CN105608727B - A kind of offshore of data-driven is surged animation synthesizing method and system - Google Patents

A kind of offshore of data-driven is surged animation synthesizing method and system Download PDF

Info

Publication number
CN105608727B
CN105608727B CN201610115647.8A CN201610115647A CN105608727B CN 105608727 B CN105608727 B CN 105608727B CN 201610115647 A CN201610115647 A CN 201610115647A CN 105608727 B CN105608727 B CN 105608727B
Authority
CN
China
Prior art keywords
surging
height
water surface
surged
target shape
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
Application number
CN201610115647.8A
Other languages
Chinese (zh)
Other versions
CN105608727A (en
Inventor
朱登明
商柳
张娟
魏毅
王兆其
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Institute of Computing Technology of CAS
Original Assignee
Institute of Computing Technology of CAS
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Institute of Computing Technology of CAS filed Critical Institute of Computing Technology of CAS
Priority to CN201610115647.8A priority Critical patent/CN105608727B/en
Publication of CN105608727A publication Critical patent/CN105608727A/en
Application granted granted Critical
Publication of CN105608727B publication Critical patent/CN105608727B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T13/00—Animation
    • G06T13/20—3D [Three Dimensional] animation
    • G06T13/60—3D [Three Dimensional] animation of natural phenomena, e.g. rain, snow, water or plants

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Processing Or Creating Images (AREA)

Abstract

The present invention proposes that a kind of offshore of data-driven is surged animation synthesizing method and system, it is related to image procossing and computer vision field, this method includes being rebuild to the monocular video of water surface video, obtain the height field data of the monocular video, opening operation is carried out to the water surface video to rebuild and closed operation reconstruction, acquisition is surged mask, is handled the mask of surging by Morphological Thinning Algorithm, and acquisition is surged framework information；According to the mask of surging, the framework information of surging, quick surging height variation model is created, by the quick surging height variation model and according to the height field data, getting frame sequence chart；The target shape of user's input surge is obtained, and the original shape of surging in the frame sequence figure is changing into the target shape, and the target shape is generated into animation.The present invention improves the durability for rebuilding data；The present invention make to surge information and control mode it is simpler intuitive.

Description

A kind of offshore of data-driven is surged animation synthesizing method and system
Technical field
The present invention relates to image procossing and computer vision field, more particularly to a kind of offshore of data-driven is surged animation Synthetic method and system.
Background technology
The simulation of natural phenomena is always one of the hot issue of computer graphics study, wherein fluid simulation, especially It is the rapid modeling of the large area water surface and wave and controllable cartoon technique, is developed, virtually in video display special efficacy, advertisement, 3d gaming The various fields such as reality are more and more widely used, and have important application value, however, natural wave is a kind of non- Often complicated physical phenomenon, due to the shadow by many factors such as whirling motions inside underwater topography, barrier, wind energy input, wave Ring, most of wave height is uneven, different in size, when and show as opposite being in periodic ripples, when and also have form violent Variation, the broken sea that can be crimped and divide, the spray also to splash, geometry have extremely strong irregular characteristic, for many years Research on fluid simulation, develops to complicated hydrodinamical model from simple empirical model, the water of synthesis it is true Sense is continuously improved, and required calculation amount is also continuously increased, therefore, how quickly to be simulated to wave, and generate controllable wave Unrestrained animation is still a challenging project.
Method And Principle based on corrugated moulding is simple, and computational efficiency is very high, but the wave feature generated is single, and the sense of reality is not By force；Method based on physics is the mainstream of nowadays wave simulation method, by solving hydrodinamical model come mimetic surface, is closed At wave effect it is true to nature, but calculate complicated, simulation precision is very low.In recent years, the water surface method for reconstructing based on measured data Start to grow up, image data is embodied by the nature actual water captured to collecting device and is rebuild, can quickly be obtained Water surface, and the sense of reality is very strong, but limited by collecting device, it is relatively more single on motion morphology to reconstruct the water surface come One, the existing method based on measured data focuses on the research of different method for reconstructing mostly, is the process of " reproduction ", to how It makes full use of and rebuilds the research not yet of the more various wave animation of data generation, in addition, existing controlled fluid cartoon technique Also the method for being all based on physics has considerable restraint in computational efficiency and control mode.
Some concepts related to the present invention：
Reflectivity (reflectance)：Indicate that the numerical value of object reflection incident light size is called reflectivity, by body surface Reflectible light quantity and its ratio between light quantity for being received be calculated, commonly use percentage and fractional representation.When light beam is close to just When incident (incidence angle θ is approximately equal to 0), reflectivity calculation formula is：Wherein n1,n2It is two media respectively True refractive index (i.e. relative to the refractive index of vacuum), refractive index refer to showing of changing of angle when light enters different medium As usingCharacterization, θ1, θ2The angle of respectively incidence angle and refraction angle, i.e. light and normal, usually, light are facing Only with the physical property of medium, the wavelength and incidence angle of light be related for reflectivity on interface.
Comentropy (informationentropy)：Usual information number be difficult measurement, the proposition of " comentropy " concept, Solves the problems, such as the quantisation metric to information.This word of comentropy is that Shannon is borrowed from thermodynamics, in thermodynamics Hot entropy is the physical quantity for indicating molecular state confusion degree, and the concept of Shannon comentropy describes the uncertainty of information source, can Comentropy is understood as the probability of occurrence of certain specific information, it is however generally that, when a kind of information probability of occurrence higher, Show that it is transmitted more extensive, in other words, the degree higher being cited, it is believed that from the perspective of information propagation, letter Breath entropy can indicate the value of information, the standard of our the so scaling information value height with regard to there are one.
Water surface elevation field (hight field)：Indicate the relative altitude situation of each position of the water surface.In the animation of the water surface In, human eye it can be seen that part be essentially the situation that springs up of the water surface, therefore can be the water surface by the reconstruction abbreviation of water surface animation The reconstruction of Partial Height, by carrying out the subdivision of gridding to the water surface, the configuration of the water surface can be expressed as all subdivision points The altitude information of the combination of height, each final point has been combined into the height field of the water surface.
It is computationally intensive needed for traditional method based on physics in the above common method, it is time-consuming longer, and to parameter setting More sensitive, for the detailed simulation of complex characteristic, there are certain technical difficulty.The water surface of existing data-driven is rebuild Method then needs complicated data acquisition device design mostly, and has to water body and require or pre-process to a certain degree, Using structure light, predetermined pattern as label etc., these requirements can meet in room conditions, but be difficult to obtain reality outdoors It applies.
Invention content
In view of the deficiencies of the prior art, the present invention proposes that a kind of offshore of data-driven is surged and animation synthesizing method and is System.
The present invention proposes that a kind of offshore of data-driven is surged animation synthesizing method, including：
Step 1, the monocular video of water surface video is rebuild, the height field data of the monocular video is obtained, to institute It states water surface video and carries out opening operation reconstruction and closed operation reconstruction, obtain mask of surging, gushed to described by Morphological Thinning Algorithm Unrestrained mask is handled, and framework information of surging is obtained；
Step 2, according to the mask of surging, the framework information of surging, quick surging height variation model is created, is passed through The quick surging height variation model and according to the height field data, getting frame sequence chart；
Step 3, the target shape of user's input surge is obtained, and the original shape of surging in the frame sequence figure is changed Animation is generated at the target shape, and by the target shape.
The quick surging height variation model is specially：In order to keep the thin of the water surface while surging smooth decline Section divides surging height h for the basal surface height h that surges when startingbaseWith water surface details height hdetailTwo parts：
H=hbase+hdetail
Wherein, by height field data described in Gaussian smoothing, the approximate water surface elevation of the water surface video is extracted, it will be described Height field data subtracts the approximate water surface elevation, obtains water surface details height hdetail；
Line part of surging finally drops to height same as surrounding, and the height finally dropped to is hend,
Wherein w indicates the image of surging in the water surface video,It surges described in expression mask, dilate () table Show expansive working, in each frame of the water surface video, surge the height h for needing to declinediffFor：
hdiff=(h-hend)(1-ratio)
Wherein ratio, which indicates to surge, highly compares the difference in height and original height difference of basal surface of surging after line part declines The calculation of ratio, different drop modes, ratio is different, if declined according to exponential form, the ratio of the i-th frame takes Value is：
Wherein N indicates the totalframes of the water surface video, if it is linear forms, then
In order to make to surge, nature declines, by hdiffAs needing the height declined on skeleton of surging, and the point around skeleton It is adopted according to the grown form surged using radial basis function to control the practical falling head for line each point in the radial direction of surging Use Gaussian function G as radial basis function：
It is the position of the i-th row of wave skeleton.
User by parameter curve and sketches the contours two kinds of input modes of curve in the step 3, obtains the target shape.
The shape surged is indicated with the pixel on curve, then original shape is expressed as with target shape：
Wherein,It surges described in expression original shape, piThe point surged described in expression on original shape, xi、yiIt indicates Point piTransverse and longitudinal coordinate；Indicate the target shape, qiIndicate the point in the target shape, x 'i、y′iIndicate its point qiTransverse and longitudinal coordinate, n indicate point number,Indicate integer, put in surge original state and the target shape Number is identical, and coordinate is all integer, and yi=y 'i；
By the original shape of surging to the target shape, shape sequence is obtained WhereinIt is obtained by linear differenceThat is the target shape of the i-th frame, w For the image of surging in the water surface video：
NframeIndicate the total quantity of the frame of deformation process experience.
The cavity that translation leaves of surging is filled up using nonparametric sample texture synthetic method.
The present invention also proposes that a kind of offshore of data-driven is surged animation compound system, including：
Acquisition is surged framework information module, is rebuild for the monocular video to water surface video, is obtained the monocular and regard The height field data of frequency carries out opening operation reconstruction to the water surface video and is rebuild with closed operation, obtains mask of surging, pass through form It learns thinning algorithm to handle the mask of surging, obtains framework information of surging；
Getting frame sequence module creates height of quickly surging for mask, the framework information of surging of surging according to Variation model is spent, by the quick surging height variation model and according to the height field data, getting frame sequence chart；
Generate animation, the target shape for obtaining user's input surge, and by surging in the frame sequence figure Original shape is changing into the target shape, and the target shape is generated animation.
The quick surging height variation model is specially：In order to keep the thin of the water surface while surging smooth decline Section divides surging height h for the basal surface height h that surges when startingbaseWith water surface details height hdetailTwo parts：
H=hbase+hdetail
Wherein, by height field data described in Gaussian smoothing, the approximate water surface elevation of the water surface video is extracted, it will be described Height field data subtracts the approximate water surface elevation, obtains water surface details height hdetail；
Line part of surging finally drops to height same as surrounding, and the height finally dropped to is hend,
Wherein w indicates the image of surging in the water surface video,It surges described in expression mask, dilate () table Show expansive working, in each frame of the water surface video, surge the height h for needing to declinediffFor：
hdiff=(h-hend)(1-ratio)
Wherein ratio, which indicates to surge, highly compares the difference in height and original height difference of basal surface of surging after line part declines The calculation of ratio, different drop modes, ratio is different, if declined according to exponential form, the ratio of the i-th frame takes Value is：
Wherein N indicates the totalframes of the water surface video, if it is linear forms, then
In order to make to surge, nature declines, by hdiffAs needing the height declined on skeleton of surging, and the point around skeleton It is adopted according to the grown form surged using radial basis function to control the practical falling head for line each point in the radial direction of surging Use Gaussian function G as radial basis function：
It is the position of the i-th row of wave skeleton.
User by parameter curve and sketches the contours two kinds of input modes of curve in the generation animation, obtains the target Shape.
The shape surged is indicated with the pixel on curve, then original shape is expressed as with target shape：
Wherein,It surges described in expression original shape, piThe point surged described in expression on original shape, xi、yiIt indicates Point piTransverse and longitudinal coordinate；Indicate the target shape, qiIndicate the point in the target shape, x 'i、y′iIndicate its point qiTransverse and longitudinal coordinate, n indicate point number,Indicate integer, put in surge original state and the target shape Number is identical, and coordinate is all integer, and yi=y 'i；
By the original shape of surging to the target shape, shape sequence is obtained WhereinIt is obtained by linear differenceThat is the target shape of the i-th frame, w For the image of surging in the water surface video：
NframeIndicate the total quantity of the frame of deformation process experience.
The cavity that translation leaves of surging is filled up using nonparametric sample texture synthetic method.
As it can be seen from the above scheme the advantage of the invention is that：
The information of surging that the present invention rebuilds video data is deducted a percentage and is edited, and is that Data Synthesis is various to be gushed using rebuilding The preliminary trial of unrestrained animation improves the durability for rebuilding data；The present invention make to surge information and control mode it is simpler straight It sees；The present invention improves the diversity for animation of surging, and can quickly generate user controllable height and shape, and have high realism The animation of surging of abundant details.
Description of the drawings
Fig. 1 is that the offshore of data-driven is surged animation synthesizing method flow chart；
Fig. 2 a are wave images；
Fig. 2 b are the images for being superimposed mask；
Fig. 2 c are the images for being superimposed skeleton；
Fig. 3 a are video images；
Fig. 3 b are the height field figures rebuild；
Fig. 3 c are linear decline result figures；
Fig. 3 d are index decreased result figures；
Fig. 4 a are the conic section figures of input；
Fig. 4 b are the curve graphs sketched the contours；
Fig. 4 c are the refinement curve graphs for sketching the contours curve extraction；
Fig. 5 a are initial wave charts；
Fig. 5 b are the wave charts for generating cavity；
Fig. 5 c are hole-filling design sketch；
Fig. 6 is the wave transformation effect diagram for sketching the contours curve.
Specific implementation mode
In order to solve the technical problems existing in the prior art, the present invention proposes that a kind of offshore of data-driven is surged animation Synthetic method, including：
Step 1, the monocular video of water surface video is rebuild, the height field data of the monocular video is obtained, to institute It states water surface video and carries out opening operation reconstruction and closed operation reconstruction, obtain mask of surging, gushed to described by Morphological Thinning Algorithm Unrestrained mask is handled, and framework information of surging is obtained；
Step 2, according to the mask of surging, the framework information of surging, quick surging height variation model is created, is passed through The quick surging height variation model and according to the height field data, getting frame sequence chart；
Step 3, the target shape of user's input surge is obtained, and the original shape of surging in the frame sequence figure is changed Animation is generated at the target shape, and by the target shape.
The quick surging height variation model is specially：In order to keep the thin of the water surface while surging smooth decline Section divides surging height h for the basal surface height h that surges when startingbaseWith water surface details height hdetailTwo parts：
H=hbase+hdetail
Wherein, by height field data described in Gaussian smoothing, the approximate water surface elevation of the water surface video is extracted, it will be described Height field data subtracts the approximate water surface elevation, obtains water surface details height hdetail；
Line part of surging finally drops to height same as surrounding, and the height finally dropped to is hend,
Wherein w indicates the image of surging in the water surface video,It surges described in expression mask, dilate () table Show expansive working, in each frame of the water surface video, surge the height h for needing to declinediffFor：
hdiff=(h-hend)(1-ratio)
Wherein ratio, which indicates to surge, highly compares the difference in height and original height difference of basal surface of surging after line part declines The calculation of ratio, different drop modes, ratio is different, if declined according to exponential form, the ratio of the i-th frame takes Value is：
Wherein N indicates the totalframes of the water surface video, if it is linear forms, then
In order to make to surge, nature declines, by hdiffAs needing the height declined on skeleton of surging, and the point around skeleton It is adopted according to the grown form surged using radial basis function to control the practical falling head for line each point in the radial direction of surging Use Gaussian function G as radial basis function：
It is the position of the i-th row of wave skeleton.
User by parameter curve and sketches the contours two kinds of input modes of curve in the step 3, obtains the target shape.
The shape surged is indicated with the pixel on curve, then original shape is expressed as with target shape：
Wherein,It surges described in expression original shape, piThe point surged described in expression on original shape, xi、yiIt indicates Point piTransverse and longitudinal coordinate；Indicate the target shape, qiIndicate the point in the target shape, x 'i、y′iIndicate its point qiTransverse and longitudinal coordinate, n indicate point number,Indicate integer, put in surge original state and the target shape Number is identical, and coordinate is all integer, and yi=y 'i；
By the original shape of surging to the target shape, shape sequence is obtained WhereinIt is obtained by linear differenceThat is the target shape of the i-th frame, w For the image of surging in the water surface video：
NframeIndicate the total quantity of the frame of deformation process experience.W is the image of surging in the water surface video,For It surges original shape,For target shape of surging.
The cavity that translation leaves of surging is filled up using nonparametric sample texture synthetic method.
The present invention also proposes that a kind of offshore of data-driven is surged animation compound system, including：
Acquisition is surged framework information module, is rebuild for the monocular video to water surface video, is obtained the monocular and regard The height field data of frequency carries out opening operation reconstruction to the water surface video and is rebuild with closed operation, obtains mask of surging, pass through form It learns thinning algorithm to handle the mask of surging, obtains framework information of surging；
Getting frame sequence module creates height of quickly surging for mask, the framework information of surging of surging according to Variation model is spent, by the quick surging height variation model and according to the height field data, getting frame sequence chart；
Generate animation, the target shape for obtaining user's input surge, and by surging in the frame sequence figure Original shape is changing into the target shape, and the target shape is generated animation.
The quick surging height variation model is specially：In order to keep the thin of the water surface while surging smooth decline Section divides surging height h for the basal surface height h that surges when startingbaseWith water surface details height hdetailTwo parts：
H=hbase+hdetail
Wherein, by height field data described in Gaussian smoothing, the approximate water surface elevation of the water surface video is extracted, it will be described Height field data subtracts the approximate water surface elevation, obtains water surface details height hdetail；
Line part of surging finally drops to height same as surrounding, and the height finally dropped to is hend,
Wherein w indicates the image of surging in the water surface video,It surges described in expression mask, dilate () table Show expansive working, in each frame of the water surface video, surge the height h for needing to declinediffFor：
hdiff=(h-hend)(1-ratio)
Wherein ratio, which indicates to surge, highly compares the difference in height and original height difference of basal surface of surging after line part declines The calculation of ratio, different drop modes, ratio is different, if declined according to exponential form, the ratio of the i-th frame takes Value is：
Wherein N indicates the totalframes of the water surface video, if it is linear forms, then
In order to make to surge, nature declines, by hdiffAs needing the height declined on skeleton of surging, and the point around skeleton It is adopted according to the grown form surged using radial basis function to control the practical falling head for line each point in the radial direction of surging Use Gaussian function G as radial basis function：
It is the position of the i-th row of wave skeleton.
User by parameter curve and sketches the contours two kinds of input modes of curve in the generation animation, obtains the target Shape.
The shape surged is indicated with the pixel on curve, then original shape is expressed as with target shape：
Wherein,It surges described in expression original shape, piThe point surged described in expression on original shape, xi、yiIt indicates Point piTransverse and longitudinal coordinate；Indicate the target shape, qiIndicate the point in the target shape, x 'i、y′iIndicate its point qiTransverse and longitudinal coordinate, n indicate point number,Indicate integer, put in surge original state and the target shape Number is identical, and coordinate is all integer, and yi=y 'i；
By the original shape of surging to the target shape, shape sequence is obtained WhereinIt is obtained by linear differenceThat is the target shape of the i-th frame, w For the image of surging in the water surface video：
NframeIndicate the total quantity of the frame of deformation process experience.
The cavity that translation leaves of surging is filled up using nonparametric sample texture synthetic method.
The specific implementation process of the method for the present invention is introduced with reference to the accompanying drawings, and Fig. 1 surges dynamic for the offshore of data-driven Draw synthetic method flow chart.
Disclosed by the invention is that the offshore of data-driven is surged animation synthesizing method, in order to for the progress of surging in video Various animation compound needs to extract the correlated characteristic surged from the water scene of sequence of video images first, then by repairing Change feature to drive the more various animation of motion generation of surging, specifically：
1) obtains the height field data that water surface monocular video is rebuild
Height field data is reconstructed from water surface monocular video according to current existing algorithm, algorithm can refer to paper 《Water Surface Modeling from A Single Viewpoint Video》
2) proposes a kind of surge shape extraction and representation method based on skeleton, indicates the shape surged in video image.
21) mask of surging is extracted
Using mathematical morphology relevant rudimentary algorithm, opening operation reconstruction is carried out to water surface video image and closed operation is rebuild, Brightness line part mask outstanding of surging is extracted by extracting local maximum region, Fig. 2 a are the wave for needing to extract feature Unrestrained original image, Fig. 2 b are that the mask extracted is superimposed upon on original image.
22) framework information is extracted
It obtains surging after mask, the mask of surging of two-value is handled using Morphological Thinning Algorithm, is removed Point, then will surge line of the line thinning to the connected not breakpoint of bottom line, the central axes to be surged are surged Framework information, Fig. 2 c are that the framework information extracted is superimposed upon on original image.
3) proposes a kind of quick surging height variation model that details is kept, and is carried out soon to rebuilding surging height out The variation of speed.
In order to keep the details of the water surface while surging smooth decline, surging height h is divided for base of surging when starting Floor height hbaseWith water surface details height hdetailTwo parts：
H=hbase+hdetail
Smooth approximate face is declined, is finally superimposed details again, avoids losing water surface details during the decline Variation and the reduction sense of reality.Wherein, by Gaussian smoothing original height field, the approximate water surface elevation of the water surface is extracted, then original Height field subtracts the water surface details h that the approximate water surface is precipitateddetail。
Line part of surging finally drops to height same with surrounding, and the height finally dropped to is hend,
WhereinIndicate that the mask of surging extracted, dilate () indicate expansive working.To surging, mask carries out After morphological dilations, the region that mask obtains surging around line is subtracted, takes the average value of this region height final as surging The average height of the water surface around the height dropped to, that is, the line of surging of protrusion.
In each frame, surge the height h for needing to declinediffFor：
hdiff=(h-hend)(1-ratio)
H indicates that present frame is surged the original height in line part, h-hendIndicate height of the line compared to water surface protrusion around of surging Degree is poor, is the fluctuation range surged during decline, and ratio is the difference in height and original that basal surface of surging highly is compared after declining The ratio of beginning difference in height, value are 0 to 1.
The calculation of different drop modes, ratio is different.Decline according to exponential form, then the ratio of the i-th frame is answered Value is：
Wherein N is totalframes.If it is linear forms, then
Decline naturally, by h to make to surgediffAs needing the height declined on skeleton of surging, and the point around skeleton The practical falling head for line each point in the radial direction of surging is controlled using radial basis function.According to the grown form surged, adopt Use Gaussian function G as radial basis function.
It is the position of the i-th row of wave skeleton, is the symmetry axis of Gaussian kernel,Indicate radial upper ripple Wave a little arrives the distance of skeletal point.
Fig. 3 b are the frame sequence for the height field that Fig. 3 a are rebuild, and Fig. 3 c are the frame sequence for the height field linear decline that the present invention synthesizes Row, Fig. 3 d are the frame sequence for the height field index decreased that the present invention synthesizes.
4) proposes a kind of target shape approaching method of surging of skeleton driven, generates original shape of surging and is changing into target The animation of shape.
41) target shape of input surge
User is by parameter curve and sketching the contours two kinds of input modes of curve, the target shape for the deformation that obtains surging, such as Fig. 4 a For the conic section of input, Fig. 4 b are the curve sketched the contours.It to be refined using thinning algorithm to sketching the contours curve, effect such as Fig. 4 c It is shown.
42) shape surged in every frame is obtained using interpolation algorithm.
In order to deform, surges original shape and target shape will realize one-to-one correspondence.With the picture on curve Vegetarian refreshments indicates the shape surged, then original shape and target shape are expressed as：
Wherein,Expression is surged original shape, piIt is the point on original shape,Indicate target shape, qiIt is Point in target shape, the coordinate put in shape is all integer, and yi=y 'i。
By original shape of surging to target shape, shape sequence can be obtained WhereinIt is obtained by linear differenceThat is the target shape of the i-th frame.
NframeIndicate the total quantity of the frame of deformation process experience.
43) cavity that translation leaves of surging is filled up.
Fig. 5 a are original figure of surging, and after line translation of surging per frame, height field cavity can be left on home position, such as Shown in Fig. 5 b.Here nonparametric sample texture synthetic method is used, paper is specifically shown in《Texture synthesis by non- parametric sampling》, and optimize its neighborhood similarity metric function, while considering neighborhood similarity and Time Continuous Property, to carry out hole-filling, to generate around line of surging than shallower wave fine crushing, effect is as shown in Figure 5 c.
Fig. 6 is the frame sequence figure that the original shape of surging that the present invention synthesizes is changing into target shape.

Claims (8)

  1. The animation synthesizing method 1. a kind of offshore of data-driven is surged, which is characterized in that including：
    Step 1, the monocular video of water surface video is rebuild, the height field data of the monocular video is obtained, to the water Plane video carries out opening operation reconstruction and is rebuild with closed operation, obtains mask of surging, is covered to described surge by Morphological Thinning Algorithm Film is handled, and framework information of surging is obtained；
    Step 2, according to the mask of surging, the framework information of surging, quick surging height variation model is created, by described Quick surging height variation model and according to the height field data, getting frame sequence chart；
    Step 3, the target shape of user's input surge is obtained, and the original shape of surging in the frame sequence figure is changing into institute Target shape is stated, and the target shape is generated into animation；
    The wherein described quick surging height variation model is specially：In order to keep the thin of the water surface while surging smooth decline Section divides surging height h for the basal surface height h that surges when startingbaseWith water surface details height hdetailTwo parts：
    H=hbase+hdetail
    Wherein, by height field data described in Gaussian smoothing, the approximate water surface elevation of the water surface video is extracted, by the height Field data subtracts the approximate water surface elevation, obtains water surface details height hdetail；
    Line part of surging finally drops to height same as surrounding, and the height finally dropped to is hend,
    Wherein w indicates the image of surging in the water surface video,It surges described in expression mask, dilate () indicates swollen Swollen operation, in each frame of the water surface video, surge the height h for needing to declinediffFor：
    hdiff=(h-hend)(1-ratio)
    Wherein ratio indicate surge line part decline after highly compare surge basal surface difference in height and original height difference ratio The calculation of rate, different drop modes, ratio is different, if declined according to exponential form, the ratio values of the i-th frame For：
    Wherein N indicates the totalframes of the water surface video, if it is linear forms, then
    In order to make to surge, nature declines, by hdiffAs needing the height declined on skeleton of surging, and the point around skeleton uses diameter The practical falling head that line each point in the radial direction of surging is controlled to basic function, according to the grown form surged, using Gauss Function G is as radial basis function：
    It is the position of the i-th row of wave skeleton.
  2. The animation synthesizing method 2. offshore of data-driven as described in claim 1 is surged, which is characterized in that in the step 3 User is by parameter curve and sketches the contours two kinds of input modes of curve, obtains the target shape.
  3. The animation synthesizing method 3. offshore of data-driven as claimed in claim 2 is surged, which is characterized in that with the picture on curve Vegetarian refreshments indicates the shape surged, then original shape is expressed as with target shape：
    Wherein,It surges described in expression original shape, piThe point surged described in expression on original shape, xi、yiIndicate point pi's Transverse and longitudinal coordinate；Indicate the target shape, qiIndicate the point in the target shape, x'i、y'iIndicate the cross of its point qi Ordinate, n indicate that the number of point, N indicate integer, and the original state of surging is identical as the number put in the target shape, Coordinate is all integer, and yi=y'i；
    By the original shape of surging to the target shape, shape sequence is obtained WhereinIt is obtained by linear differenceThat is the target shape of the i-th frame, W is the image of surging in the water surface video：
    NframeIndicate the total quantity of the frame of deformation process experience.
  4. The animation synthesizing method 4. offshore of data-driven as claimed in claim 3 is surged, which is characterized in that adopted using nonparametric Sample texture synthesis method fills up the cavity that translation leaves of surging.
  5. The animation compound system 5. a kind of offshore of data-driven is surged, which is characterized in that including：
    Acquisition is surged framework information module, is rebuild for the monocular video to water surface video, is obtained the monocular video Height field data carries out opening operation reconstruction to the water surface video and is rebuild with closed operation, obtains mask of surging, thin by morphology Change algorithm to handle the mask of surging, obtains framework information of surging；
    Getting frame sequence module creates quick surging height and becomes for mask, the framework information of surging of surging according to Change model, by the quick surging height variation model and according to the height field data, getting frame sequence chart；
    Generation animation, the target shape for obtaining user's input surge, and surging in the frame sequence figure is initial Change in shape generates animation at the target shape, and by the target shape；
    The wherein described quick surging height variation model is specially：In order to keep the thin of the water surface while surging smooth decline Section divides surging height h for the basal surface height h that surges when startingbaseWith water surface details height hdetailTwo parts：
    H=hbase+hdetail
    Wherein, by height field data described in Gaussian smoothing, the approximate water surface elevation of the water surface video is extracted, by the height Field data subtracts the approximate water surface elevation, obtains water surface details height hdetail；
    Line part of surging finally drops to height same as surrounding, and the height finally dropped to is hend,
    Wherein w indicates the image of surging in the water surface video,It surges described in expression mask, dilate () indicates swollen Swollen operation, in each frame of the water surface video, surge the height h for needing to declinediffFor：
    hdiff=(h-hend)(1-ratio)
    Wherein ratio indicate surge line part decline after highly compare surge basal surface difference in height and original height difference ratio The calculation of rate, different drop modes, ratio is different, if declined according to exponential form, the ratio values of the i-th frame For：
    Wherein N indicates the totalframes of the water surface video, if it is linear forms, then
    In order to make to surge, nature declines, by hdiffAs needing the height declined on skeleton of surging, and the point around skeleton uses diameter The practical falling head that line each point in the radial direction of surging is controlled to basic function, according to the grown form surged, using Gauss Function G is as radial basis function：
    It is the position of the i-th row of wave skeleton.
  6. The animation compound system 6. the offshore of data-driven as claimed in claim 5 is surged, which is characterized in that the generation animation User by parameter curve and sketches the contours two kinds of input modes of curve in module, obtains the target shape.
  7. The animation compound system 7. the offshore of data-driven as claimed in claim 6 is surged, which is characterized in that with the picture on curve Vegetarian refreshments indicates the shape surged, then original shape is expressed as with target shape：
    Wherein,It surges described in expression original shape, piThe point surged described in expression on original shape, xi、yiIndicate point pi's Transverse and longitudinal coordinate；Indicate the target shape, qiIndicate the point in the target shape, x'i、y'iIndicate its point qiCross Ordinate, n indicate that the number of point, N indicate integer, and the original state of surging is identical as the number put in the target shape, Coordinate is all integer, and yi=y'i；
    By the original shape of surging to the target shape, shape sequence is obtained WhereinIt is obtained by linear differenceThat is the target shape of the i-th frame, W is the image of surging in the water surface video：
    NframeIndicate the total quantity of the frame of deformation process experience.
  8. The animation compound system 8. the offshore of data-driven as claimed in claim 7 is surged, which is characterized in that adopted using nonparametric Sample texture synthesis method fills up the cavity that translation leaves of surging.
CN201610115647.8A 2016-03-01 2016-03-01 A kind of offshore of data-driven is surged animation synthesizing method and system Active CN105608727B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610115647.8A CN105608727B (en) 2016-03-01 2016-03-01 A kind of offshore of data-driven is surged animation synthesizing method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610115647.8A CN105608727B (en) 2016-03-01 2016-03-01 A kind of offshore of data-driven is surged animation synthesizing method and system

Publications (2)

Publication Number Publication Date
CN105608727A CN105608727A (en) 2016-05-25
CN105608727B true CN105608727B (en) 2018-08-10

Family

ID=55988642

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610115647.8A Active CN105608727B (en) 2016-03-01 2016-03-01 A kind of offshore of data-driven is surged animation synthesizing method and system

Country Status (1)

Country Link
CN (1) CN105608727B (en)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106023287B (en) * 2016-05-31 2019-06-18 中国科学院计算技术研究所 A kind of the interactive three-dimensional animation synthesizing method and system of data-driven
CN106362398B (en) * 2016-10-14 2020-02-07 网易（杭州）网络有限公司 Method and device for determining position of camera lens
CN107085629B (en) * 2017-03-28 2020-05-12 华东师范大学 Fluid simulation method based on coupling of video reconstruction and Euler model
CN109840935A (en) * 2017-12-12 2019-06-04 中国科学院计算技术研究所 Wave method for reconstructing and system based on depth acquisition equipment
CN110084872B (en) * 2019-03-25 2020-12-25 中国科学院计算技术研究所 Data-driven smoke animation synthesis method and system
CN110717946A (en) * 2019-09-25 2020-01-21 浙江万胜智能科技股份有限公司 Method for screening flame target from video image

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101673410A (en) * 2008-09-12 2010-03-17 中国科学院计算技术研究所 Vector building drawing based method for reconstructing three-dimensional model
CN102831628A (en) * 2012-07-02 2012-12-19 中国科学院计算技术研究所 Method and equipment for updating Voronoi map of dynamic scene
CN103021010A (en) * 2012-12-01 2013-04-03 王爽 Wave effect realtime rendering method based on graphics processing unit (GPU)
CN103279971A (en) * 2013-06-06 2013-09-04 中国科学院计算技术研究所 Method and system for redirection of sports style

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI353559B (en) * 2006-12-18 2011-12-01 Inst Information Industry Apparatus, method, application program and compute

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101673410A (en) * 2008-09-12 2010-03-17 中国科学院计算技术研究所 Vector building drawing based method for reconstructing three-dimensional model
CN102831628A (en) * 2012-07-02 2012-12-19 中国科学院计算技术研究所 Method and equipment for updating Voronoi map of dynamic scene
CN103021010A (en) * 2012-12-01 2013-04-03 王爽 Wave effect realtime rendering method based on graphics processing unit (GPU)
CN103279971A (en) * 2013-06-06 2013-09-04 中国科学院计算技术研究所 Method and system for redirection of sports style

Also Published As

Publication number Publication date
CN105608727A (en) 2016-05-25

Similar Documents

Publication Publication Date Title
CN105608727B (en) A kind of offshore of data-driven is surged animation synthesizing method and system
Wang et al. Nerf-art: Text-driven neural radiance fields stylization
Kato et al. Neural 3d mesh renderer
US10846903B2 (en) Single shot capture to animated VR avatar
Chai et al. Dynamic hair manipulation in images and videos
Pyun et al. An example-based approach for facial expression cloning
US8902232B2 (en) Facial performance synthesis using deformation driven polynomial displacement maps
CN103854306A (en) High-reality dynamic expression modeling method
Deussen et al. Weighted linde-buzo-gray stippling
WO2021140510A2 (en) Large-scale generation of photorealistic 3d models
Eyiyurekli et al. Interactive free-form level-set surface-editing operators
Zhu et al. Hifa: High-fidelity text-to-3d generation with advanced diffusion guidance
CN110335275B (en) Fluid surface space-time vectorization method based on three-variable double harmonic and B spline
Bharadwaj et al. Flare: Fast learning of animatable and relightable mesh avatars
Zhang et al. Avatarstudio: High-fidelity and animatable 3d avatar creation from text
CN105761295B (en) A kind of the water surface method for reconstructing and system of data-driven
Meyer et al. PEGASUS: Physically Enhanced Gaussian Splatting Simulation System for 6DOF Object Pose Dataset Generation
Kozhekin et al. An approach to surface retouching and mesh smoothing
Chaudhry et al. Modelling and Simulation of Lily flowers using PDE Surfaces
CN117315211B (en) Digital human synthesis and model training method, device, equipment and storage medium thereof
Liu et al. A feature-based approach for individualized human head modeling
Li et al. 3D Personalized Face Modeling Based on KINECT2
Shengming Research of 3D face reconstruction based on one image
Yang et al. A learning-based system for generating exaggerative caricature from face images with expression
US8896607B1 (en) Inverse kinematics for rigged deformable characters

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