CN105608451B - Human face portrait generation method based on subspace ridge regression - Google Patents
Human face portrait generation method based on subspace ridge regression Download PDFInfo
- Publication number
- CN105608451B CN105608451B CN201610143041.5A CN201610143041A CN105608451B CN 105608451 B CN105608451 B CN 105608451B CN 201610143041 A CN201610143041 A CN 201610143041A CN 105608451 B CN105608451 B CN 105608451B
- Authority
- CN
- China
- Prior art keywords
- photo
- portrait
- subset
- training
- block
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
Abstract
The invention discloses a kind of human face portrait generation method based on subspace ridge regression, it is of low quality or time-consuming mainly to solve the problems, such as that existing method generates portrait.Implementation step is: (1) dividing training portrait sample set, training photo sample set and test sample collection；(2) all images are divided into image block, and form set of blocks；(3) training photo set of blocks is divided into multiple subsets with corresponding portrait set of blocks；(4) the mapping coefficient matrix between each pair of photo-portrait block subset is calculated；(5) test photo block is divided into corresponding subset；(6) according to each test photo block subset and the corresponding coefficient matrix of subset where it, synthesis portrait block subset is solved；(7) merge synthesis portrait block subset to obtain synthesis portrait set of blocks (8) to combine all synthesis portrait fast, generate pseudo- draw a portrait.The present invention has synthesis portrait quality high, fireballing advantage, can be used for face retrieval and identification in public safety field.
Description
Technical field
The invention belongs to technical field of image processing, in particular to a kind of human face portrait generation method can be used for public peace
Face retrieval and identification in full field.
Background technique
In criminal investigation is chased, public security department has citizen's picture data library, in conjunction with face recognition technology to determine crime
Suspect's identity, but general more difficult acquisition suspect's photo in practice, but can be under the cooperation of artist and witness
The sketch for obtaining suspect is drawn a portrait to carry out subsequent face retrieval and identification.Due to having between portrait and common human face photo
Very big difference is directly difficult to acquire satisfied recognition effect with traditional face identification method.By citizen's picture data
Photo in library is combined into the gap drawn a portrait and can effectively reduced on their textures, and then improves discrimination.
Currently, the human face portrait generation method of mainstream is broadly divided into three classes: first, the method based on sub-space learning；Its
Two, the method based on rarefaction representation；Third, the method based on Bayesian learning.
Q.Liu et al. is in document " Q.Liu, X.Tang, H.Jin, H.Lu, and S.Ma.A Nonlinear Approach
for Face Sketch Synthesis and Recognition.In Proc.IEEE Int.Conference on
Proposed in Computer Vision, pp.1005-1010,2005 " it is a kind of based on the method being locally linear embedding into, by that will instruct
Practice the facial image piecemeal in collection and test set, the K that it is found in training set to each image block of test photo is a similar
Photo neighbour's block, then calculate the linear combination weight of photo neighbour block using the thought that is locally linear embedding into, and using pair
The K portrait neighbour block and Combining weights answered obtain pseudo- portrait block and are finally fused into pseudo- portrait.Deficiency existing for this method
Place is that fixed neighbour's number leads to the blur effect of composograph.
X.Gao et al. is in document " X.Gao, N.Wang, D.Tao, et al.Face Sketch-Photo Synthesis
and Retrieval using Spares Representation.IEEE Transactions on Circuits
2012,22 (8): Systems for Video Technology is proposed a kind of based on rarefaction representation in 1213-1226. "
Human face portrait synthetic method.This method generates the initial estimation of synthesis portrait using existing portrait synthetic method first, then
The method for recycling rarefaction representation synthesizes the detailed information of synthesis portrait, is finally overlapped initial estimation and detailed information
Obtain synthesis portrait to the end.The defect of this method is: synthesizing the quality overwhelming majority of portrait dependent on initial portrait synthesis side
Method, the synthesis of the synthesis and high-frequency information initially drawn a portrait do not carry out under same frame.
X.Wang et al. is in document " X.Wang and X.Tang.Face Photo-Sketch Synthesis and
Recognition.IEEE Transactions on Pattern Analysis and Machine Intelligence.31
(11): 1955-1967 proposes a kind of method based on markov network model in 2009 ", by by training set and test
The facial image of concentration is divided into the localized mass that size is identical, overlaps each other, and is shone using markov network model foundation face
Relational model between piece and human face portrait, by realizing that human face portrait generates to model solution.Deficiency existing for this method
Place is, since each photo block to be synthesized finally only selects a training portrait block to carry out portrait synthesis, to cause blocking artifact serious
And it can not synthesize the portrait block being not present in training portrait set of blocks.
Summary of the invention
It is an object of the invention to be directed to the deficiency of above-mentioned existing method, a kind of face based on subspace ridge regression is proposed
Portrait generation method, to improve the quality of the pseudo- portrait generated, and the speed that significantly promotion portrait generates.
Realize that the technical solution of the object of the invention includes the following steps:
(1) M photos are taken out to concentration from photo-portrait and forms training photo sample set Tp, and take out and training photo
Sample set TpIn the one-to-one M of photo portrait composition training portrait sample set Ts, remaining photo-portrait surveys composition
Sample set is tried, is concentrated from test sample and chooses test photo L, 2≤M≤U-1, U is photo-of the photo-portrait to concentration
Portrait is to number；
(2) all images are divided into the image block of same size and identical overlapping degree, every image is divided into N block,
And form set of blocks:
Test photo L is divided into the image block and composition set of blocks of same size and identical overlapping degree by (2a):
P={ p1,…,pj,…,pN}；
Wherein, P indicates test photo set of blocks, pjIndicate j-th of photo block in test photo, j ∈ 1 ..., N, N；
(2b) will training photo sample set TpIn M photos be divided into the image block of same size and identical overlapping degree
And form set of blocks:
Wherein, R indicates training photo sample set of blocks,Indicate j-th of photo block in i-th photo, i ∈ 1 ...,
M, j ∈ 1 ..., N；
(2c) will train the sample set T that draws a portraitsIn M portrait be divided into the image block of same size and identical overlapping degree
And form set of blocks:
Wherein, Q indicates training portrait sample set of blocks,Indicate i-th portrait in j-th of portrait block, i ∈ 1 ...,
M, j ∈ 1 ..., N；
(3) training photo sample set of blocks R is divided into multiple subsets with corresponding portrait sample set of blocks Q:
RC={ R1,…,Rk,…,Rc}
QC={ Q1,…,Qk,…,Qc}
Wherein, RCIndicate the set of all trained photo sample block subset compositions, RkIndicate each training photo sample block
Subset, QCIndicate the set of all training portrait sample block subset compositions, QkIndicate each training portrait sample block subset, k ∈
1 ..., c, c are subset total number；
(4) photo sample block subset R is trained according to eachkWith corresponding training portrait sample block subset Qk, according to the following formula
Solve the coefficient matrix w of relationship between each pair of subset of mappingk, merge and obtain all subset coefficient matrix set WC=
{w1,…,wk,…wc}:
Wherein, wkIndicate kth to training photo-portrait sample block subset between coefficient matrix,Indicate photo sample
The transposed matrix of block subset matrix, I indicate unit matrix；
(5) the test photo block tested in photo set of blocks P is divided into different training photos-portrait sample block subset
Centering:
PC={ P1,…,Pk,…,Pc}；
Wherein, PCIndicate the set of all test photo sample block subset compositions, PkIndicate each test photo sample block
Subset；
(6) photo sample block subset P is tested according to eachkAnd the corresponding coefficient matrix w of subset where itk, under
Formula solves synthesis portrait block subset Sk:
Sk=wkPk；
(7) by all synthesis portrait block subset SkMerge, obtain synthesis portrait set of blocks:
S={ S1,…,Sk,…,Sc}；
Wherein, SkIndicate k-th of subset in synthesis portrait set of blocks, k ∈ 1 ..., c, c is subset total number；
(8) N number of synthesis portrait block in synthesis portrait set of blocks S is combined, obtains conjunction corresponding with test photo L
At portrait.
Photo-portrait block is divided into several subsets due to using the strategy divided and rule by the present invention, in different subsets
Using the linear relationship between simple formula mapping photo-portrait block and coefficient matrix is acquired, thus high quality puppet can generated
Formation speed is substantially improved while portrait.
To be described in further detail the step of with reference to the accompanying drawing, realization to the present invention.
Detailed description of the invention
Fig. 1 is implementation flow chart of the invention；
Fig. 2 is the comparison knot for the pseudo- portrait that the present invention generates on CUHK student database with existing two methods
Fruit figure.
Specific embodiment
Referring to Fig.1, the step of present invention realizes is as follows:
Step 1, training portrait sample set, training photo sample set and test sample collection are divided.
M photos are taken out to concentration from photo-portrait and form training photo sample set Tp, and take out and training photo sample
Collect TpIn the one-to-one M of photo portrait composition training portrait sample set Ts, by remaining photo-portrait to composition test specimens
This collection is concentrated from test sample and chooses test photo L, 2≤M≤U-1, and U is photo-portrait of the photo-portrait to concentration
To number.
Step 2, figure is carried out to the portrait in training portrait sample set, the photo in training photo sample set and test photo
As block divides.
Test photo L is divided into the image block and composition set of blocks of same size and identical overlapping degree by (2a):
P={ p1,…,pj,…,pN}；
Wherein, P indicates test photo set of blocks, pjIndicate j-th of photo block in test photo, j ∈ 1 ..., N；
(2b) will training photo sample set TpIn M photos be divided into the image block of same size and identical overlapping degree
And form set of blocks:
Wherein, R indicates training photo sample set of blocks,Indicate j-th of photo block in i-th photo, i ∈ 1 ...,
M, j ∈ 1 ..., N；
(2c) will train the sample set T that draws a portraitsIn M portrait be divided into the image block of same size and identical overlapping degree
And form set of blocks:
Wherein, Q indicates training portrait sample set of blocks,Indicate i-th portrait in j-th of portrait block, i ∈ 1 ...,
M, j ∈ 1 ..., N.
Step 3, training photo sample set of blocks and training portrait sample set of blocks are divided into different subsets.
(3a) will train photo sample set of blocks R according to region different demarcation in the photo of photo sample block place to RF、RHTwo
In a subset:
Wherein, RFIndicate the set of blocks that the training photo sample block for being located at facial area forms,It indicates in i-th photo
F-th of photo block of facial area, i ∈ 1 ..., M, f ∈ 1 ..., F, F are the total number of every photo facial area photo block；
RHIndicate the set of blocks that the training photo sample block for being located at hair zones forms,Indicate hair in i-th photo
H-th of photo block in region, i ∈ 1 ..., M, h ∈ 1 ..., H, H are the total number of every photo hair zones photo block；
Hair zones training photo sample set of blocks is divided into different sons using existing K mean cluster algorithm by (3b)
Collection:
Wherein,Indicate the set of all hair zones training photo sample block subset compositions,Indicate each head
Send out regional training photo sample block subset, kh∈1,…,ch, chFor hair zones training photo sample block subset total number；
Facial area training photo sample block is divided into different subsets according to position difference by (3c):
Wherein,Indicate the set of all facial area training photo sample block subset compositions, kf∈1,…,cf, cfFor
Facial area trains photo sample block subset total number, i ∈ 1 ..., M；
(3d) merges all hair zones training photo sample block subset with facial area training photo sample block subset
Obtain total training photo sample block subset set:
RC={ R1,…,Rk,…,Rc}；
Wherein, RCIndicate the set of all trained photo sample block subset compositions, RkIndicate each training photo sample block
Subset, k ∈ 1 ..., c, c are subset total number；
(3e) will training portrait sample block be divided into identical subset correspondingly with training photo sample block, and constitutes
Total training portrait sample block subset set:
QC={ Q1,…,Qk,…,Qc}；
Wherein, QCIndicate the set of all training portrait sample block subset compositions, QkIndicate each training portrait sample block
Subset, k ∈ 1 ..., c；
Step 4, the coefficient matrix of relationship between mapping each pair of photo-portrait block subset is solved.
Each is trained photo sample block subset R by (4a)kWith corresponding training portrait sample block subset QkBetween it is linear
Relationship map are as follows:
Qk=wkRk+Γ
Wherein, wkFor coefficient matrix, Γ is residual matrix；
(4b) coefficient matrix w in (4a) formula in order to obtainkUnique solution, need to following formula unconstrained minimization problem into
Row optimization:
(4c) solves the analytic solutions w of formula minimization problem in (4b) using following formulak:
(4d) repeats (4c), until solving the coefficient matrix w of relationship between each pair of subset of mappingk, merge
To all subset coefficient matrix set WC={ w1,…,wk,…wc}。
Step 5, test photo set of blocks is divided into corresponding sketch-photo sample block subset pair.
PC={ P1,…,Pk,…,Pc}；
Wherein, PCIndicate the set of all test photo sample block subset compositions, PkIndicate each test photo sample block
Subset；
Step 6, pseudo- portrait block subset is calculated.
Photo sample block subset P is tested according to eachkAnd the corresponding coefficient matrix w of subset where itk, according to the following formula
Solve synthesis portrait block subset Sk:
Sk=wkPk。
Step 7, by all synthesis portrait block subset SkMerge, obtain synthesis portrait set of blocks:
S={ S1,…,Sk,…,Sc}；
Wherein, SkIndicate k-th of subset in synthesis portrait set of blocks, k ∈ 1 ..., c, c is subset total number.
Step 8, the synthesis portrait block in synthesis portrait set of blocks S is combined, obtains conjunction corresponding with test photo L
At portrait.
In anabolic process, each synthesis portrait block is arranged according to the sequence of positions of test photo block, to having
Two synthesis portrait block of overlapping region, their pixel values in overlapping region are averaged, are obtained corresponding with test photo L
Puppet portrait.
Effect of the invention can be described further by following emulation experiment.
1. simulated conditions
It is Intel (R) Core i7-4790 3.6GHZ, memory 16G, WINDOWS 7 behaviour that the present invention, which is in central processing unit,
Make in system, the emulation carried out using the MATLAB software that Mathworks company, the U.S. develops.
Control methods used in experiment includes following 2 kinds:
First is that being denoted as LLE in experiment based on the method being locally linear embedding into；Bibliography is Q.Liu, X.Tang,
H.Jin,H.Lu,and S.Ma.A Nonlinear Approach for Face Sketch Synthesis and
Recognition.In Proc.IEEE Int.Conference on Computer Vision,pp.1005-1010,2005；
Second is that the method based on markov weight field model, MWF is denoted as in experiment；Bibliography is H.Zhou,
Z.Kuang,and K.Wong.Markov Weight Fields for Face Sketch Synthesis.In
Proc.IEEE Int.Conference on Computer Vision,pp.1091-1097,2012。
Representation data library used in experiment is CUHK student representation data library disclosed in Hong Kong Chinese University.
2. emulation content
Experiment 1: enterprising in CUHK student representation data library using the present invention and existing LLE method and MWF method
The generation of the pseudo- portrait of row, as a result such as Fig. 2, wherein Fig. 2 (a) is test photo, and Fig. 2 (b) is the pseudo- portrait that LLE method generates, Fig. 2
It (c) is that the puppet that MWF method generates is drawn a portrait, Fig. 2 (d) is the pseudo- portrait that the method for the present invention generates.
From Figure 2 it can be seen that photo-portrait block is divided into several sons since the method for the present invention uses the strategy divided and rule
Collection, so that the relationship between the photo of each pair of subset-portrait sample block is more linear, the portrait of the puppet obtained from is of high quality.
Experiment 2: using two evaluation indexes of structural similarity SSIM and characteristic similarity FSIM to the three kinds of methods of experiment 1
Assembly average carries out quality evaluation to the pseudo- portrait of generation respectively, and SSIM and FSIM are bigger, illustrates the quality of pseudo- portrait generated
Better, the comparing result of three kinds of methods is as shown in table 1:
1 three kinds of methods of table generate the quality evaluation of pseudo- portrait
| Algorithm/evaluation index | SSIM | FSIM |
| LLE | 0.4811 | 0.7064 |
| MWF | 0.4996 | 0.7121 |
| The present invention | 0.5012 | 0.7203 |
As seen from Table 1, the average SSIM and FSIM for the pseudo- portrait that the method for the present invention generates are above two kinds of control methods, say
The pseudo- portrait and original portrait similarity degree that bright the method for the present invention generates are higher, can obtain and preferably generate effect, further
Demonstrate advance of the invention.
Claims (3)
1. the human face portrait generation method based on subspace ridge regression, comprising:
(1) M photos are taken out to concentration from photo-portrait and forms training photo sample set Tp, and take out and training photo sample set
TpIn the one-to-one M of photo portrait composition training portrait sample set Ts, by remaining photo-portrait to composition test sample
Collection is concentrated from test sample and chooses test photo L, 2≤M≤U-1, and U is photo-portrait pair of the photo-portrait to concentration
Number；
(2) all images are divided into the image block of same size and identical overlapping degree, every image is divided into N block, and group
At set of blocks:
Test photo L is divided into the image block and composition set of blocks of same size and identical overlapping degree by (2a):
P={ p1..., pj..., pN}；
Wherein, P indicates test photo set of blocks, pjIndicate j-th of photo block in test photo, j ∈ 1 ..., N；
(2b) will training photo sample set TpIn M photos be divided into the image block and group of same size and identical overlapping degree
At set of blocks:
Wherein, R indicates training photo sample set of blocks,Indicate j-th of photo block in i-th photo, i ∈ 1 ..., M, j ∈
1 ..., N；
(2c) will train the sample set T that draws a portraitsIn M portrait be divided into the image block and group of same size and identical overlapping degree
At set of blocks:
Wherein, Q indicates training portrait sample set of blocks,Indicate j-th of portrait block in i-th portrait, i ∈ 1 ..., M, j ∈
1 ..., N；
(3) training photo sample set of blocks R is divided into multiple subsets with corresponding portrait sample set of blocks Q:
RC={ R1..., Rk..., Rc}
QC={ Q1..., Qk..., Qc}
Wherein, RCIndicate the set of all trained photo sample block subset compositions, RkIndicate each training photo sample block
Collection, QCIndicate the set of all training portrait sample block subset compositions, QkIndicate each training portrait sample block subset, k ∈
1 ..., c, c are subset total number；
(4) photo sample block subset R is trained according to eachkWith corresponding training portrait sample block subset Qk, solve according to the following formula
The coefficient matrix w of relationship between each pair of subset is mapped outk, merge and obtain all subset coefficient matrix set WC={ w1...,
wk... wc}:
Wherein, wkIndicate kth to training photo-portrait sample block subset between coefficient matrix,Indicate photo sample block
Collect the transposed matrix of matrix, I indicates unit matrix；
(5) the N number of test photo block tested in photo set of blocks P is divided into different training photos-portrait sample block subset
Centering:
PC={ P1..., Pk..., Pc}；
Wherein, PCIndicate the set of all test photo sample block subset compositions, PkIndicate each test photo sample block
Collection；
(6) photo sample block subset P is tested according to eachkAnd the corresponding coefficient matrix w of subset where itk, ask according to the following formula
Solve synthesis portrait block subset Sk:
Sk=wkPk；
(7) by all synthesis portrait block subset SkMerge, obtain synthesis portrait set of blocks:
S={ S1..., Sk..., Sc}；
Wherein, SkIndicate k-th of subset in synthesis portrait set of blocks, k ∈ 1 ..., c, c is subset total number；
(8) N number of synthesis portrait block in synthesis portrait set of blocks S is combined, obtains synthesis picture corresponding with test photo L
Picture.
2. based on the human face portrait generation method of subspace ridge regression according to claim 1, which is characterized in that wherein described
(3) training photo sample set of blocks R is divided into multiple subsets with corresponding portrait sample set of blocks Q, its step are as follows:
(3a) will train photo sample set of blocks R according to region different demarcation in the photo of photo sample block place to RF、RHTwo sons
In collection:
Wherein, RFIndicate the set of blocks that the training photo sample block for being located at facial area forms,Indicate i-th photo septum reset
F-th of photo block in region, i ∈ 1 ..., M, f ∈ 1 ..., F, F are the total number of every photo facial area photo block；
RHIndicate the set of blocks that the training photo sample block for being located at hair zones forms,Indicate hair zones in i-th photo
H-th of photo block, i ∈ 1 ..., M, h ∈ 1 ..., H, H be every photo hair zones photo block total number；
Hair zones training photo sample set of blocks is divided into different subsets using existing K mean cluster algorithm by (3b):
Wherein,Indicate the set of all hair zones training photo sample block subset compositions,Indicate each hair area
Train photo sample block subset, k in domainh∈ 1 ..., ch, chFor hair zones training photo sample block subset total number；
Facial area training photo sample block is divided into different subsets according to position difference by (3c):
Wherein,Indicate the set of all facial area training photo sample block subset compositions, kf∈ 1 ..., cf, cfFor face
Regional training photo sample block subset total number, i ∈ 1 ..., M；
All hair zones training photo sample block subset is merged to obtain by (3d) with facial area training photo sample block subset
Total training photo sample block subset set: RC={ R1..., Rk..., Rc}
(3e) will training portrait sample block be divided into identical subset correspondingly with training photo sample block, and constitutes always
Training portrait sample block subset set: QC={ Q1..., Qk..., Qc}。
3. based on the human face portrait generation method of subspace ridge regression according to claim 1, which is characterized in that wherein described
(4) to subset coefficient matrix set W inCIt is solved, is carried out as follows:
Each is trained photo sample block subset R by (4a)kWith corresponding training portrait sample block subset QkBetween linear relationship
Mapping are as follows:
Qk=wkRk+Γ
Wherein, wkFor coefficient matrix, Γ is residual matrix；
(4b) coefficient matrix w in (4a) formula in order to obtainkUnique solution, need to following formula unconstrained minimization problem carry out it is excellent
Change:
(4c) solves the analytic solutions w of formula minimization problem in (4b) using following formulak:
(4d) repeats (4c), until solving the coefficient matrix w of relationship between each pair of subset of mappingk, merge and owned
Subset coefficient matrix set WC={ w1..., wk... wc}。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610143041.5A CN105608451B (en) | 2016-03-14 | 2016-03-14 | Human face portrait generation method based on subspace ridge regression |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610143041.5A CN105608451B (en) | 2016-03-14 | 2016-03-14 | Human face portrait generation method based on subspace ridge regression |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105608451A CN105608451A (en) | 2016-05-25 |
| CN105608451B true CN105608451B (en) | 2019-11-26 |
Family
ID=55988377
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610143041.5A Active CN105608451B (en) | 2016-03-14 | 2016-03-14 | Human face portrait generation method based on subspace ridge regression |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105608451B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110909601B (en) * | 2019-10-18 | 2022-12-09 | 武汉虹识技术有限公司 | Beautiful pupil identification method and system based on deep learning |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101958000A (en) * | 2010-09-24 | 2011-01-26 | 西安电子科技大学 | Face image-picture generating method based on sparse representation |
| CN103902991A (en) * | 2014-04-24 | 2014-07-02 | 西安电子科技大学 | Face recognition method based on forensic sketches |
| CN105138951A (en) * | 2015-07-08 | 2015-12-09 | 西安电子科技大学 | Face sketch-photo recognition method based on graph model representation |
-
2016
- 2016-03-14 CN CN201610143041.5A patent/CN105608451B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101958000A (en) * | 2010-09-24 | 2011-01-26 | 西安电子科技大学 | Face image-picture generating method based on sparse representation |
| CN101958000B (en) * | 2010-09-24 | 2012-08-15 | 西安电子科技大学 | Face image-picture generating method based on sparse representation |
| CN103902991A (en) * | 2014-04-24 | 2014-07-02 | 西安电子科技大学 | Face recognition method based on forensic sketches |
| CN105138951A (en) * | 2015-07-08 | 2015-12-09 | 西安电子科技大学 | Face sketch-photo recognition method based on graph model representation |
Non-Patent Citations (1)
| Title |
|---|
| Local Regression Model for Automatic Face Sketch Generation;Naye Ji et al.;《2011 Sixth International Conference on Image and Graphics》;20111231;第412-417页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105608451A (en) | 2016-05-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | MCnet: Multiple context information segmentation network of no-service rail surface defects | |
| CN109753992A (en) | The unsupervised domain for generating confrontation network based on condition adapts to image classification method | |
| Jiang et al. | Masked swin transformer unet for industrial anomaly detection | |
| Kong et al. | Detect and locate: Exposing face manipulation by semantic-and noise-level telltales | |
| CN108564066A (en) | A kind of person recognition model training method and character recognition method | |
| CN106778512A (en) | Face identification method under the conditions of a kind of unrestricted based on LBP and depth school | |
| CN105138951B (en) | Human face portrait-photo array the method represented based on graph model | |
| CN108960201A (en) | A kind of expression recognition method extracted based on face key point and sparse expression is classified | |
| CN106022241A (en) | Face recognition method based on wavelet transformation and sparse representation | |
| CN105956570A (en) | Lip characteristic and deep learning based smiling face recognition method | |
| CN106056088B (en) | The single sample face recognition method of criterion is generated based on adaptive virtual sample | |
| Liu et al. | A fusion face recognition approach based on 7-layer deep learning neural network | |
| CN105740787B (en) | Identify the face identification method of color space based on multicore | |
| Luo et al. | Research on data augmentation for object detection based on X-ray security inspection picture | |
| CN103605993B (en) | Image-to-video face identification method based on distinguish analysis oriented to scenes | |
| Georgopoulos et al. | Enhancing facial data diversity with style-based face aging | |
| CN105608451B (en) | Human face portrait generation method based on subspace ridge regression | |
| Dong et al. | Learning-based texture synthesis and automatic inpainting using support vector machines | |
| Yaseen et al. | A Novel Approach Based on Multi-Level Bottleneck Attention Modules Using Self-Guided Dropblock for Person Re-Identification | |
| CN106023079B (en) | The two stages human face portrait generation method of joint part and global property | |
| CN106023120B (en) | Human face portrait synthetic method based on coupling neighbour's index | |
| CN116343294A (en) | Pedestrian re-identification method suitable for generalization of field | |
| Zhang et al. | Motion magnification multi-feature relation network for facial microexpression recognition | |
| CN104992185B (en) | Human face portrait generation method based on super-pixel | |
| Zhang et al. | Learn to walk across ages: Motion augmented multi-age group gait video translation |
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 |