrid operator image redirection method based on visual similarity measurement | |
| CN103761766A (en) | Three-dimensional object model texture mapping algorithm based on tone mapping and image smoothing | |
| CN107301674A (en) | A kind of three-dimensional reconstruction reconstruction method based on window interpolation | |
| CN103871089B (en) | Image superpixel meshing method based on fusion | |
| CN110070608B (en) | Method for automatically deleting three-dimensional reconstruction redundant points based on images | |
| CN106875359B (en) | A kind of sample block image repair method based on layering boot policy | |
| CN104978558B (en) | The recognition methods of target and device | |
| CN105427242A (en) | Interactive network constraint deformation image adaptive scaling method based on content perception |
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 |