CN106126865B - Sense of reality fluid weight emulation mode - Google Patents
Sense of reality fluid weight emulation mode Download PDFInfo
- Publication number
- CN106126865B CN106126865B CN201610620770.5A CN201610620770A CN106126865B CN 106126865 B CN106126865 B CN 106126865B CN 201610620770 A CN201610620770 A CN 201610620770A CN 106126865 B CN106126865 B CN 106126865B
- Authority
- CN
- China
- Prior art keywords
- block
- frequency spectrum
- density
- fluid
- particle
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Processing Or Creating Images (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of heavy emulation mode of sense of reality fluid, this method is intended to the sample frame using video fluid, obtains the heavy simulation result of the sense of reality.This method is first using the physical drives data for calculating sample frame fluid, pass through wavelet transformation, the radio-frequency component of driving data is obtained in domain space, and then during fluid emulates again, the driving data in LBM simulation process is enhanced using the high frequency detail ingredient of extraction, to obtain the heavy simulation result of realistic details.The present invention can be effectively applied in the research and application of the reverse engineering of fluid, and the emulation again of the fluid scene sense of reality, acquisition and the close sense of reality effect of sample fluid appearance may be implemented, overcome the problem of lacking details in the existing fluid sense of reality.
Description
Technical field
The present invention relates to sense of reality fluid weight emulation modes, using wavelet transform, in frequency domain high-frequency region, extract
The details of physical drives data；During the deduction that fluid emulates again, the dynamic details enhancing of the radio-frequency component of extraction is arrived
In the frequency spectrum driving data of simulation process, so as in spectrum space realize fluid emulate again in driving data increasing
By force, and then the heavy simulation result with abundant details is obtained.
Background technique
Sense of reality fluid emulates again and has widely been applied to game design at present, fluid animation emulates, the spy of film
Imitate the fields such as design, fluid Military Simulation.In recent years, the research of video flowing weight emulation technology is concerned, and the purpose is to fill
Divide using the driving data in video fluid, obtains the real-time simulation effect of the sense of reality.At present in research, people are used in low-dimensional
The time performance of emulation is obtained in space by the method for reducing dimension, to occur lacking the sense of reality in weight simulation process
The problem of details.Currently, in existing technology, although there is the method for some fluid sense of reality emulation, utilize video
It is still a challenging problem that fluid, which realizes that the fluid of the sense of reality emulates again,.
Summary of the invention
It is emulated again the purpose of the present invention is in view of the deficiencies of the prior art, and for the sense of reality in video fluid reverse engineering
Needs, a kind of measure based on physical motion data-driven, domain space enhances physical drives data is provided, utilize
This method can be realized the fluid sense of reality in video fluid reverse engineering and emulate again.
Realizing the specific technical solution of the object of the invention is:
A kind of sense of reality fluid weight emulation mode, which is characterized in that using the first two sample frame of video, be denoted as F1And F2,
Using the frequency domain high frequency components of density, realize that the fluid sense of reality emulates again, the specific steps are as follows:
Step 1: pretreatment
Calculate F1In any one particle height Hi, utilize F1And F2Calculate F1The speed v of middle particleiAnd density Pi,
Middle i is nonnegative integer, and 0≤i < N, N are the particle numbers of fluid scene；
(a) the frequency spectrum set of blocks S of standard density is establishedF, specifically:
By F1In the processing of all particle piecemeals all blocks are determined as and according to the direction of motion of particle in every piece
It is different classes of；Division and the label for first carrying out particle block, by F1In all particles be divided into square according to specified size b × b
Block, if the resolution ratio of image be X × Y, horizontal segmentation block numberVertical segmentation block numberIt indicates to be rounded downwards；Frequency spectrum set of blocks S is constructed using this WA frequency spectrum blocksF；WhereinFor F1In it is any one
A block, it is located at w block horizontal in spectrum space and vertical h block, and w and h are nonnegative integer, 0≤w < W, 0≤h < A；
Assuming that F1In any one blockIn the mean values of all particle 2D speed beAnd u isHorizontal direction point
Amount, v areVertically-oriented component, thenType T utilize (1) formula determine:
Calculate F1In all particle densities, be called standard density, by standard density according to the specified number of plies carry out it is discrete small
Wave conversion, it is assumed that number of plies L, then transformed high frequency region one shares 3L, for any one area in this 3L high frequency region
Ak(0≤k < 3L), piecemeal according to the method described above: being horizontally divided into W block, and vertical segmentation is A block；In each high frequency region, by WA
A frequency spectrum blocks construct four set as followsSpecifically: for AkAny one block in (0≤k < 3L)
bj(0≤j < W*H), if AkIn SFThe type of middle corresponding blocks is assumed to be T, then, by bjMerging set
According to this quadrat method, the set of each four frequency spectrums in region is constructedClaimFor standard density frequency
Compose block collection；
(b) to the frequency spectrum block sequencing inside each set, specifically:
Count each setThe frequency spectrum of internal all particles and, and will setFrequency
Block is composed, is sorted from large to small according to statistical result, the set after sequence is denoted asClaimFor mark
Quasi- density spectral block ordered set；
Step 2: according to the following steps, carrying out N altogetherRSecondary emulation is deduced:
(a) when it is previous be I time emulation；If meeting I≤NR, NREmulation altogether carry out deduction number, continue with
Otherwise lower step jumps to step 3；
(b) speed v is utilizedi(0≤i < Z) initializes LBM model, emulation is made to start to deduce, then right according to the following steps
Density after deduction is enhanced, and Z is the particle number of fluid scene, if streaming video F1Resolution ratio be g × m, then Z
It may be calculated Z=gm；During deduction, the density p of fluid scene particle in I time deduction is obtainedl(0≤l < Z) (claims pl
To deduce density, Z is the particle number of fluid scene), wherein I is natural number (1≤I≤NR)；Then by plAccording to step 1
(a) and (b) establishes plFrequency spectrum blocks ordered set
(c) enhancing of density is deduced
Tentative standard density spectral block ordered setIn have c frequency spectrum blocks, deduce density
Frequency spectrum blocks ordered setIn have e frequency spectrum blocks；
For specified threshold θ (0 < θ < 1), in deducing for the m times,In preceding θ c
Frequency spectrum blocks carry out enhancing processing；Specifically: deducing the frequency spectrum blocks b of densityiBlock number i calculates standard density frequency spectrum blocks BsBlock number s:
S=ci/e (2)
Selection standard density spectral block Bs, then for biThe density spectral p of middle particletEnhancing are as follows:
pt=KtQb/XB (3)
Wherein KtFor BsThe density spectral of the particle of middle serial number t；QbForThe sum of all frequency spectrums, X in frequency spectrum blocksBFor
The sum of all frequency spectrums in frequency spectrum blocks；
(d) 2 are gone to step, is deduced next time；
Step 3: emulation terminates again.
The present invention utilizes the physical drives data of video fluid, realizes that realistic video fluid emulates again, overcomes existing
The problem of lacking details in simulation study again, may be implemented the video flowing weight emulation with abundant details, be suitable for
Including the multiple fluids type such as advection and wave.It is thin that a kind of sense of reality for obtaining poplar example frame in domain space is proposed in the present invention
Save and emulate deduce during, using obtain fluid density high frequency dynamic details enhance Simulation drive data, so as to
To realize that the fluid of the sense of reality emulates again.
Detailed description of the invention
Fig. 1 is the sense of reality of embodiment of the present invention fluid weight simulation sequence result figure；
Fig. 2 is that the embodiment of the present invention is emulating result figure obtained in marginal testing again；
Fig. 3 is the comparative result figure of the sense of reality of embodiment of the present invention fluid weight simulation result and existing method.
Specific embodiment
Embodiment
Invention is further illustrated with reference to the accompanying drawing.
The present embodiment implements stream using the streaming video (such as the streaming videos such as 54ab110) in DynTex dynamic texture library
The body scene sense of reality emulates again.Implemented under Windows7 operating system in PC machine, hardware configuration is 2.66GHz
Intel Core(TM)2Duo CPU、4GB RAM。
(referred to as sample frame, is denoted as F to two frames of selection streaming video (such as the streaming videos such as 54ab110)1And F2) realize stream
Body copying, since simulation initialisation condition derives from sample frame, it is therefore an objective in the physical drives number of streaming video sample frame
After initialization, sense of reality fluid weight simulation process is realized, the present invention follows the steps below:
Step 1: pretreatment
Calculate F1In any one particle height Hi, utilize F1And F2Calculate F1The speed v of middle particleiAnd density Pi,
Middle i is nonnegative integer, and 0≤i < N, N are the particle numbers of fluid scene；
(a) the frequency spectrum set of blocks S of standard density is establishedF, specifically:
By F1In the processing of all particle piecemeals all blocks are determined as and according to the direction of motion of particle in every piece
It is different classes of；Division and the label for first carrying out particle block, by F1In all particles be divided into square according to specified size b × b
Block, if the resolution ratio of image be X × Y, horizontal segmentation block numberVertical segmentation block numberIt indicates to be rounded downwards；Frequency spectrum set of blocks S is constructed using this WA frequency spectrum blocksF；WhereinFor F1In it is any one
A block, it is located at w block horizontal in spectrum space and vertical h block, and w and h are nonnegative integer, 0≤w < W, 0≤h < A；
Assuming that F1In any one blockIn the mean values of all particle 2D speed beAnd u isHorizontal direction point
Amount, v areVertically-oriented component, thenType T utilize (1) formula determine:
Calculate F1In all particle densities, be called standard density, by standard density according to the specified number of plies carry out it is discrete small
Wave conversion, it is assumed that number of plies L, then transformed high frequency region one shares 3L, for any one area in this 3L high frequency region
Ak(0≤k < 3L), piecemeal according to the method described above: being horizontally divided into W block, and vertical segmentation is A block；In each high frequency region, by WA
A frequency spectrum blocks construct four set as followsSpecifically: for AkAny one block in (0≤k < 3L)
bj(0≤j < W*H), if AkIn SFThe type of middle corresponding blocks is assumed to be T, then, by bjMerging set
According to this quadrat method, the set of each four frequency spectrums in region is constructedClaimFor standard density
Frequency spectrum blocks collection；
(b) to the frequency spectrum block sequencing inside each set, specifically:
Count each setThe frequency spectrum of internal all particles and, and will setFrequency
Block is composed, is sorted from large to small according to statistical result, the set after sequence is denoted asClaimFor mark
Quasi- density spectral block ordered set；
Step 2: according to the following steps, carrying out N altogetherRSecondary emulation is deduced:
(a) when it is previous be I time emulation；If meeting I≤NR, NREmulation altogether carry out deduction number, continue with
Otherwise lower step jumps to step 3；
(b) speed v is utilizedi(0≤i < Z) initializes LBM model, emulation is made to start to deduce, then right according to the following steps
Density after deduction is enhanced, and Z is the particle number of fluid scene, if streaming video F1Resolution ratio be g × m, then Z
It may be calculated Z=gm；During deduction, the density p of fluid scene particle in I time deduction is obtainedl(0≤l < Z) (claims pl
To deduce density, Z is the particle number of fluid scene), wherein I is natural number (1≤I≤NR)；Then by plAccording to step 1
(a) and (b) establishes plFrequency spectrum blocks ordered set
(c) enhancing of density is deduced
Tentative standard density spectral block ordered setIn have c frequency spectrum blocks, deduce density
Frequency spectrum blocks ordered setIn have e frequency spectrum blocks；
For specified threshold θ (0 < θ < 1), in deducing for the m times,In preceding θ c
Frequency spectrum blocks carry out enhancing processing；Specifically: deducing the frequency spectrum blocks b of densityiBlock number i calculates standard density frequency spectrum blocks BsBlock number s:
S=ci/e (2)
Selection standard density spectral block Bs, then for biThe density spectral p of middle particletEnhancing are as follows:
pt=KtQb/XB (3)
Wherein KtFor BsThe density spectral of the particle of middle serial number t；QbForThe sum of all frequency spectrums, X in frequency spectrum blocksBFor
The sum of all frequency spectrums in frequency spectrum blocks；
(d) 2 are gone to step, is deduced next time.
Step 3: emulation terminates again.
Fig. 1 is emulated again using the present invention, obtained time series result.It can be bright from the heavy simulation result in figure
It is aobvious to find out, the timing results of available realistic details when being emulated using the present invention again.This illustrates the present invention
In validity.
Fig. 2 is sense of reality fluid of embodiment of the present invention weight simulation result diagram, is emulating result obtained in marginal testing again
Figure, the measure that the result of Cong Tuzhong can be seen that in the present invention using frequency components enhancing is very effective, in the example of marginal testing
The details of middle flow surface still is able to embody.
Fig. 3 is the comparative result figure of the sense of reality of embodiment of the present invention fluid weight simulation result and existing method.In Fig. 3 result
In, the column of right side one are sense of reality fluid of embodiment of the present invention weight simulation results, and left side is the emulation again realized using existing method
As a result, can be seen that from the result of comparison, the measure that sense of reality fluid emulates again in the present invention is highly effective, and emulation details is bright again
It is aobvious.
Claims (1)
1. a kind of sense of reality fluid weight emulation mode, which is characterized in that using the first two sample frame of video, be denoted as F1And F2, benefit
With the frequency domain high frequency components of density, realize that the fluid sense of reality emulates again, the specific steps are as follows:
Step 1: pretreatment
Calculate F1In any one particle height Hi, utilize F1And F2Calculate F1The speed v of middle particleiAnd density Pi, wherein i be
Nonnegative integer, and 0≤i < N, N are the particle numbers of fluid scene；
(a) the frequency spectrum set of blocks S of standard density is establishedF, specifically:
By F1In the processing of all particle piecemeals all blocks are determined as inhomogeneity and according to the direction of motion of particle in every piece
Not；Division and the label for first carrying out particle block, by F1In all particles according to specified size b × b be divided into square block, such as
The resolution ratio of fruit image is X × Y, then, horizontal segmentation block numberVertical segmentation block number It indicates
It is rounded downwards；Frequency spectrum set of blocks S is constructed using this WA frequency spectrum blocksF；WhereinFor F1In any one block, it be located at frequency spectrum sky
Between middle level w block and vertical h block, w and h are nonnegative integer, 0≤w < W, 0≤h < A；
Assuming that F1In any one blockIn the mean values of all particle 2D speed beAnd u isHorizontal direction component, v beVertically-oriented component, thenType T utilize (1) formula determine:
Calculate F1In all particle densities, be called standard density, by standard density according to the specified number of plies carry out discrete wavelet transformer
It changes, it is assumed that number of plies L, then transformed high frequency region one shares 3L, for any one area A in this 3L high frequency regionk(0
≤ k < 3L), piecemeal according to the method described above: being horizontally divided into W block, and vertical segmentation is A block；In each high frequency region, by WA frequency
Spectrum block constructs four set as followsSpecifically: for AkAny one block b in (0≤k < 3L)j(0
≤ j < W*H), if AkIn SFThe type of middle corresponding blocks is assumed to be T, then, by bjMerging setIn, according to
This quadrat method constructs the set of each four frequency spectrums in regionClaimFor standard density frequency spectrum
Block collection；
(b) to the frequency spectrum block sequencing inside each set, specifically:
Count each setThe frequency spectrum of internal all particles and, and will setFrequency spectrum
Block is sorted from large to small according to statistical result, and the set after sequence is denoted asClaimFor standard
Density spectral block ordered set；
Step 2: according to the following steps, carrying out N altogetherRSecondary emulation is deduced:
(a) when it is previous be I time emulation；If meeting I≤NR, NRThe deduction number that emulation carries out altogether, continues following step
Suddenly, otherwise, step 3 is jumped to；
(b) speed v is utilizedi(0≤i < Z) initializes LBM model, makes emulation start to deduce, then according to the following steps to deduction
Density afterwards is enhanced, and Z is the particle number of fluid scene, if streaming video F1Resolution ratio be g × m, then Z can be with
It is calculated as Z=gm；During deduction, the density p of fluid scene particle in I time deduction is obtainedl(0≤l < Z), plTo deduce
Density, Z are the particle numbers of fluid scene, and wherein I is natural number (1≤I≤NR)；Then by plAccording to (a) and (b) of step 1
Establish plFrequency spectrum blocks ordered set
(c) enhancing of density is deduced
Tentative standard density spectral block ordered setIn have c frequency spectrum blocks, deduce density frequency spectrum
Block ordered setIn have e frequency spectrum blocks；
For specified threshold θ (0 < θ < 1), in deducing for the m times,In preceding θ c frequency spectrum
Block carries out enhancing processing；Specifically: deducing the frequency spectrum blocks b of densityiBlock number i calculates standard density frequency spectrum blocks BsBlock number s:
S=ci/e (2)
Selection standard density spectral block Bs, then for biThe density spectral p of middle particletEnhancing are as follows:
pt=KtQb/XB (3)
Wherein KtFor BsThe density spectral of the particle of middle serial number t；QbForThe sum of all frequency spectrums, X in frequency spectrum blocksBForFrequency spectrum
The sum of all frequency spectrums in block；
(d) 2 are gone to step, is deduced next time；
Step 3: emulation terminates again.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610620770.5A CN106126865B (en) | 2016-08-01 | 2016-08-01 | Sense of reality fluid weight emulation mode |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610620770.5A CN106126865B (en) | 2016-08-01 | 2016-08-01 | Sense of reality fluid weight emulation mode |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106126865A CN106126865A (en) | 2016-11-16 |
| CN106126865B true CN106126865B (en) | 2019-04-26 |
Family
ID=57255271
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610620770.5A Expired - Fee Related CN106126865B (en) | 2016-08-01 | 2016-08-01 | Sense of reality fluid weight emulation mode |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106126865B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103413346A (en) * | 2013-04-09 | 2013-11-27 | 华东师范大学 | Method for real-time reconstruction of realistic fluid and system thereof |
| CN104517299A (en) * | 2014-12-19 | 2015-04-15 | 华东师范大学 | Method for restoring and resimulating physical video fluid driving model |
-
2016
- 2016-08-01 CN CN201610620770.5A patent/CN106126865B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103413346A (en) * | 2013-04-09 | 2013-11-27 | 华东师范大学 | Method for real-time reconstruction of realistic fluid and system thereof |
| CN104517299A (en) * | 2014-12-19 | 2015-04-15 | 华东师范大学 | Method for restoring and resimulating physical video fluid driving model |
Non-Patent Citations (4)
| Title |
|---|
| Numerical study of flow past an impulsively started cylinder by the lattice-Boltzmann method;YANBING LI 等;《Journal of Fluid Mechanics》;20041125;第519卷;273-300 |
| Subspace Fluid Re-Simulation;Theodore Kim 等;《ACM Transactions on Graphics》;20130731;第32卷(第4期);62:1-62:11 |
| 流体真实性仿真的国内外研究进展;王树杰 等;《系统仿真学报》;20100228;第22卷(第2期);280-286 |
| 真实感流体实时重建;俞铭琪 等;《计算机辅助设计与图形学学报》;20130515;第25卷(第5期);622-630 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106126865A (en) | 2016-11-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hu et al. | Pose-guided photorealistic face rotation | |
| Wang et al. | Action recognition from depth maps using deep convolutional neural networks | |
| US20120316848A1 (en) | Sph fluid simulation method and system for multi-level vorticity, recording medium for the same | |
| CN109858461A (en) | A kind of method, apparatus, equipment and storage medium that dense population counts | |
| CN109033945B (en) | Human body contour extraction method based on deep learning | |
| Raveendran et al. | Blending liquids | |
| CN105551072B (en) | Realize the method and system of actor model part real time kinematics | |
| CN104240277B (en) | Augmented reality exchange method and system based on Face datection | |
| Diba et al. | Temporal 3d convnets using temporal transition layer | |
| CN106651926A (en) | Regional registration-based depth point cloud three-dimensional reconstruction method | |
| CN103902989B (en) | Human action video frequency identifying method based on Non-negative Matrix Factorization | |
| CN105513014B (en) | A kind of multi-frame image super-resolution reconstruction method and its reconstructing system | |
| CN110060286A (en) | A kind of monocular depth estimation method | |
| CN103942090B (en) | Data-driven real-time hair motion simulation method | |
| WO2016127421A1 (en) | Real-time motion simulation method for hair and object collisions | |
| CN110598646B (en) | Depth feature-based unconstrained repeated action counting method | |
| CN104680567B (en) | According to curvature, threedimensional model is carried out the method for frame cluster | |
| CN1920880A (en) | Video flow based people face expression fantasy method | |
| CN103839280B (en) | A kind of human body attitude tracking of view-based access control model information | |
| CN106126865B (en) | Sense of reality fluid weight emulation mode | |
| Zeng et al. | Denoising stochastic progressive photon mapping renderings using a multi-residual network | |
| CN102831616B (en) | Video stream motion vector calculation method | |
| CN106485703B (en) | Fuzzy detection method based on image gradient dct transform | |
| CN106294955B (en) | A kind of sense of reality fluid weight emulation mode of physical data driving | |
| CN103871096B (en) | Sense of reality fluid Scene Composition methods in three dimensions |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190426 Termination date: 20210801 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |