CN109584327A - Face aging analogy method, device and equipment - Google Patents
Face aging analogy method, device and equipment Download PDFInfo
- Publication number
- CN109584327A CN109584327A CN201811324007.3A CN201811324007A CN109584327A CN 109584327 A CN109584327 A CN 109584327A CN 201811324007 A CN201811324007 A CN 201811324007A CN 109584327 A CN109584327 A CN 109584327A
- Authority
- CN
- China
- Prior art keywords
- aging
- pixel
- image
- facial image
- face
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/001—Texturing; Colouring; Generation of texture or colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/40—Filling a planar surface by adding surface attributes, e.g. colour or texture
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to face aging analogy method, device and equipment, method comprising steps of to mutual face alignment simulation facial image triangulation to be aging obtain first kind triangular mesh, the second class triangular mesh is obtained to target burn facial image triangulation；Obtain the triangular coordinate of the first kind pixel in first kind triangular mesh；Each first kind pixel corresponding second class triangular facet in the second class triangular mesh, and the triangular coordinate system based on corresponding triangular coordinate and corresponding second class triangular facet are obtained, the new cartesian coordinate of each first kind pixel is obtained；Acquire the pixel value of each new cartesian coordinate corresponding second class pixel on target burn facial image；According to pixel value of the first kind pixel on simulation facial image to be aging and the sum of the pixel value of corresponding second class pixel, aging image to be output is obtained.The present invention improves the timeliness of Aging simulation image fidelity to be output and image simulation.
Description
Technical field
The present invention relates to technical field of image processing, in particular to face aging analogy method, device and equipment.
Background technique
In recent years, computer graphic image technology is to develop more swift and violent emerging technology, using this technology to people
The aging course of face carries out the project that quantitative prediction is increasingly becoming hot research.For the Aging simulation of face appearance, tradition
The common method of face aging simulation will be substantially used in technology to be divided into: the ageing process simulation based on face prototype；It is based on
The face aging of reconstruct synthesizes；Face aging simulation based on study.
During realization, inventor has found that at least there are the following problems in traditional technology:
There is program redundancy to a certain extent in the face aging simulation algorithm part of traditional technology, some algorithm need according to
Mass data is relied to carry out model training etc..Its process is complex, the more demanding and used time is partially long, be easy to cause to real-time and
Fidelity impacts.
Summary of the invention
It is an object of the invention to be directed to the deficiency of traditional technology, provide a kind of face aging analogy method, device and
Equipment.
Embodiment according to the present invention, the present invention provides a kind of face aging analogy methods, comprising the following steps:
First kind triangular mesh is obtained to the simulation facial image triangulation to be aging of mutual face alignment, to target
Aging facial image triangulation obtains the second class triangular mesh；
The pixel for the simulation facial image to be aging for including using first kind triangular mesh is as first kind pixel, base
In it is to be aging simulation facial image each first kind pixel former cartesian coordinate and first kind pixel in the first kind three
In hexagonal lattice, three vertex of the first kind triangular facet at place constitute triangular coordinate system, obtain each first kind pixel
Triangular coordinate in the first kind triangular facet of place；
Obtain corresponding second class in the second class triangular mesh of the first kind triangular facet where each first kind pixel
Triangular facet, and sat based on the triangle that three vertex are constituted in corresponding triangular coordinate and corresponding second class triangular facet
Mark system, obtains new cartesian coordinate of each first kind pixel on target burn facial image；
The pixel for the target burn image for including using the second class triangular mesh is as the second class pixel, according to each new
Cartesian coordinate acquires the pixel value of each new cartesian coordinate corresponding second class pixel on target burn facial image；
According to pixel value of the first kind pixel on simulation facial image to be aging and corresponding second class pixel
The sum of pixel value obtains picture of the first kind pixel on the aging image to be output determined based on simulation facial image to be aging
Element value, obtains aging image to be output.
In one of the embodiments, according to first kind pixel it is to be aging simulation facial image on pixel value with it is right
The sum of the pixel value for the second class pixel answered, obtain first kind pixel based on it is to be aging simulation facial image determine to
The pixel value on aging image is exported, the step of obtaining aging image to be output includes:
It will be determined according to the former cartesian coordinate of the corresponding first kind pixel of new cartesian coordinate, and according to texture mapping
Aging image to be output on pixel value, be mapped to the corresponding first kind pixel of simulation facial image to be aging, obtain to
Export aging image.
The method for obtaining aging image to be output in one of the embodiments, includes:
Operation is carried out based on following formula:
SRCnew(i, j)=SRC (i, j) (1-t)+REF (inew,jnew)·t(1)
Wherein, SRCnew(i, j) is the aging figure to be output that first kind pixel is determined based on simulation facial image to be aging
As upper pixel value；SRC (i, j) is pixel value of the first kind pixel on simulation facial image to be aging；REF(inew,
jnew) be new cartesian coordinate corresponding second class pixel on target burn facial image pixel value；T is preset use
In the adjustment constant of adjustment degree of aging；(i, j) is described to be output old to be determined based on the simulation facial image to be aging
Change the former cartesian coordinate of the corresponding first kind pixel on image；(inew, jnew) indicate the first kind pixel
The new cartesian coordinate；
Using operation result as first kind pixel in the output figure to be aging determined based on simulation facial image to be aging
As upper pixel value, output image to be aging is obtained.
Carry out triangulation in the simulation facial image to be aging that is aligned to mutual face in one of the embodiments, and
First kind triangular mesh is obtained, triangulation is carried out to target burn facial image and obtains the second class triangular mesh, is wrapped
It includes:
It demarcates in the first kind predetermined patterns characteristic point in simulation facial image to be aging and target burn facial image
The second class predetermined patterns characteristic point；First kind predetermined patterns characteristic point is that mutual face does not carry out human face characteristic point mark when being aligned
The pixel at fixed, simulation facial image to be aging face position；Second class predetermined patterns characteristic point is the alignment of mutual face
When do not carry out the pixel at human face characteristic point calibration, target burn image face position；
In each human face characteristic point according to each first kind predetermined patterns characteristic point and simulation facial image to be aging, determine
Simulation facial image to be aging human face region carry out triangulation, obtain first kind triangular mesh；According to each second
Each human face characteristic point of class predetermined patterns characteristic point and target burn facial image, the people of determining target burn facial image
Face region carries out triangulation, obtains the second class triangular net.
The first kind predetermined patterns characteristic point to be aging simulated in facial image is being demarcated in one of the embodiments,
And before the second class predetermined patterns characteristic point in target burn facial image, further includes:
Each one of each human face characteristic point for simulating facial image and target burn facial image to be aging are calibrated respectively
Face characteristic point, and simulation facial image to be aging is subjected to face with target burn image and is aligned.
Carry out triangulation in the simulation facial image to be aging that is aligned to mutual face in one of the embodiments, and
Obtain first kind triangular mesh, to target burn facial image carry out triangulation and obtain the second class triangular mesh it
Afterwards, further includes:
Each pixel for traversing simulation facial image to be aging, identifies each first kind three in first kind triangular mesh
All first kind pixels that edged surface separately includes.
In one of the embodiments, obtain first kind pixel based on it is to be aging simulation facial image determine to
The pixel value on aging image is exported, after obtaining aging image to be output, further includes:
According to the second class triangular mesh, black white binarization is carried out to target burn facial image, obtains target burn figure
As the black white binarization image in the second class triangular mesh；
According to black white binarization image, graph cut is carried out to aging image to be output, obtains final output aging mould
Quasi- image.
Embodiment according to the present invention additionally provides a kind of face aging simulator, comprising:
Triangular mesh module is obtained, the simulation facial image to be aging for being aligned to mutual face carries out triangulation
And first kind triangular mesh is obtained, triangulation is carried out to target burn facial image and obtains the second class triangular mesh；
Triangular coordinate conversion module, the picture of the simulation facial image to be aging for including by first kind triangular mesh
Element value be used as first kind pixel, based on it is to be aging simulation facial image each first kind pixel former cartesian coordinate and
First kind pixel is in first kind triangular mesh, the triangular coordinate that three vertex of the first kind triangular facet at place are constituted
System, the triangular coordinate where obtaining each first kind pixel in first kind triangular facet；
New cartesian coordinate converts module, for obtaining the first kind triangular facet where each first kind pixel in the second class
Corresponding second class triangular facet in triangular mesh, and it is based on corresponding triangular coordinate and corresponding second class triangular facet
In three vertex constitute triangular coordinate system, obtain new Descartes of each first kind pixel on target burn facial image
Coordinate；
Aging data acquisition module, the pixel of the target burn image for including using the second class triangular mesh as
It is corresponding on target burn facial image to acquire each new cartesian coordinate according to each new cartesian coordinate for second class pixel
The pixel value of second class pixel；
Aging simulation module, for according to first kind pixel it is to be aging simulation facial image on pixel value with it is corresponding
The second class pixel the sum of pixel value, obtain first kind pixel based on it is to be aging simulation facial image determine to defeated
Pixel value on aging image out obtains aging image to be output.
Embodiment according to the present invention, the present invention also provides a kind of face aging analog machine, including memory and
Processor, memory are stored with computer program, which is characterized in that processor realizes face of the present invention when executing computer program
Each step of aging analogue method.
Embodiment according to the present invention, the present invention also provides a kind of computer storage mediums, are stored thereon with calculating
Machine program realizes each step of the present inventor's face aging analogue method when computer program is executed by processor.
A technical solution in above-mentioned technical proposal is had the following advantages and beneficial effects:
Face aging analogy method, device and equipment of the invention, by the mould to be aging for obtaining mutual face alignment
The first kind triangular mesh and the second class triangular mesh that anthropomorphic face image and target burn facial image triangulation obtain,
It is accurate to determine and obtain the region of simulation to be aging.And then it is convenient for according to first kind triangular mesh, and be based on simulation to be aging
The former cartesian coordinate of each first kind pixel of facial image and the triangular facet in first kind triangular mesh converse each
The triangular coordinate of first kind pixel, from there through the second class triangular facet and each triangle in the second class triangular mesh
Shape coordinate, inversion swap out the pixel value of each first kind pixel corresponding pixel on target burn facial image, and then
To aging image to be output.The present invention constructs the triangular mesh of simulation facial image and target burn facial image to be aging,
Each first kind pixel correspondence of simulation facial image to be aging is swapped out in target burn face figure based on triangular coordinate system inversion
As upper new cartesian coordinate.And then the pixel value of each pixel of target burn image is accurately acquired, in conjunction with simulation to be aging
The pixel value of each first kind pixel of facial image treats Aging simulation facial image and carries out Aging simulation.Aging of the invention
Simulation process is relatively simple, and the topological deformation process easy to accomplish to image improves Aging simulation image fidelity to be output,
And the timeliness of image simulation, more natural face aging image can be effectively synthesized.
Detailed description of the invention
Fig. 1 shows the flow diagram of the face aging analogy method of an embodiment of the invention offer；
Fig. 2 shows the present inventor's face aging analogue methods to obtain first kind triangular mesh and the second class triangular mesh
Flow diagram；
Fig. 3 show the present inventor's face aging analogue method not mutual face alignment simulation facial image to be aging and
The schematic diagram of target burn image；
Fig. 4 shows the effect diagram that face is aligned in the present inventor's face aging analogue method；
Fig. 5 shows the simulation facial image to be aging that mutual face alignment is demarcated in the present inventor's face aging analogue method
First kind predetermined patterns characteristic point schematic diagram；
Fig. 6 shows the schematic diagram that triangulation is carried out in the present inventor's face aging analogue method；
Fig. 7 shows the schematic diagram carried out after texture mapping in the present inventor's face aging analogue method；
Fig. 8 shows the black white binarization image schematic diagram in the present inventor's face aging analogue method；
Fig. 9 shows output Aging simulation image schematic diagram final in the present inventor's face aging analogue method；
Figure 10 shows the flow diagram that the present inventor's face aging analogue method carries out black white binarization；
The detailed process for the face aging analogy method that the another embodiment that Figure 11 shows face of the present invention provides
Schematic diagram；
Figure 12 shows the structural block diagram for the face aging simulator that one embodiment of the present of invention mode provides；
Figure 13 shows the structural block diagram for the face aging analog machine that one embodiment of the present of invention mode provides.
Specific embodiment party examination
To make the objectives, technical solutions, and advantages of the present invention more clearly, referring to the drawings in conjunction with the embodiments, to this
Invention is further wanted to illustrate.
Ageing process analogy method based on face prototype, has constructed the face prototype of different age group, and by this
Shape and the difference of texture are added into facial image by caricature technology between a little face prototypes, thus directly linear by people
Face aging is to target age section.This method not only allows for face shape and changes at any time, while considering texture information
Variation.Secondly, to inspire the method for synthesizing aging effect by illumination, and rendering to obtain with this there are also another mode
The aging face of any target age.Since this method has carried out using optical flow method the adjustment and processing of illumination to average face,
So obtained average face has the shape being more clear and texture.In the aging synthesis phase of face, using optical flow method to people
Face carries out deformation process, compared to the face aging synthetic method based on prototype is realized by addition and subtraction operation, obtains preferable
Aging simulation effect.
Another side synthesizes the face aging based on reconstruct, is the facial image of a given input, in target age
Several aging face bases are selected in section, and are weighted combination to it, to synthesize input picture aging to target age section
Facial image.If arbitrarily selecting a collection of facial image as aging face base in all library images of target age section,
And it is subjected to linear combination using identical weighting coefficient, then these selected aging face bases are equivalent to be average
Face.Image super-resolution method of some investigators based on coefficient expression study, proposes a kind of based on the old of matrix decomposition
Change analogy method.It needs to train a super-resolution algorithms on the library image in different age group when its core concept.It is first
First, it needs to carry out down-sampling to the facial image of input, then utilizes the super-resolution algorithms pair of the target age section trained
It carries out super-resolution rebuilding, the facial image of the instant aging of image after reconstruction.Face aging simulation based on study, for
The facial image of one input, this method are used to the difference for predicting to be displaced between input face and target age section face, pass through
This displacement difference come to input face deform, the instant aging facial image of the facial image obtained after deformation.Researcher
The characteristic point position of face in different age group is trained using SVM (support vector machines).The facial image inputted for one,
It can use the position that trained model prediction goes out characteristic point when inputting face aging to other age brackets, thus with image
Deformation technology will input face aging deformation values target age section.It is asked existing for the real-time and fidelity for the above method
Topic, it is lower the invention proposes scheme complexity and more natural face aging image can be synthesized.
Referring to Fig. 1, in one embodiment, the present invention provides a kind of face aging analogy methods, comprising the following steps:
Step S110: triangulation is carried out to the simulation facial image to be aging of mutual face alignment and obtains first kind triangle
Shape net carries out triangulation to target burn facial image and obtains the second class triangular mesh.
Specifically, face be aligned to will it is to be aging simulation facial image in face and target burn facial image in
The different shape of face, the process that the two is normalized, by one of face shape as much as possible close to another face
Shape.Facial feature points detection is carried out to the face in two face figures respectively, determines respective human face region range, into
And simulation facial image to be aging can be subjected to face with target burn facial image by way of affine transformation and be aligned.Into one
Step ground, when first kind triangle gridding is that simulation facial image to be aging carries out triangulation, in simulation facial image to be aging
Pixel within the scope of determining human face region is connected as each triangular facet, and the polygonal mesh being made of each triangular facet.Second
When class triangular mesh is that target burn image carries out triangulation, within the scope of the human face region that target burn image determines
Pixel is connected as each triangular facet, and the polygonal mesh being made of each triangular facet.Wherein, triangle subdivision can be
Delaunay Triangulation.
The present invention can obtain uniform and smooth face by obtaining first kind triangular subdibison and the second class triangular subdibison
Piece grid, so that needing on simulation face figure to be aging and target burn image can the company of foundation between the pixel of calculation processing
Relationship is connect, help the topology deformation for realizing image in Aging simulation, and be less prone to loss of significance and algorithmic error etc..
Step S120: the pixel for the simulation facial image to be aging for including using first kind triangular mesh is as the first kind
Pixel, the former cartesian coordinate and first kind pixel of each first kind pixel based on simulation facial image to be aging exist
In first kind triangular mesh, three vertex of the first kind triangular facet at place constitute triangular coordinate system, obtain each first
Triangular coordinate where class pixel in first kind triangular facet.
Specifically, in cartesian coordinate system, three tops of each first kind triangular facet in first kind triangular mesh
Point separately constitutes a triangle, and constitutes a triangular coordinate system.By first kind pixel in first kind triangular mesh
The triangle of three vertex composition of the first kind triangular facet at middle place, as the core of triangular coordinate system, i.e. first kind pixel
The first kind pixel is calculated in the corresponding first kind centered on the triangle that the first kind triangular facet at place forms in point
Triangular coordinate system in triangular facet.
Specifically, any one first kind triangular facet in first kind triangular mesh is taken, three of them vertex is respectively A, B
And C, three vertex form a triangle, any one first kind pixel P in the first kind triangular facet, note point P is taken to exist
Coordinate in the triangular coordinate system of vertex A, B and C composition is (xa, xb).The triangle of point P can be then obtained based on following formula
Coordinate:
T=MC
Wherein, T=(xa xb)T；C=(px py 1)T；
kt=(Ay-Cy)(Bx-Cx)-(Ax-Cx)(By-Cy)。
The former cartesian coordinate of first kind pixel P is (Px, Py), the former cartesian coordinate of point A is (Ax, Ay), point B's
Former cartesian coordinate is (Bx, By), the former cartesian coordinate of point C is (Cx, Cy)。
The triangular coordinate system and each first that the present invention is made up of first kind triangular facet each in first kind triangular mesh
Class pixel accurately obtains each first kind pixel on simulation facial image to be aging, in target when facilitating Aging simulation
The pixel value for the pixel that should be mapped on aging facial image improves the efficiency of traversal processing first kind pixel.
Step S130: the first kind triangular facet where obtaining each first kind pixel is corresponding in the second class triangular mesh
The second class triangular facet, and based on three tops in the corresponding triangular coordinate and the corresponding second class triangular facet
The triangular coordinate system that point is constituted, obtains new cartesian coordinate of each first kind pixel on target burn facial image.
Specifically, three vertex of each second class triangular facet in the second class triangular mesh separately constitute a triangle
Shape, and constitute a triangular coordinate system.Based on above-mentioned steps S110, first kind triangular facet in first kind triangular mesh and
The second class triangular facet in second class triangular mesh is one-to-one relationship, can obtain each first kind pixel institute as a result,
First kind triangular facet in the second class triangular mesh corresponding second class triangular facet.
Specifically, a first kind pixel P in first kind triangular mesh and first kind pixel P institute are taken
The corresponding second class triangular mesh of first kind triangular facet in the second class triangular facet.The second class triangular facet is taken, remembers it
Three vertex are respectively A, B and C, and three vertex form a triangle.The new Descartes of point P can be then obtained based on following formula
Coordinate:
P=OW
Wherein, P=(Px Py)；W=(xa xb 1-xa-xb)。
The former cartesian coordinate of first kind pixel P is (Px, Py), the former cartesian coordinate of point A is (Ax, Ay), point B's
Former cartesian coordinate is (Bx, By), the former cartesian coordinate of point C is (Cx, Cy), the triangular coordinate of point P is (xa, xb)。
The present invention is based on simulation facial images and target burn facial image to be aging when face alignment to distinguish triangulation
Afterwards, each second class triangular facet corresponds in each first kind triangular facet and the second class triangular mesh in first kind triangular mesh
Relationship, possibility of reversal swaps out new cartesian coordinate of each first kind pixel on target burn facial image.The present invention helps
When treating Aging simulation facial image progress face aging, accurately to the face position texture mapping of aging and aging tune
It is whole.
Step S140: the pixel for the target burn image for including using the second class triangular mesh is as the second class pixel
Point acquires each new cartesian coordinate corresponding second class pixel on target burn facial image according to each new cartesian coordinate
The pixel value of point.
Specifically, new cartesian coordinate is the first kind triangle of simulation facial image to be aging in Aging simulation
Each first kind pixel, the corresponding cartesian coordinate in the second class triangular mesh of target burn facial image in grid,
And then obtain the pixel value of the second class pixel of the cartesian coordinate.
The present invention can be by the new cartesian coordinate of each first kind pixel of inverse transformation, so as in target burn facial image
Upper each first kind pixel of search carries out the pixel value mapped needed for aging adjustment when simulating aging, and it is old to help to improve face
Change the fidelity and operation efficiency of simulation.
Step S150: according to first kind pixel in the pixel value to be aging simulated on facial image and corresponding second class
The sum of pixel value of pixel obtains first kind pixel in the aging figure to be output determined based on simulation facial image to be aging
As upper pixel value, aging image to be output is obtained.
Specifically, by pixel value of the first kind pixel on simulation facial image to be aging and corresponding second class picture
The pixel value of vegetarian refreshments merges, and prevents aging image fault to be output.
The present invention is based on triangular coordinate system, first kind triangular mesh and the second class triangular mesh is established, pass through
Triangular coordinate coordinate system inversion renews cartesian coordinate, and then can be in each first kind picture for treating Aging simulation facial image
When vegetarian refreshments carries out aging adjustment, the second class pixel can be accurately searched on the corresponding coordinate of target burn facial image
Pixel value, while can guarantee the continuity during Aging simulation, prevent the appearance aging of aging image to be output incomplete
Situation.
Face aging analogy method of the invention, by the simulation facial image and mesh to be aging that obtain mutual face alignment
The first kind triangular mesh and the second class triangular mesh that mark aging facial image triangulation obtains are accurate to determine and obtain
Simulated domain to be aging.Each the and then convenient for according to first kind triangular mesh, and based on simulation facial image to be aging
The former cartesian coordinate of a kind of pixel and the triangular facet in first kind triangular mesh, converse each first kind pixel
Triangular coordinate, from there through in the second class triangular mesh the second class triangular facet and each triangular coordinate, inversion swap out
The pixel value of each first kind pixel corresponding pixel on target burn facial image, and then obtain aging figure to be output
Picture.The present invention constructs the triangular mesh of simulation facial image and target burn facial image to be aging, is based on triangular coordinate
It is that inversion swaps out the corresponding new Descartes on target burn facial image of each first kind pixel of simulation facial image to be aging
Coordinate.And then the pixel value of each pixel of target burn image is accurately acquired, in conjunction with simulation facial image each first to be aging
The pixel value of class pixel treats Aging simulation facial image and carries out Aging simulation.Aging simulation process of the invention is more simple
Single, the topological deformation process easy to accomplish to image improves Aging simulation image fidelity to be output and image simulation
Timeliness can be effectively synthesized more natural face aging image.
In a specific embodiment, according to first kind pixel it is to be aging simulation facial image on pixel value with
The sum of the pixel value of corresponding second class pixel, obtain first kind pixel based on it is to be aging simulation facial image to defeated
Pixel value on aging image out, the step of obtaining aging image to be output include:
Step S4: it according to the former cartesian coordinate of the corresponding first kind pixel of new cartesian coordinate, and is reflected according to texture
It penetrates the pixel value on determining aging image to be output, is mapped to the corresponding first kind pixel of simulation facial image to be aging
Point obtains aging image to be output.
Specifically, texture mapping is that the texture pixel on target burn facial image is mapped to simulation face to be aging
The process of pixel in image.Specifically, new cartesian coordinate corresponding first is searched on simulation facial image to be aging
The former cartesian coordinate of class pixel, according to texture mapping by the pixel value on determining aging image to be output, be mapped to
On the corresponding first kind pixel of Aging simulation facial image, and then obtain aging image to be output.
The present invention is based on the former cartesian coordinate of texture mapping and first kind pixel is corresponding with new cartesian coordinate
The precision and operation efficiency of Aging simulation can be improved in relationship.
In a specific embodiment, the method for obtaining aging image to be output includes:
Operation is carried out based on following formula:
SRCnew(i, j)=SRC (i, j) (1-t)+REF (inew,jnew)·t(1)
Wherein, SRCnew(i, j) is the aging figure to be output that first kind pixel is determined based on simulation facial image to be aging
As upper pixel value；SRC (i, j) is pixel value of the first kind pixel on simulation facial image to be aging；REF(inew,
jnew) be new cartesian coordinate corresponding second class pixel on target burn facial image pixel value；T is preset use
In the adjustment constant of adjustment degree of aging.
Using operation result as first kind pixel in the output figure to be aging determined based on simulation facial image to be aging
As upper pixel value, output image to be aging is obtained.
The present invention is the distortion level for reducing aging image to be output, and first kind pixel is added in simulation face to be aging
Pixel value on image, and will be used to adjust the adjustment constant of degree of aging as coefficient, and then by the operation result of above formula
As the pixel value on output image to be aging of the first kind pixel based on simulation facial image to be aging.The present invention is by each
The pixel value of a kind of pixel converts to complete to treat Aging simulation facial image and carry out Aging simulation, obtains to be aging defeated
Image out improves the efficiency and precision of Aging simulation, so that output image to be aging is close to real human face.
Referring to fig. 2, in conjunction with Fig. 3, Fig. 4, Fig. 5 and Fig. 6, in a specific embodiment, what is be aligned to mutual face
Simulation facial image to be aging carries out triangulation and obtains first kind triangular mesh, carries out three to target burn facial image
Angle subdivision and obtain the second class triangular mesh, comprising:
Step S210: first kind predetermined patterns characteristic point and target burn in calibration simulation facial image to be aging
The second class predetermined patterns characteristic point in facial image；First kind predetermined patterns characteristic point is that mutual face does not carry out people when being aligned
The pixel at the face position of simulation facial image face characteristic point calibration, to be aging；Second class predetermined patterns characteristic point is mutual
Phase face does not carry out the pixel at human face characteristic point calibration, target burn image face position when being aligned.
Specifically, target burn facial image can be the image of any one target burn age bracket aging character.
The present invention, can be by simulation face to be aging by calibrating first kind predetermined patterns characteristic point and the second class predetermined patterns characteristic point
The Aging simulation region of image is defined, and guarantees that face aging region is complete, prevents from omitting Aging simulation position, to be aging defeated
The problem of texture of image is uneven out and the discontinuous or distortion of image fracture.
Step S220: special in each face according to each first kind predetermined patterns characteristic point and simulation facial image to be aging
Point is levied, the human face region of the simulation facial image to be aging determined carries out triangulation, obtains first kind triangular mesh；In root
According to each second class predetermined patterns characteristic point and each human face characteristic point of target burn facial image, determining target burn face
The human face region of image carries out triangulation, obtains the second class triangular net.
Specifically, figure b is face alignment as shown in figure 4, figure a is the simulation facial image to be aging of face alignment
Target burn facial image.As shown in figure 5, the small ash point in Fig. 5 is each first kind predetermined patterns characteristic point in forehead position, wherein
First kind predetermined patterns characteristic point can with but be not limited to the characteristic point at forehead position, can be waited for always for the figure a in as shown in Figure 3
Change simulation facial image with it is as shown in Figure 3 in the mutual face of figure b target burn analog image be aligned when, do not carry out face characteristic
The characteristic point at the face position of point calibration.Further, in the human face region and target for determining simulation facial image to be aging
After the human face region of aging facial image, triangulation is carried out respectively, obtains effect picture as shown in FIG. 6.
Present invention determine that the human face region of Aging simulation out, accurately obtains each first kind pixel and exists when helping Aging simulation
The pixel value for the pixel that should be mapped on target burn facial image, and improve the fidelity and operation of face aging simulation
Efficiency.
Referring to Fig. 5, in a specific embodiment, portion is preset in the first kind demarcated in simulation facial image to be aging
Before the second class predetermined patterns characteristic point in position characteristic point and target burn facial image, further includes:
Step S8: each human face characteristic point and target burn face figure of simulation facial image to be aging are calibrated respectively
Each human face characteristic point of picture, and simulation facial image to be aging is subjected to face with target burn image and is aligned.
Specifically, being demarcated by human face characteristic point, can detect and identifying the shape of face and face area in image
Domain.Wherein, in central point, the prenasale and two of simulation facial image to be aging and two eyeballs of target burn facial image
When a corners of the mouth point alignment, it can be considered that two faces have been aligned.As shown in figure 5, Fig. 5 is target burn facial image to be aging
Simulate the effect picture that facial image carries out face alignment.
The present invention is demarcated by human face characteristic point and face is aligned, and helps to treat Aging simulation facial image progress people
When face aging, accurately the face position texture mapping of aging and aging are adjusted.
Referring to Fig. 6, in a specific embodiment, carried out in the simulation facial image to be aging being aligned to mutual face
Triangulation and obtain first kind triangular mesh, to target burn facial image carry out triangulation and obtain the second class triangle
After shape grid, further includes:
Step S12: each pixel of the traversal simulation facial image to be aging is identified in the first kind triangle
All first kind pixels that each first kind triangular facet separately includes in grid.
Specifically, first where may recognize that first kind pixel in first kind triangular mesh by the step
Class triangular facet, and exclude the pixel for the simulation facial image to be aging being not included in first kind triangular mesh.Specifically,
In image as shown in FIG. 6, the pixel and a first kind triangular facet of a simulation facial image to be aging, meter are given
Calculate triangular coordinate (x of this under the triangular coordinate system using the triangle as corea, xb), enable xc=1-xa-xbIf xa,
xb, xcIn at least one be negative, then point in triangular exterior；If xa, xb, xcIt is nonnegative number and wherein at least one is
0, then the pixel is in triangle edges；If xa, xb, xcAll positive numbers, then point is in triangle interior.And then may recognize that
Whether the pixel of Aging simulation facial image is in first kind triangular mesh, and each in first kind triangular mesh
Which first kind triangular facet a kind of pixel is included in.
Face aging analogy method of the invention, the simulation people to be aging that can will be not included in first kind triangular mesh
The pixel of face image excludes, and then identifies the first kind triangular facet that there emerged a where each first kind pixel.The present invention can convert
The triangular coordinate of first kind pixel out facilitates the precision for improving texture mapping when carrying out Aging simulation, simplifies operation
Process and operation efficiency.
Referring to Figure 10, in conjunction with Fig. 4, Fig. 7, Fig. 8 and Fig. 9, in a specific embodiment, first kind pixel is being obtained
Based on it is to be aging simulation facial image determine aging image to be output on pixel value, obtain aging image to be output it
Afterwards, further includes:
Step S310: according to the second class triangular mesh, black white binarization is carried out to target burn facial image, obtains mesh
Mark black white binarization image of the aging image in the second class triangular mesh.
Specifically, can obtain black white binarization threshold value carries out black white binarization, the second class triangular mesh will be in
The second class pixel in second class triangular facet is set to white, black is otherwise set to, as a result, as shown in figure 8, obtaining target burn
Black white binarization image of the image in the second class triangular mesh.
Step S320: according to black white binarization image, carrying out graph cut to aging image to be output, obtains final defeated
Aging simulation image out.
Specifically, being obtained shown in the gradient fields M and Fig. 4 (a) within the scope of the human face region as shown in Fig. 4 (b) respectively
The gradient fields N of image within the scope of human face region further obtains gradient as shown in Figure 7 according to gradient fields M and gradient fields N
Field Z, Fig. 7 are that the schematic diagram after texture mapping, aging image to be output as to be fused are carried out in people's face aging analogue method.
Further, divergence is obtained by gradient fields Z as shown in Figure 7.Specifically, the pixel of fusion results is obtained based on following formula
Value:
Ax=b
Wherein, A is the coefficient matrix constructed according to the image size of Fig. 7, and b is divergence, and x is the pixel value of fusion results.
Final output Aging simulation image as shown in Figure 9 can be obtained by graph cut according to above formula.
Face aging analogy method of the invention, aloow final output aging image seamless and it is more aobvious naturally,
Improve the fidelity of Aging simulation.
Referring to Figure 11, as a preferred embodiment, face aging analogy method of the invention is further illustrated, including with
Lower step:
Step S10: image input.
It inputs shown in simulation facial image RCS to be aging such as Fig. 3 (a) and target burn facial image such as such as Fig. 3 (b) institute
Show.
Step S20: Face datection and facial modeling.
Aging simulation facial image SRC and target burn facial image REF is treated, Face datection and face are carried out
Positioning feature point.
Step S30: face alignment.
Target burn facial image REF is subjected to face alignment to simulation facial image SRC to be aging, alignment effect is as schemed
Shown in 4.
Step S40: the calibration of forehead characteristic point.
Face length and the wide general standard ratio of face " three five, front yards " using people, mark the characteristic point at forehead position
It is fixed, as shown in figure 5,7 each forehead characteristic points that the position of small ash point calibrates immediately.
Step S50:Delaunay triangulation
Deformation based on triangular coordinate system.Treat respectively Aging simulation facial image RCS and target burn image REF into
Row trigonometric ratio Delaunay Triangulation, obtains first kind triangular mesh and the second class triangle is grid.Wherein, to be aging
The triangulation effect for simulating facial image is as shown in Figure 6.
Step S60: the traversal of simulation facial image RCS and first kind triangle gridding to be aging.
Each first kind pixel in simulation facial image RCS to be aging is traversed, is identified in first kind triangular mesh
All first kind pixels that each first kind triangular facet separately includes.And then it is not handled in Aging simulation not in first kind triangle
Pixel in face.
Step S70: former cartesian coordinate turns triangular coordinate.
Judge first kind triangular facet locating for first kind pixel (i, j), and records three tops of the first kind triangular facet
Point is a, b, c.The triangle for using three vertex to be constituted calculates first kind pixel as the core of triangular coordinate system
The triangular coordinate of (i, j).The wherein former cartesian coordinate of (i, j) first kind pixel.
Step S80: texture mapping.
RSC is corresponded to using simulation facial image to be aging and target burn facial image REF triangular facet closes correspondingly
System, and pass through its Xin Dika in target burn facial image REF of the triangular coordinate inverse transformation of first kind pixel (i, j)
That coordinate.Complete the texture mapping deformed based on triangular coordinate according to the following formula: the effect after texture mapping is as shown in fig. 7, i.e.
Obtain aging image to be output.
SRCnew(i, j)=SRC (i, j) (1-t)+REF (inew,jnew)·t(1)
Step S90: graph cut.
The to be output aging image to be fused to Fig. 7 carries out graph cut.The second class in the second class triangular mesh
The second class pixel in triangular facet is set to white, is otherwise set to black, exists as a result, as shown in figure 8, obtaining target burn image
Black white binarization image (mask) in second class triangular mesh.Further, the mask as shown in Fig. 4 (b) is obtained respectively
The gradient fields N of image within the scope of human face region shown in the gradient fields M and Fig. 4 (a) in region, further, according to gradient fields M
Obtaining gradient fields Z, Fig. 7 as shown in Figure 7 with gradient fields N is that the signal after texture mapping is carried out in people's face aging analogue method
Figure, aging image to be output as to be fused.Further, divergence is obtained by gradient fields Z as shown in Figure 7.Further
Ground constructs coefficient matrix according to the image size of Fig. 7, is based on coefficient matrix and divergence, and utilize graph cut algorithm, so that
The region mask is inside and outside to realize seamless fusion, and syncretizing effect is as shown in figure 9, i.e. final output Aging simulation image.
The present invention constructs the triangular mesh of simulation facial image and target burn facial image to be aging, is based on triangle
Coordinate system inversion swaps out the corresponding new flute on target burn facial image of each first kind pixel of simulation facial image to be aging
Karr coordinate.And then the pixel value of each pixel of target burn image is accurately acquired, it is each in conjunction with simulation facial image to be aging
The pixel value of first kind pixel treats Aging simulation facial image and carries out Aging simulation.Aging simulation process of the invention compared with
To be simple, the topological deformation process easy to accomplish to image improves Aging simulation image fidelity to be output and image mould
Quasi- timeliness can be effectively synthesized more natural face aging image.
Referring to Figure 12, in one embodiment, the present invention provides a kind of face aging simulators, comprising:
Triangular mesh module 110 is obtained, the simulation facial image to be aging for being aligned to mutual face carries out triangle
Subdivision and obtain first kind triangular mesh, to target burn facial image carry out triangulation and obtain the second class network of triangle
Lattice.
Triangular coordinate conversion module 120, the simulation facial image to be aging for including by first kind triangular mesh
Pixel as first kind pixel, the former cartesian coordinate of each first kind pixel based on simulation facial image to be aging
And the triangle that first kind pixel is in first kind triangular mesh, three vertex of the first kind triangular facet at place are constituted
Coordinate system, the triangular coordinate where obtaining each first kind pixel in first kind triangular facet.
New cartesian coordinate converts module 130, for obtaining the first kind triangular facet where each first kind pixel the
Corresponding second class triangular facet in two class triangular mesh, and it is based on corresponding triangular coordinate and corresponding second class three
The triangular coordinate system that three vertex are constituted in edged surface, obtains new flute of each first kind pixel on target burn facial image
Karr coordinate.
Aging data acquisition module 140, the pixel of the target burn image for including by the second class triangular mesh
It is right on target burn facial image to be acquired according to each new cartesian coordinate for each new cartesian coordinate as the second class pixel
The pixel value for the second class pixel answered.
Aging simulation module 150, for according to first kind pixel it is to be aging simulation facial image on pixel value with
The sum of the pixel value of corresponding second class pixel obtains first kind pixel and is simulating what facial image determined based on to be aging
Pixel value on aging image to be output obtains aging image to be output.
Face aging simulator of the invention, by the simulation facial image and mesh to be aging that obtain mutual face alignment
The first kind triangular mesh and the second class triangular mesh that mark aging facial image triangulation obtains are accurate to determine and obtain
The region of simulation to be aging.Each the and then convenient for according to first kind triangular mesh, and based on simulation facial image to be aging
The former cartesian coordinate of a kind of pixel and the triangular facet in first kind triangular mesh, converse each first kind pixel
Triangular coordinate, from there through in the second class triangular mesh the second class triangular facet and each triangular coordinate, inversion swap out
The pixel value of each first kind pixel corresponding pixel on target burn facial image, and then obtain aging figure to be output
Picture.The present invention constructs the triangular mesh of simulation facial image and target burn facial image to be aging, is based on triangular coordinate
It is that inversion swaps out the corresponding new Descartes on target burn facial image of each first kind pixel of simulation facial image to be aging
Coordinate.And then the pixel value of each pixel of target burn image is accurately acquired, in conjunction with simulation facial image each first to be aging
The pixel value of class pixel treats Aging simulation facial image and carries out Aging simulation.Aging simulation process of the invention is more simple
Single, the topological deformation process easy to accomplish to image improves Aging simulation image fidelity to be output and image simulation
Timeliness can be effectively synthesized more natural face aging image.
In a specific embodiment, Aging simulation module includes:
Texture map unit, for the former cartesian coordinate according to the corresponding first kind pixel of new cartesian coordinate, and
According to texture mapping by the pixel value on determining aging image to be output, it is mapped to simulation facial image to be aging corresponding
A kind of pixel obtains aging image to be output.
The present invention is based on the former cartesian coordinate of texture mapping and first kind pixel is corresponding with new cartesian coordinate
The precision and operation efficiency of Aging simulation can be improved in relationship.
In a specific embodiment, Aging simulation module further include:
Aging adjustment unit, for carrying out operation based on following formula:
SRCnew(i, j)=SRC (i, j) (1-t)+REF (inew,jnew)·t(1)
Wherein, SRCnew(i, j) is the aging figure to be output that first kind pixel is determined based on simulation facial image to be aging
As upper pixel value；SRC (i, j) is pixel value of the first kind pixel on simulation facial image to be aging；REF(inew,
jnew) be new cartesian coordinate corresponding second class pixel on target burn facial image pixel value；T is preset use
In the adjustment constant of adjustment degree of aging；(i, j) is described to be output old to be determined based on the simulation facial image to be aging
Change the former cartesian coordinate of the corresponding first kind pixel on image；(inew, jnew) indicate the first kind pixel
The new cartesian coordinate.
Data processing unit is used for using operation result as first kind pixel true based on simulation facial image to be aging
Pixel value on fixed output image to be aging, obtains output image to be aging.
The present invention is the distortion level for reducing aging image to be output, and first kind pixel is added in simulation face to be aging
Pixel value on image, and will be used to adjust the adjustment constant of degree of aging as coefficient, and then by the operation result of above formula
As the pixel value on output image to be aging of the first kind pixel based on simulation facial image to be aging.The present invention is by each
The pixel value of a kind of pixel converts to complete to treat Aging simulation facial image and carry out Aging simulation, obtains to be aging defeated
Image out improves the efficiency and precision of Aging simulation, so that output image to be aging is close to real human face.
In a specific embodiment, obtaining triangular mesh module includes:
Characteristic point demarcates unit, the first kind predetermined patterns characteristic point for demarcating in simulation facial image to be aging, with
And the second class predetermined patterns characteristic point in target burn facial image；First kind predetermined patterns characteristic point is the alignment of mutual face
When do not carry out human face characteristic point calibration, it is to be aging simulation facial image face position pixel；Second class predetermined patterns
Characteristic point is the pixel for not carrying out human face characteristic point calibration, target burn image face position when mutual face is aligned.
Triangulation unit, for according to each first kind predetermined patterns characteristic point and simulation facial image to be aging
The human face region of each human face characteristic point, the simulation facial image to be aging determined carries out triangulation, obtains first kind triangle
Grid；In each human face characteristic point according to each second class predetermined patterns characteristic point and target burn facial image, determining mesh
The human face region for marking aging facial image carries out triangulation, obtains the second class triangular net.
Present invention determine that the human face region of Aging simulation out, accurately obtains each first kind pixel and exists when helping Aging simulation
The pixel value for the pixel that should be mapped on target burn facial image, and improve the fidelity and operation of face aging simulation
Efficiency.
In a specific embodiment, further includes:
Face alignment module, for calibrating each human face characteristic point and target of simulation facial image to be aging respectively
Each human face characteristic point of aging facial image, and simulation facial image to be aging is subjected to face with target burn image and is aligned.
The present invention is demarcated by human face characteristic point and face is aligned, and helps to treat Aging simulation facial image progress people
When face aging, accurately the face position texture mapping of aging and aging are adjusted.
In a specific embodiment, further includes:
Pixel spider module is identified for traversing each pixel of simulation facial image to be aging in the first kind three
All first kind pixels that each first kind triangular facet separately includes in hexagonal lattice.
Face aging simulator of the invention, the simulation people to be aging that can will be not included in first kind triangular mesh
The pixel of face image excludes, and then identifies the first kind triangular facet that there emerged a where each first kind pixel.The present invention can convert
The triangular coordinate of first kind pixel out facilitates the precision for improving texture mapping when carrying out Aging simulation, simplifies operation
Process and operation efficiency.
In a specific embodiment, further includes:
Binarization block, for carrying out black white binarization to target burn facial image according to the second class triangular mesh,
Obtain black white binarization image of the target burn image in the second class triangular mesh.
Image co-registration module, for carrying out graph cut to aging image to be output, obtaining according to black white binarization image
Final output Aging simulation image.
Face aging simulator of the invention, aloow final output aging image seamless and it is more aobvious naturally,
Improve the fidelity of Aging simulation.
Specific about face aging simulator limits the limit that may refer to above for face aging analogy method
Fixed, details are not described herein.Modules in above-mentioned face aging simulator can fully or partially through software, hardware and its
Combination is to realize.Above-mentioned each module can be embedded in the form of hardware or independently of in the processor in computer equipment, can also be with
It is stored in the memory in computer equipment in a software form, in order to which processor calls the above modules of execution corresponding
Operation.
Referring to Figure 13, in one embodiment, the present invention provides a kind of face aging analog machine, the face aging moulds
It proposes for can be server, internal structure chart can be as shown in figure 13.The system of the face Aging simulation equipment includes logical
Cross processor, the memory, network interface of system bus connection.Wherein, the processor is for providing calculating and control ability.It should
The memory of face aging analog machine includes non-volatile memory medium, built-in storage.Non-volatile memory medium storage
There are operating system, computer program and database.The built-in storage is the operating system and calculating in non-volatile memory medium
The operation of machine program provides environment.The database of the face Aging simulation equipment is for storing image data.The face aging mould
Standby network interface is proposed for communicating with external terminal by network connection.The computer program can when being executed by processor
To realize face aging analogy method.
It will be understood by those skilled in the art that structure shown in Figure 11, only part relevant to application scheme
The block diagram of structure, does not constitute the restriction for the computer equipment being applied thereon to application scheme, and specific computer is set
Standby may include perhaps combining certain components or with different component layouts than more or fewer components as shown in the figure.
In one embodiment, the present invention provides a kind of computer readable storage mediums, are stored thereon with computer journey
Sequence realizes each step of face aging analogy method when computer program is executed by processor.Those of ordinary skill in the art can
To understand all or part of the process realized in above-described embodiment method, be can be instructed by computer program it is relevant hard
Part is completed, and the computer program can be stored in a non-volatile computer read/write memory medium, the computer journey
Sequence is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein, made in each embodiment provided herein
Any reference to memory, storage, database or other media, may each comprise non-volatile and/or volatile storage
Device.Nonvolatile memory may include read-only memory (ROM), programming ROM (PROM), electrically programmable ROM (EPROM), electricity
Erasable programmable ROM (EEPROM) or flash memory.Volatile memory may include random access memory (RAM) or external height
Fast buffer storage.By way of illustration and not limitation, RAM is available in many forms, such as static state RAM (SRAM), dynamic ram
(DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronization link
(Synchlink) DRAM (SLDRAM), memory bus (Rambus) directly RAM (RDRAM), direct memory bus dynamic
RAM (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
In several embodiments provided herein, it should be understood that disclosed device and method can also pass through
Other modes are realized.The apparatus embodiments described above are merely exemplary, for example, flow chart and structure in attached drawing
Figure shows architectural framework in the cards, function and the behaviour of the device of multiple embodiments according to the present invention, method and product
Make.In this regard, each box in flowchart or block diagram can represent a part of a module, section or code, mould
A part of block, program segment or code includes one or more executable instructions for implementing the specified logical function.Also it answers
When note that in the implementation as replacement, function marked in the box can also be to be different from being marked in attached drawing
Sequence occurs.For example, two continuous boxes can actually be basically executed in parallel, they sometimes can also be by opposite suitable
Sequence executes, and this depends on the function involved.It is also noted that each box, Yi Jijie in structure chart and/or flow chart
The combination of box in composition and/or flow chart can use the dedicated hardware based system for executing defined function or movement
System is to realize, or can realize using a combination of dedicated hardware and computer instructions.
In addition, each functional module or unit in each embodiment of the present invention can integrate one independence of formation together
Part, be also possible to modules individualism, an independent part can also be integrated to form with two or more modules.
If function is realized and when sold or used as an independent product in the form of software function module, can store
In a computer readable storage medium.Based on this understanding, technical solution of the present invention is substantially in other words to existing
Having the part for the part or the technical solution that technology contributes can be embodied in the form of software products, the computer
Software product is stored in a storage medium, including some instructions are used so that a computer equipment (can be intelligent hand
Machine, personal computer, server or network equipment etc.) execute all or part of the steps of each embodiment method of the present invention.
And storage medium above-mentioned includes: that USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), arbitrary access are deposited
The various media that can store program code such as reservoir (RAM, Random Access Memory), magnetic or disk.
The above is merely preferred embodiments of the present invention, be not intended to limit the invention, it is all in spirit of the invention and
Within principle, any modification, equivalent substitution, improvement and etc. done be should be included within the scope of the present invention.
Claims (10)
1. a kind of face aging analogy method, which comprises the following steps:
First kind triangular mesh is obtained to the simulation facial image triangulation to be aging of mutual face alignment, to target burn
Facial image triangulation obtains the second class triangular mesh；
The pixel for the simulation facial image to be aging for including using the first kind triangular mesh is as first kind pixel
Point, the former cartesian coordinate and the first kind pixel of each first kind pixel based on the simulation facial image to be aging
Point obtains in the first kind triangular mesh, the triangular coordinate system that three vertex of the first kind triangular facet at place are constituted
To the triangular coordinate where each first kind pixel in the first kind triangular facet；
The first kind triangular facet where obtaining each first kind pixel is corresponding in the second class triangular mesh
The second class triangular facet, and based on three tops in the corresponding triangular coordinate and the corresponding second class triangular facet
The triangular coordinate system that point is constituted, obtains new Descartes of each first kind pixel on the target burn facial image
Coordinate；
The pixel for the target burn image for including using the second class triangular mesh as the second class pixel, according to
Each new cartesian coordinate acquires each new cartesian coordinate corresponding second class on the target burn facial image
The pixel value of pixel；
According to pixel value of the first kind pixel on the simulation facial image to be aging and described corresponding described the
The sum of the pixel value of two class pixels obtains the first kind pixel based on the simulation facial image determination to be aging
Pixel value on aging image to be output obtains the aging image to be output.
2. face aging analogy method according to claim 1, which is characterized in that according to the first kind pixel in institute
State pixel value and the sum of the pixel value of the corresponding second class pixel on simulation facial image to be aging, acquisition described the
Pixel value of a kind of pixel on the aging image to be output determined based on the simulation facial image to be aging, is obtained described
The step of aging image to be output includes:
According to the former cartesian coordinate of the corresponding first kind pixel of the new cartesian coordinate, and will according to texture mapping
Pixel value on the aging image to be output of the determination is mapped to the simulation facial image to be aging corresponding described first
Class pixel obtains the aging image to be output.
3. face aging analogy method according to claim 1, which is characterized in that obtain the aging image to be output
Method includes:
Operation is carried out based on following formula:
SRCnew(i, j)=SRC (i, j) (1-t)+REF (inew,jnew)·t (1)
Wherein, SRCnew(i, j) be the first kind pixel based on described in the simulation facial image to be aging determination to defeated
Pixel value on aging image out；SRC (i, j) is picture of the first kind pixel on the simulation facial image to be aging
Element value；REF(inew,jnew) it is the new cartesian coordinate corresponding second class picture on the target burn facial image
The pixel value of vegetarian refreshments；T is preset for adjusting the adjustment constant of degree of aging；(i, j) is based on the simulation people to be aging
The former cartesian coordinate of the corresponding first kind pixel on the aging image to be output that face image determines；(inew,
jnew) indicate the new cartesian coordinate of the first kind pixel；
Scheme using operation result as first kind pixel in the output to be aging determined based on simulation facial image to be aging
As upper pixel value, the output image to be aging is obtained.
4. face aging analogy method according to claim 1, which is characterized in that be aging being aligned to mutual face
Simulation facial image carries out triangulation and obtains first kind triangular mesh, carries out triangulation to target burn facial image and obtains
To the second class triangular mesh, comprising:
Demarcate the first kind predetermined patterns characteristic point in the simulation facial image to be aging and the target burn face figure
The second class predetermined patterns characteristic point as in；The first kind predetermined patterns characteristic point is that mutual face does not carry out face when being aligned
The pixel at the face position of characteristic point calibration, the described simulation facial image to be aging；The second class predetermined patterns feature
Point does not carry out the pixel at the face position of human face characteristic point calibration, described target burn image when being aligned for mutual face；
According to each first kind predetermined patterns characteristic point and it is described it is to be aging simulation facial image each human face characteristic point,
The human face region of the determining simulation facial image to be aging carries out triangulation, obtains the first kind triangular mesh；
In each human face characteristic point according to each second class predetermined patterns characteristic point and the target burn facial image, determine
The human face region of the target burn facial image carries out triangulation, obtains the second class triangular net.
5. face aging analogy method according to claim 4, which is characterized in that demarcating the simulation face to be aging
The second class predetermined patterns characteristic point in first kind predetermined patterns characteristic point and the target burn facial image in image
Before, further includes:
Each human face characteristic point and the target burn facial image of the simulation facial image to be aging are calibrated respectively
Each human face characteristic point, and the simulation facial image to be aging is subjected to face with the target burn image and is aligned.
6. according to claim 1 to face aging analogy method described in 5 any one, which is characterized in that mutual face
Alignment simulation facial image to be aging carry out triangulation obtain first kind triangular mesh, to target burn facial image into
Row triangulation obtains after the second class triangular mesh, further includes:
Each pixel for traversing the simulation facial image to be aging identifies each described in the first kind triangular mesh
All first kind pixels that first kind triangular facet separately includes.
7. face aging analogy method according to claim 6, which is characterized in that exist obtaining the first kind pixel
The pixel value on aging image to be output determined based on the simulation facial image to be aging, obtains the aging figure to be output
As after, further includes:
According to the second class triangular mesh, black white binarization is carried out to the target burn facial image, obtains the mesh
Mark black white binarization image of the aging image in the second class triangular mesh；
According to the black white binarization image, graph cut is carried out to the aging image to be output, it is old to obtain final output
Change analog image.
8. a kind of face aging simulator characterized by comprising
Triangular mesh module is obtained, the simulation facial image to be aging for being aligned to mutual face carries out triangulation and obtains
First kind triangular mesh carries out triangulation to target burn facial image and obtains the second class triangular mesh；
Triangular coordinate conversion module, the simulation facial image to be aging for including by the first kind triangular mesh
Pixel as first kind pixel, the former Descartes of each first kind pixel based on the simulation facial image to be aging
Coordinate and the first kind pixel are in the first kind triangular mesh, three vertex of the first kind triangular facet at place
The triangular coordinate system of composition, the triangular coordinate where obtaining each first kind pixel in the first kind triangular facet；
New cartesian coordinate converts module, for obtaining the first kind triangular facet where each first kind pixel in institute
State corresponding second class triangular facet in the second class triangular mesh, and based on the corresponding triangular coordinate and corresponding
The triangular coordinate system that three vertex are constituted in the second class triangular facet, obtains each first kind pixel in the target
New cartesian coordinate on aging facial image；
Aging data acquisition module, the pixel of the target burn image for including by the second class triangular mesh
Each new cartesian coordinate is acquired in the target burn according to each new cartesian coordinate as the second class pixel
The pixel value of corresponding second class pixel on facial image；
Aging simulation module, for according to the first kind pixel it is described it is to be aging simulation facial image on pixel value with
The sum of the pixel value of the corresponding second class pixel obtains the first kind pixel based on the mould to be aging
The pixel value on aging image to be output that anthropomorphic face image determines, obtains the aging image to be output.
9. a kind of face aging analog machine, which is characterized in that including memory and processor, the memory is stored with calculating
Machine program, which is characterized in that the processor is realized described in any one of claims 1 to 7 when executing the computer program
The step of method.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program
The step of method described in any one of claims 1 to 7 is realized when being executed by processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811324007.3A CN109584327B (en) | 2018-11-08 | 2018-11-08 | Face aging simulation method, device and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811324007.3A CN109584327B (en) | 2018-11-08 | 2018-11-08 | Face aging simulation method, device and equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109584327A true CN109584327A (en) | 2019-04-05 |
| CN109584327B CN109584327B (en) | 2023-07-25 |
Family
ID=65921811
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811324007.3A Active CN109584327B (en) | 2018-11-08 | 2018-11-08 | Face aging simulation method, device and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109584327B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110163832A (en) * | 2019-05-24 | 2019-08-23 | 北京百度网讯科技有限公司 | Face fusion method, apparatus and terminal |
| CN111275610A (en) * | 2020-01-08 | 2020-06-12 | 杭州趣维科技有限公司 | Method and system for processing face aging image |
| CN111652791A (en) * | 2019-06-26 | 2020-09-11 | 广州虎牙科技有限公司 | Face replacement display method, face replacement display device, live broadcast method, live broadcast device, electronic equipment and storage medium |
| CN111652024A (en) * | 2019-07-05 | 2020-09-11 | 广州虎牙科技有限公司 | Face display method, face live broadcast method, face display device, live broadcast device, electronic equipment and storage medium |
| CN112330529A (en) * | 2020-11-03 | 2021-02-05 | 上海镱可思多媒体科技有限公司 | Dlid-based face aging method, system and terminal |
| CN112884642A (en) * | 2021-03-23 | 2021-06-01 | 福州大学 | Real-time facial aging simulation method based on face recognition technology |
| CN113762173A (en) * | 2021-09-09 | 2021-12-07 | 北京地平线信息技术有限公司 | Training method and device for human face light stream estimation and light stream value prediction model |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104599325A (en) * | 2013-10-30 | 2015-05-06 | 中国科学院苏州纳米技术与纳米仿生研究所 | Three-dimensional face resampling method |
| CN105117712A (en) * | 2015-09-15 | 2015-12-02 | 北京天创征腾信息科技有限公司 | Single-sample human face recognition method compatible for human face aging recognition |
| CN105787974A (en) * | 2014-12-24 | 2016-07-20 | 中国科学院苏州纳米技术与纳米仿生研究所 | Establishment method for establishing bionic human facial aging model |
-
2018
- 2018-11-08 CN CN201811324007.3A patent/CN109584327B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104599325A (en) * | 2013-10-30 | 2015-05-06 | 中国科学院苏州纳米技术与纳米仿生研究所 | Three-dimensional face resampling method |
| CN105787974A (en) * | 2014-12-24 | 2016-07-20 | 中国科学院苏州纳米技术与纳米仿生研究所 | Establishment method for establishing bionic human facial aging model |
| CN105117712A (en) * | 2015-09-15 | 2015-12-02 | 北京天创征腾信息科技有限公司 | Single-sample human face recognition method compatible for human face aging recognition |
Non-Patent Citations (1)
| Title |
|---|
| 陆飞等: ""三角形坐标系下人脸图像连续性研究"", 《计算机辅助设计与图形学学报》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110163832B (en) * | 2019-05-24 | 2021-06-08 | 北京百度网讯科技有限公司 | Face fusion method and device and terminal |
| CN110163832A (en) * | 2019-05-24 | 2019-08-23 | 北京百度网讯科技有限公司 | Face fusion method, apparatus and terminal |
| CN111652791B (en) * | 2019-06-26 | 2024-03-08 | 广州虎牙科技有限公司 | Face replacement display method, face replacement live broadcast device, electronic equipment and storage medium |
| CN111652791A (en) * | 2019-06-26 | 2020-09-11 | 广州虎牙科技有限公司 | Face replacement display method, face replacement display device, live broadcast method, live broadcast device, electronic equipment and storage medium |
| CN111652024A (en) * | 2019-07-05 | 2020-09-11 | 广州虎牙科技有限公司 | Face display method, face live broadcast method, face display device, live broadcast device, electronic equipment and storage medium |
| CN111652024B (en) * | 2019-07-05 | 2023-09-01 | 广州虎牙科技有限公司 | Face display and live broadcast method and device, electronic equipment and storage medium |
| CN111275610B (en) * | 2020-01-08 | 2023-08-18 | 杭州小影创新科技股份有限公司 | Face aging image processing method and system |
| CN111275610A (en) * | 2020-01-08 | 2020-06-12 | 杭州趣维科技有限公司 | Method and system for processing face aging image |
| CN112330529A (en) * | 2020-11-03 | 2021-02-05 | 上海镱可思多媒体科技有限公司 | Dlid-based face aging method, system and terminal |
| CN112884642A (en) * | 2021-03-23 | 2021-06-01 | 福州大学 | Real-time facial aging simulation method based on face recognition technology |
| CN112884642B (en) * | 2021-03-23 | 2022-07-08 | 福州大学 | Real-time facial aging simulation method based on face recognition technology |
| CN113762173A (en) * | 2021-09-09 | 2021-12-07 | 北京地平线信息技术有限公司 | Training method and device for human face light stream estimation and light stream value prediction model |
| CN113762173B (en) * | 2021-09-09 | 2024-05-07 | 北京地平线信息技术有限公司 | Training method and device for face optical flow estimation and optical flow value prediction model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109584327B (en) | 2023-07-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109584327A (en) | Face aging analogy method, device and equipment | |
| CN109325437B (en) | Image processing method, device and system | |
| JP5829371B2 (en) | Facial animation using motion capture data | |
| CN108550176A (en) | Image processing method, equipment and storage medium | |
| KR102602112B1 (en) | Data processing method, device, and medium for generating facial images | |
| CN109493417A (en) | Three-dimension object method for reconstructing, device, equipment and storage medium | |
| CN112950769A (en) | Three-dimensional human body reconstruction method, device, equipment and storage medium | |
| CN110889893B (en) | Three-dimensional model representation method and system for expressing geometric details and complex topology | |
| CN111507259B (en) | Face feature extraction method and device and electronic equipment | |
| Fratarcangeli | Position‐based facial animation synthesis | |
| CN109345604A (en) | Image processing method, computer equipment and storage medium | |
| Wang et al. | Generative Deformable Radiance Fields for Disentangled Image Synthesis of Topology‐Varying Objects | |
| Zhang et al. | [Retracted] Virtual Reality Design and Realization of Interactive Garden Landscape | |
| CN117422802B (en) | Three-dimensional figure digital reconstruction method, device, terminal equipment and storage medium | |
| CN102682473A (en) | Virtual clothing real-time physical modeling method | |
| Iglesias et al. | Hybridizing mesh adaptive search algorithm and artificial immune systems for discrete rational Bézier curve approximation | |
| CN116825127A (en) | Voice-driven digital person generation method based on nerve field | |
| CN116958396A (en) | Image relighting method and device and readable storage medium | |
| CN115953330A (en) | Texture optimization method, device, equipment and storage medium for virtual scene image | |
| CN109934926A (en) | Model data processing method, device, readable storage medium storing program for executing and equipment | |
| CN116883524A (en) | Image generation model training, image generation method and device and computer equipment | |
| US11893673B2 (en) | Morph target animation | |
| CN114119821A (en) | Hair rendering method, device and equipment of virtual object | |
| Capell et al. | Physically based rigging for deformable characters | |
| EP4111420A1 (en) | Face mesh deformation with detailed wrinkles |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |