CN110222628A - A kind of face restorative procedure based on production confrontation network - Google Patents
A kind of face restorative procedure based on production confrontation network Download PDFInfo
- Publication number
- CN110222628A CN110222628A CN201910475787.XA CN201910475787A CN110222628A CN 110222628 A CN110222628 A CN 110222628A CN 201910475787 A CN201910475787 A CN 201910475787A CN 110222628 A CN110222628 A CN 110222628A
- Authority
- CN
- China
- Prior art keywords
- network
- image
- face
- information
- generates
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- G06V40/172—Classification, e.g. identification
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of face restorative procedures based on production confrontation network, comprising the following steps: S1, acquisition human face data are simultaneously pre-processed；S2, establish confrontation network model: confrontation network model includes two deep neural networks: generating network G and differentiates network D, generates facial image by generating network G；By differentiating that network D judges the true and false of image；S3, carry out face reparation: random adds mask to test image, simulate true picture defect area, this Incomplete image is inputted to generate in network G and generates facial image, the mask region for generating image is substituted into the corresponding position of missing image, then carry out graph cut.The present invention differentiates that loss iteration update the parameter of network using context loss and context loss and overall situation and partial situation two, can generate the simultaneously more natural and true to nature facial image of completion.
Description
Technical field
It is the invention belongs to deep learning and technical field of image processing, in particular to a kind of that network is fought based on production
Face restorative procedure.
Background technique
Image restoration technology is one important branch of field of image processing in recent years, belong to pattern-recognition, machine learning,
The multi-disciplinary cross-cutting issue such as statistics, computer vision.Image repair refers to image caused in image retention process
Loss of learning carries out reconstruction or removes the reparation after the extra object in image.Nowadays, researcher proposes respectively
The method of the image repair of kind various kinds is widely used in the fields such as older picture reparation, historical relic's protection, the extra object of removal.
Currently, it mainly includes three directions that digital picture, which repairs algorithm: structure-based image repair algorithm is based on texture
Image repair algorithm and image repair algorithm based on deep learning.Wherein, document Bertalmio, Marcelo, Sapiro,
2000,4 (9): et al.Image inpainting [J] .Siggraph is proposed a kind of based on partial differential in 417--424.
The reparation algorithm of equation model determines diffusion information and dispersal direction using the marginal information in image region to be repaired, then by
Inside defect area edge-diffusion to defect area, to realize the effect for repairing image.Document Criminisi A, P é rez
P,Toyama K.Region filling and object removal by exemplar-based image
Inpainting [J] .IEEE Transactions on image processing, 2004,13 (9): mentions in 1200-1212.
The Texture Synthesis based on sample block is gone out, the main thought of the algorithm is a picture on random selection zone boundary to be repaired
Vegetarian refreshments according to the textural characteristics of image, chooses sizeable texture block, with the texture block then centered on the pixel
For unit, optimal texture block is matched in the search of image known region, then by the content copy in optimal texture block to be repaired
Region is sufficiently reserved the structure and texture information of absent region.Document Pinho M S, Finamore W.Context-based
2002,38 (20): LZW encoder [J] .Electronics Letters has trained an encoder-in 1172-1174.
Decoder combines the confrontation direct forecast image absent region of network objectives function.
However, it is excessively local currently based on the image repair algorithm application range of structure and texture, it is mainly used for small scale
The reparation of absent region, with the expansion of absent region, repairing effect runs down, and it is endless that there are semantic informations in reparation result
It is whole, the problems such as image obscures, it is unable to reach the requirement of reparation.And the image repair algorithm based on deep learning rests on supervision
On the basis of study, many restrictive factors are brought for image repair.In unsupervised learning field, 2014 by
The production confrontation type network (GAN) that Goodfellow is proposed achieves initiative progress, and therefore, production confrontation network is answered
Research hotspot will be become by using image repair field.
Summary of the invention
There are two the technical problems to be solved by the invention: first is that existing generation fights network, there are network training shakinesses
Fixed and mode crash issue；Second is that existing face, which repairs image, does not meet the not high problem of visual cognition, similarity.For above-mentioned
Problem, the present invention are provided a kind of lost using context and update net with two differentiations loss iteration of context loss and overall situation and partial situation
The parameter of network can generate the face reparation based on production confrontation network of the simultaneously more natural and true to nature facial image of completion
Method.
The purpose of the present invention is achieved through the following technical solutions: it is a kind of based on production confrontation network face repair
Compound method, comprising the following steps:
S1, acquisition human face data are simultaneously pre-processed: collecting great amount of images as data set, the image being collected into is carried out
Pretreatment, is cut into the face training image being sized；
S2, it establishes confrontation network model: fighting network mould using the face training image handled well as data set generation
Two deep neural networks of type: generating network G and differentiates network D, by missing image information input to network G is generated, passes through
It generates network G and generates facial image；It is whole until differentiating that network reaches equilibrium state by differentiating that network D judges the true and false of image
A network model is optimal；
S3, carry out face reparation: random adds mask to test image, true picture defect area is simulated, by this defect
Image is input in step S2 and generates facial image in trained generation network G, and the mask region for generating image is substituted into
The corresponding position of missing image, then carry out graph cut and obtain the facial image of final repairing intact.
Further, the step S1 concrete methods of realizing are as follows: recognition of face is carried out to the image being collected into, extracts face
Information, the mark positioned on the face according to every, by the image cropping being collected at the face training image being sized.
Further, the step S2 concrete methods of realizing are as follows: confrontation network is made of two depth convolutional neural networks:
It generates network G and differentiates network D；
Generating network G is a decoding-coding network, is to be made of two symmetrical VGG16 networks, input is 224*
The missing image information of 224*3 dimension, the reparation image information of 224*224*3 dimension is generated by coding-decoding network；
Differentiate that network D is made of a positive VGG16 network, input the reparation image information tieed up for 224*224*3,
The probability that input data belongs to training data rather than generates sample is obtained by VGG16 network convolution algorithm；
It generates network G and is used to the facial image that the information generation that analogue data is concentrated is similar to truthful data, differentiate network D
Truthful data x is come from for distinguishing the image of input and still generates network G, is schemed until differentiating that network D differentiates not Chu to input
Picture it is true and false, generate confrontation network be then optimal；
Generate the objective function of confrontation network are as follows:
Wherein, V (G, D) indicates to generate the objective function for needing to optimize in confrontation network；D (x) is to differentiate network for defeated
Enter the output of x, G (x) is the output for generating network and being directed to x, and D (G (x)) is first with being input to differentiation net again after generation network query function
In network；X~PdataIndicate that x obeys the distribution P of real data setdata；X~P (z, y) indicates that x obeys image data set to be repaired
It is distributed P (z, y)；Mathematic expectaion is sought in E [] expression；
It can be obtained according to the objective function for generating confrontation network, generate network and differentiate that the objective function of network is respectively as follows:
Wherein, V (G) indicates to generate the objective function of network, and V (D) indicates to differentiate the objective function of network, and M expression is blocked
Mask, M ⊙ G (z, y) indicate to generate the generation information of the non-absent region of image, the non-absent region letter of M ⊙ y expression original image
Breath；Generate confrontation network by gradient descent method minimize loss function to generate network G and differentiate network D parameter progress by
Layer is reversed to be adjusted, and training is iterated.
Further, the step S3 concrete methods of realizing are as follows: specific as follows for the process of image repair: pass through step
Trained generation network G in S2, random adds mask m to test image x, simulates true picture absent region, utilizes generation
Network G generates image G (z, y), then passes through the non-absent region information in composite artwork picture and generate the absent region of image
Information obtains composograph y ':
Y '=(1-M) ⊙ G (z, y)+M ⊙ y (4)
Wherein, the information of the non-absent region of M ⊙ y original image, (1-M) ⊙ G (z, y) indicate to generate the absent region of image
Generation information；Graph cut is carried out to the image y ' of synthesis again and obtains the facial image of final repairing intact.
The beneficial effects of the present invention are: the present invention is producing the facial image for meeting vision using generation confrontation network
On the basis of, it is lost by introducing context loss relevant to the facial image of missing information and the confrontation of missing information, and with
Global and local two differentiations loss is protected together as loss function, while using coding-decoding network as network is generated
Non- missing image information is stayed, by iterative network model, final obtain meets context loss requirement and meet visual cognition
Image is generated, finally realizes effective facial image reparation using the corresponding portion of this life image.For traditional GAN,
Solves the problems, such as unstable, periods of network disruption present in network training.Simultaneously propose using context loss with context loss and
Overall situation and partial situation two differentiates that loss iteration updates the parameter of network, additionally it is possible to generate the simultaneously more natural and true to nature face of completion
Image.
Detailed description of the invention
Fig. 1 is of the invention based on the facial image restorative procedure flow chart for generating confrontation network；
Fig. 2 is the schematic diagram that the present invention generates confrontation network；
Fig. 3 is the schematic diagram that network model is fought in the present invention；
Fig. 4 is the result figure that facial image reparation is carried out using method of the invention.
Specific embodiment
Technical solution of the present invention is further illustrated with reference to the accompanying drawing.
The present invention provides a kind of facial image restorative procedure based on generation confrontation network, can fighting network using generation
On the basis of generation meets the facial image of vision, lost by introducing relevant to the facial image of missing information context and
The confrontation of missing information is lost, and with global and local two differentiations loss together as loss function, while using coding-solution
Code network retains non-missing image information as network is generated, and by iterative network model, final obtain meets context damage
The generation image for requiring and meeting visual cognition is lost, finally realizes that effective facial image is repaired using the corresponding portion of this life image
It is multiple.
As shown in Figure 1, a kind of face restorative procedure based on production confrontation network provided by the invention, including following step
It is rapid:
S1, acquisition human face data are simultaneously pre-processed: collecting great amount of images as data set, the image being collected into is carried out
Pretreatment, is cut into the face training image being sized.Concrete methods of realizing are as follows: existing database CeleA is used,
CeleA data set is a face database, including 202599 famous faces, is trained, is made with wherein 200,000 images
It is tested with 2599 images.Recognition of face is carried out to image using openCV, the information of face is extracted, such as the top of chin
Portion, the outer of eyes, the interior edge of eyebrow etc.；The mark positioned on the face according to every, by the image cropping being collected at setting
The face training image of size, in order to which eyes and mouth can be placed in the middle, picture size is in data set in this example
224*224。
S2, it establishes confrontation network model: fighting network mould using the face training image handled well as data set generation
Two deep neural networks of type: generating network G and differentiates network D, by missing image information input to network G is generated, passes through
It generates network G and generates facial image；It is whole until differentiating that network reaches equilibrium state by differentiating that network D judges the true and false of image
A network model is optimal.
Concrete methods of realizing are as follows: be input to the facial image handled well as data set in generation confrontation network.It generates
Fight network be originated from game theory in zero-sum two-person game, it is made of two depth convolutional neural networks: generate network G and
Differentiate network D, structure is as shown in Figure 2.It generates network G and is used to the data distribution that analogue data is concentrated, generate and be similar to true number
According to facial image；Differentiate that network D is used to extract the feature of input, is equivalent to two classifiers, the image for distinguishing input is
The image generated from truthful data x or G, if sample is from truthful data, D output is true, and otherwise, output is false.Until
Differentiate that network can not differentiate the source of input picture, generates confrontation network and be then optimal.
Generating network G is a decoding-coding network, is to be made of two symmetrical VGG16 networks, input is 224*
The missing image information of 224*3 dimension, the reparation image information of 224*224*3 dimension is generated by coding-decoding network；
Differentiate that network D is made of a positive VGG16 network, input the reparation image information tieed up for 224*224*3,
The probability that input data belongs to training data rather than generates sample, this process such as Fig. 3 institute are obtained by VGG16 network convolution algorithm
Show；
It generates network G and is used to the facial image that the information generation that analogue data is concentrated is similar to truthful data, differentiate network D
Truthful data x is come from for distinguishing the image of input and still generates network G, is schemed until differentiating that network D differentiates not Chu to input
Picture it is true and false, generate confrontation network be then optimal；
Generate the objective function of confrontation network are as follows:
Wherein, V (G, D) indicates to generate the objective function for needing to optimize in confrontation network；D (x) is to differentiate network for defeated
Enter the output of x, G (x) is the output for generating network and being directed to x, and D (G (x)) is first with being input to differentiation net again after generation network query function
In network；X~PdataIndicate that x obeys the distribution P of real data setdata；X~P (z, y) indicates that x obeys image data set to be repaired
It is distributed P (z, y)；Mathematic expectaion is sought in E [] expression；
It can be obtained according to the objective function for generating confrontation network, generate network and differentiate that the objective function of network is respectively as follows:
Wherein, V (G) indicates to generate the objective function of network, and V (D) indicates to differentiate the objective function of network, and M expression is blocked
Mask, M ⊙ G (z, y) indicate to generate the generation information of the non-absent region of image, the non-absent region letter of M ⊙ y expression original image
Breath.
The context generated in network loses LG2, primarily to guaranteeing to generate the non-absent region in image and original image
Non- absent region consistency on messaging, and fight loss LG1Primarily to guaranteeing to generate absent region in image generates information
Validity and authenticity, two loss function formula difference are as follows:
LG1=-EZ~p (z, y)D(G(z,y)) (5)
LG2=EZ~p (z, y)||M⊙G(z,y)-M⊙y||1 (6)
Differentiate in network mainly include global loss LD(g)With local losses LD(l), wherein overall situation loss principal security is raw
At the continuity and accuracy of image entirety, and local losses principal security generates the accuracy of image absent region information.Two
A loss function formula is as follows:
Loss function is minimized to generation network G by gradient descent method and differentiates network D's finally, generating confrontation network
Parameter carries out successively reversed adjusting, by repetitive exercise network to improve the precision of network, so that it is similar to generate generation network
In the facial image of training set.
S3, carry out face reparation: random adds mask to test image, true picture defect area is simulated, by this defect
Image is input in step S2 and generates facial image in trained generation network G, and the mask region for generating image is substituted into
The corresponding position of missing image, then carry out graph cut and obtain the facial image of final repairing intact.Concrete methods of realizing are as follows:
Specific as follows for the process of image repair: by generation network G trained in step S2, random adds to test image x
Mask m simulates true picture absent region, generates image G (z, y) using network G is generated, then passes through in composite artwork picture not
The information of the absent region of absent region information and generation image obtains composograph y ':
Y '=(1-M) ⊙ G (z, y)+M ⊙ y (9)
Wherein, the information of the non-absent region of M ⊙ y original image, (1-M) ⊙ G (z, y) indicate to generate the absent region of image
Generation information；Graph cut is carried out to the image y ' of synthesis again and obtains the facial image of final repairing intact.The present embodiment
It is as shown in Figure 4 that partial face image repairs result.
Those of ordinary skill in the art will understand that the embodiments described herein, which is to help reader, understands this hair
Bright principle, it should be understood that protection scope of the present invention is not limited to such specific embodiments and embodiments.This field
Those of ordinary skill disclosed the technical disclosures can make according to the present invention and various not depart from the other each of essence of the invention
The specific variations and combinations of kind, these variations and combinations are still within the scope of the present invention.
Claims (4)
1. a kind of face restorative procedure based on production confrontation network, which comprises the following steps:
S1, acquisition human face data are simultaneously pre-processed: collecting great amount of images as data set, the image being collected into is located in advance
Reason, is cut into the face training image being sized；
S2, confrontation network model is established: using the face training image handled well as data set generation confrontation network model
Two deep neural networks: generating network G and differentiates network D, by missing image information input to network G is generated, passes through generation
Network G generates facial image；By differentiating that network D judges the true and false of image, until differentiating that network reaches equilibrium state, entire net
Network model is optimal；
S3, carry out face reparation: random adds mask to test image, true picture defect area is simulated, by this Incomplete image
It is input in step S2 and generates facial image in trained generation network G, the mask region for generating image is substituted into missing
The corresponding position of image, then carry out graph cut and obtain the facial image of final repairing intact.
2. a kind of face restorative procedure based on production confrontation network according to claim 1, which is characterized in that described
Step S1 concrete methods of realizing are as follows: recognition of face is carried out to the image being collected into, extracts the information of face, it is fixed on the face according to every
The mark of position, by the image cropping being collected at the face training image being sized.
3. a kind of face restorative procedure based on production confrontation network according to claim 1, which is characterized in that described
Step S2 concrete methods of realizing are as follows: confrontation network is made of two depth convolutional neural networks: generating network G and differentiates network D；
Generating network G is a decoding-coding network, is to be made of two symmetrical VGG16 networks, input is 224*224*3
The missing image information of dimension generates the reparation image information of 224*224*3 dimension by coding-decoding network；
Differentiate that network D is made of a positive VGG16 network, inputs the reparation image information tieed up for 224*224*3, pass through
VGG16 network convolution algorithm obtains the probability that input data belongs to training data rather than generates sample；
It generates network G and is used to the facial image that the information generation that analogue data is concentrated is similar to truthful data, differentiate that network D is used to
The image for distinguishing input comes from truthful data x and still generates network G, until differentiating that network D can not differentiate input picture
It is true and false, it generates confrontation network and is then optimal；
Generate the objective function of confrontation network are as follows:
Wherein, V (G, D) indicates to generate the objective function for needing to optimize in confrontation network；D (x) is to differentiate network for input x
Output, G (x) be generate network be directed to x output, D (G (x)) be first with generate network query function after again be input to differentiation network in；
X~PdataIndicate that x obeys the distribution P of real data setdata；X~P (z, y) indicates that x obeys the distribution P of image data set to be repaired
(z,y)；Mathematic expectaion is sought in E [] expression；
It generates network and differentiates that the objective function of network is respectively as follows:
Wherein, V (G) indicates to generate the objective function of network, and V (D) indicates to differentiate the objective function of network, and mask is blocked in M expression,
M ⊙ G (z, y) indicates to generate the generation information of the non-absent region of image, the non-absent region information of M ⊙ y expression original image；It is raw
It is layer-by-layer anti-to the parameter progress for generating network G and differentiation network D by gradient descent method minimum loss function at network is fought
To adjusting, it is iterated training.
4. a kind of face restorative procedure based on production confrontation network according to claim 1, which is characterized in that described
Step S3 concrete methods of realizing are as follows: specific as follows for the process of image repair: pass through generation network trained in step S2
G, random adds mask m to test image x, simulates true picture absent region, generates image G (z, y) using network G is generated,
Composograph y ' is then obtained by the information of the non-absent region information in composite artwork picture and the absent region for generating image:
Y '=(1-M) ⊙ G (z, y)+M ⊙ y (4)
Wherein, the information of the non-absent region of M ⊙ y original image, (1-M) ⊙ G (z, y) indicate to generate the life of the absent region of image
At information；Graph cut is carried out to the image y ' of synthesis again and obtains the facial image of final repairing intact.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910475787.XA CN110222628A (en) | 2019-06-03 | 2019-06-03 | A kind of face restorative procedure based on production confrontation network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910475787.XA CN110222628A (en) | 2019-06-03 | 2019-06-03 | A kind of face restorative procedure based on production confrontation network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110222628A true CN110222628A (en) | 2019-09-10 |
Family
ID=67818954
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910475787.XA Pending CN110222628A (en) | 2019-06-03 | 2019-06-03 | A kind of face restorative procedure based on production confrontation network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110222628A (en) |
Cited By (34)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110633748A (en) * | 2019-09-16 | 2019-12-31 | 电子科技大学 | Robust automatic face fusion method |
| CN110689495A (en) * | 2019-09-25 | 2020-01-14 | 安徽信息工程学院 | Image restoration method for deep learning |
| CN110689499A (en) * | 2019-09-27 | 2020-01-14 | 北京工业大学 | Face image restoration method based on dense expansion convolution self-coding countermeasure network |
| CN110738161A (en) * | 2019-10-12 | 2020-01-31 | 电子科技大学 | face image correction method based on improved generation type confrontation network |
| CN110889370A (en) * | 2019-11-26 | 2020-03-17 | 上海大学 | System and method for generating end-to-end side face synthesis front face of countermeasure network based on conditions |
| CN111091151A (en) * | 2019-12-17 | 2020-05-01 | 大连理工大学 | Method for generating countermeasure network for target detection data enhancement |
| CN111127355A (en) * | 2019-12-17 | 2020-05-08 | 上海工程技术大学 | Method for finely complementing defective light flow graph and application thereof |
| CN111275638A (en) * | 2020-01-16 | 2020-06-12 | 湖南大学 | Face restoration method for generating confrontation network based on multi-channel attention selection |
| CN111340708A (en) * | 2020-03-02 | 2020-06-26 | 北京理工大学 | Method for rapidly generating high-resolution complete face image according to prior information |
| CN111340122A (en) * | 2020-02-29 | 2020-06-26 | 复旦大学 | Multi-modal feature fusion text-guided image restoration method |
| CN111401138A (en) * | 2020-02-24 | 2020-07-10 | 上海理工大学 | Countermeasure optimization method for generating countermeasure neural network training process |
| CN111539263A (en) * | 2020-04-02 | 2020-08-14 | 江南大学 | Video face recognition method based on aggregation countermeasure network |
| CN111612798A (en) * | 2020-05-15 | 2020-09-01 | 中南大学 | Method, system and medium for repairing complete human face reticulate pattern facing human face data |
| CN111612718A (en) * | 2020-05-21 | 2020-09-01 | 中山大学 | Human face image restoration method introducing attention mechanism |
| CN111667425A (en) * | 2020-06-02 | 2020-09-15 | 大连民族大学 | Prior algorithm-based facial expression image occlusion restoration method |
| CN111814566A (en) * | 2020-06-11 | 2020-10-23 | 北京三快在线科技有限公司 | Image editing method, image editing device, electronic equipment and storage medium |
| CN111882495A (en) * | 2020-07-05 | 2020-11-03 | 东北林业大学 | Image highlight processing method based on user-defined fuzzy logic and GAN |
| CN112270651A (en) * | 2020-10-15 | 2021-01-26 | 西安工程大学 | Image restoration method for generating countermeasure network based on multi-scale discrimination |
| CN112488284A (en) * | 2019-09-11 | 2021-03-12 | 北京京东尚科信息技术有限公司 | Model training method, character image completion method and device and electronic equipment |
| CN112541866A (en) * | 2020-11-24 | 2021-03-23 | 同济大学 | Human face image restoration model based on evolutionary generation countermeasure network |
| CN112541864A (en) * | 2020-09-25 | 2021-03-23 | 中国石油大学(华东) | Image restoration method based on multi-scale generation type confrontation network model |
| CN112837236A (en) * | 2021-01-27 | 2021-05-25 | 浙江大学 | Prosthetic neural network training method and device for image completion, computer equipment and storage medium |
| CN113034355A (en) * | 2021-04-20 | 2021-06-25 | 浙江大学 | Portrait image double-chin removing method based on deep learning |
| CN113112411A (en) * | 2020-01-13 | 2021-07-13 | 南京信息工程大学 | Human face image semantic restoration method based on multi-scale feature fusion |
| CN113298736A (en) * | 2021-06-24 | 2021-08-24 | 河北工业大学 | Face image restoration method based on face pattern |
| CN113538485A (en) * | 2021-08-25 | 2021-10-22 | 广西科技大学 | Contour detection method for learning biological visual pathway |
| CN113673539A (en) * | 2021-08-19 | 2021-11-19 | 广西科技大学 | Progressive interactive contour recognition method based on deep learning model |
| CN113688799A (en) * | 2021-09-30 | 2021-11-23 | 合肥工业大学 | Facial expression recognition method for generating confrontation network based on improved deep convolution |
| CN113870128A (en) * | 2021-09-08 | 2021-12-31 | 武汉大学 | Digital mural image restoration method based on deep convolution impedance network |
| CN113920020A (en) * | 2021-09-26 | 2022-01-11 | 中国舰船研究设计中心 | Human point cloud real-time repairing method based on depth generation model |
| CN114025165A (en) * | 2021-11-03 | 2022-02-08 | 杭州海量信息技术有限公司 | Image compression method and system for maintaining face recognition precision |
| CN114387170A (en) * | 2020-10-19 | 2022-04-22 | 复旦大学 | Image repairing method for improving edge incoherence phenomenon of repairing area |
| CN117218302A (en) * | 2023-11-09 | 2023-12-12 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Doll model generation algorithm based on generation countermeasure network |
| CN118175238A (en) * | 2024-05-14 | 2024-06-11 | 威海凯思信息科技有限公司 | Image generation method and device based on AIGC |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108520503A (en) * | 2018-04-13 | 2018-09-11 | 湘潭大学 | A method of based on self-encoding encoder and generating confrontation network restoration face Incomplete image |
| CN108564611A (en) * | 2018-03-09 | 2018-09-21 | 天津大学 | A kind of monocular image depth estimation method generating confrontation network based on condition |
| CN109377448A (en) * | 2018-05-20 | 2019-02-22 | 北京工业大学 | A kind of facial image restorative procedure based on generation confrontation network |
| CN109559287A (en) * | 2018-11-20 | 2019-04-02 | 北京工业大学 | A kind of semantic image restorative procedure generating confrontation network based on DenseNet |
| CN109584178A (en) * | 2018-11-29 | 2019-04-05 | 腾讯科技（深圳）有限公司 | Image repair method, device and storage medium |
| CN109636867A (en) * | 2018-10-31 | 2019-04-16 | 百度在线网络技术（北京）有限公司 | Image processing method, device and electronic equipment |
| CN109785258A (en) * | 2019-01-10 | 2019-05-21 | 华南理工大学 | A kind of facial image restorative procedure generating confrontation network based on more arbiters |
-
2019
- 2019-06-03 CN CN201910475787.XA patent/CN110222628A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108564611A (en) * | 2018-03-09 | 2018-09-21 | 天津大学 | A kind of monocular image depth estimation method generating confrontation network based on condition |
| CN108520503A (en) * | 2018-04-13 | 2018-09-11 | 湘潭大学 | A method of based on self-encoding encoder and generating confrontation network restoration face Incomplete image |
| CN109377448A (en) * | 2018-05-20 | 2019-02-22 | 北京工业大学 | A kind of facial image restorative procedure based on generation confrontation network |
| CN109636867A (en) * | 2018-10-31 | 2019-04-16 | 百度在线网络技术（北京）有限公司 | Image processing method, device and electronic equipment |
| CN109559287A (en) * | 2018-11-20 | 2019-04-02 | 北京工业大学 | A kind of semantic image restorative procedure generating confrontation network based on DenseNet |
| CN109584178A (en) * | 2018-11-29 | 2019-04-05 | 腾讯科技（深圳）有限公司 | Image repair method, device and storage medium |
| CN109785258A (en) * | 2019-01-10 | 2019-05-21 | 华南理工大学 | A kind of facial image restorative procedure generating confrontation network based on more arbiters |
Non-Patent Citations (2)
| Title |
|---|
| CHAO-YI HUANG 等: "Ultrasound Imaging Improved by the Context Encoder Reconstruction Generative Adversarial Network", 《2018 IEEE INTERNATIONAL ULTRASONICS SYMPOSIUM (IUS)》 * |
| EMILY DENTON等: "SEMI-SUPERVISED LEARNING WITHC ONTEXT-CONDITIONAL GENERATIVE ADVERSARIAL NETWORKS", 《ARXIV》 * |
Cited By (53)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112488284A (en) * | 2019-09-11 | 2021-03-12 | 北京京东尚科信息技术有限公司 | Model training method, character image completion method and device and electronic equipment |
| CN110633748A (en) * | 2019-09-16 | 2019-12-31 | 电子科技大学 | Robust automatic face fusion method |
| CN110633748B (en) * | 2019-09-16 | 2022-06-14 | 电子科技大学 | Robust automatic face fusion method |
| CN110689495A (en) * | 2019-09-25 | 2020-01-14 | 安徽信息工程学院 | Image restoration method for deep learning |
| CN110689495B (en) * | 2019-09-25 | 2022-10-04 | 安徽信息工程学院 | Image restoration method for deep learning |
| CN110689499A (en) * | 2019-09-27 | 2020-01-14 | 北京工业大学 | Face image restoration method based on dense expansion convolution self-coding countermeasure network |
| CN110689499B (en) * | 2019-09-27 | 2023-04-25 | 北京工业大学 | Face image restoration method based on dense expansion convolution self-coding countermeasure network |
| CN110738161A (en) * | 2019-10-12 | 2020-01-31 | 电子科技大学 | face image correction method based on improved generation type confrontation network |
| CN110889370A (en) * | 2019-11-26 | 2020-03-17 | 上海大学 | System and method for generating end-to-end side face synthesis front face of countermeasure network based on conditions |
| CN110889370B (en) * | 2019-11-26 | 2023-10-24 | 上海大学 | System and method for synthesizing face by end-to-end side face based on condition generation countermeasure network |
| CN111127355A (en) * | 2019-12-17 | 2020-05-08 | 上海工程技术大学 | Method for finely complementing defective light flow graph and application thereof |
| CN111091151A (en) * | 2019-12-17 | 2020-05-01 | 大连理工大学 | Method for generating countermeasure network for target detection data enhancement |
| CN113112411B (en) * | 2020-01-13 | 2023-11-24 | 南京信息工程大学 | Human face image semantic restoration method based on multi-scale feature fusion |
| CN113112411A (en) * | 2020-01-13 | 2021-07-13 | 南京信息工程大学 | Human face image semantic restoration method based on multi-scale feature fusion |
| CN111275638B (en) * | 2020-01-16 | 2022-10-28 | 湖南大学 | Face repairing method for generating confrontation network based on multichannel attention selection |
| CN111275638A (en) * | 2020-01-16 | 2020-06-12 | 湖南大学 | Face restoration method for generating confrontation network based on multi-channel attention selection |
| CN111401138A (en) * | 2020-02-24 | 2020-07-10 | 上海理工大学 | Countermeasure optimization method for generating countermeasure neural network training process |
| CN111401138B (en) * | 2020-02-24 | 2023-11-07 | 上海理工大学 | Countermeasure optimization method for generating countermeasure neural network training process |
| CN111340122A (en) * | 2020-02-29 | 2020-06-26 | 复旦大学 | Multi-modal feature fusion text-guided image restoration method |
| CN111340122B (en) * | 2020-02-29 | 2022-04-12 | 复旦大学 | Multi-modal feature fusion text-guided image restoration method |
| CN111340708A (en) * | 2020-03-02 | 2020-06-26 | 北京理工大学 | Method for rapidly generating high-resolution complete face image according to prior information |
| CN111539263B (en) * | 2020-04-02 | 2023-08-11 | 江南大学 | Video face recognition method based on aggregation countermeasure network |
| CN111539263A (en) * | 2020-04-02 | 2020-08-14 | 江南大学 | Video face recognition method based on aggregation countermeasure network |
| CN111612798B (en) * | 2020-05-15 | 2023-04-18 | 中南大学 | Face data pair-oriented complete face reticulate pattern repairing method, system and medium |
| CN111612798A (en) * | 2020-05-15 | 2020-09-01 | 中南大学 | Method, system and medium for repairing complete human face reticulate pattern facing human face data |
| CN111612718A (en) * | 2020-05-21 | 2020-09-01 | 中山大学 | Human face image restoration method introducing attention mechanism |
| CN111667425A (en) * | 2020-06-02 | 2020-09-15 | 大连民族大学 | Prior algorithm-based facial expression image occlusion restoration method |
| CN111667425B (en) * | 2020-06-02 | 2023-05-26 | 大连民族大学 | Facial expression image shielding and repairing method based on priori algorithm |
| CN111814566A (en) * | 2020-06-11 | 2020-10-23 | 北京三快在线科技有限公司 | Image editing method, image editing device, electronic equipment and storage medium |
| CN111882495A (en) * | 2020-07-05 | 2020-11-03 | 东北林业大学 | Image highlight processing method based on user-defined fuzzy logic and GAN |
| CN112541864A (en) * | 2020-09-25 | 2021-03-23 | 中国石油大学(华东) | Image restoration method based on multi-scale generation type confrontation network model |
| CN112270651B (en) * | 2020-10-15 | 2023-12-15 | 西安工程大学 | Image restoration method for generating countermeasure network based on multi-scale discrimination |
| CN112270651A (en) * | 2020-10-15 | 2021-01-26 | 西安工程大学 | Image restoration method for generating countermeasure network based on multi-scale discrimination |
| CN114387170A (en) * | 2020-10-19 | 2022-04-22 | 复旦大学 | Image repairing method for improving edge incoherence phenomenon of repairing area |
| CN112541866B (en) * | 2020-11-24 | 2022-09-13 | 同济大学 | Human face image restoration model based on evolutionary generation countermeasure network |
| CN112541866A (en) * | 2020-11-24 | 2021-03-23 | 同济大学 | Human face image restoration model based on evolutionary generation countermeasure network |
| CN112837236A (en) * | 2021-01-27 | 2021-05-25 | 浙江大学 | Prosthetic neural network training method and device for image completion, computer equipment and storage medium |
| CN112837236B (en) * | 2021-01-27 | 2023-11-07 | 浙江大学 | Method, device, computer equipment and storage medium for training repairing neural network for image complement |
| CN113034355A (en) * | 2021-04-20 | 2021-06-25 | 浙江大学 | Portrait image double-chin removing method based on deep learning |
| CN113298736B (en) * | 2021-06-24 | 2022-03-04 | 河北工业大学 | Face image restoration method based on face pattern |
| CN113298736A (en) * | 2021-06-24 | 2021-08-24 | 河北工业大学 | Face image restoration method based on face pattern |
| CN113673539A (en) * | 2021-08-19 | 2021-11-19 | 广西科技大学 | Progressive interactive contour recognition method based on deep learning model |
| CN113673539B (en) * | 2021-08-19 | 2023-06-20 | 广西科技大学 | Progressive interactive contour recognition method based on deep learning model |
| CN113538485A (en) * | 2021-08-25 | 2021-10-22 | 广西科技大学 | Contour detection method for learning biological visual pathway |
| CN113538485B (en) * | 2021-08-25 | 2022-04-22 | 广西科技大学 | Contour detection method for learning biological visual pathway |
| CN113870128A (en) * | 2021-09-08 | 2021-12-31 | 武汉大学 | Digital mural image restoration method based on deep convolution impedance network |
| CN113870128B (en) * | 2021-09-08 | 2024-10-15 | 武汉大学 | Digital mural image restoration method based on depth convolution countermeasure network |
| CN113920020A (en) * | 2021-09-26 | 2022-01-11 | 中国舰船研究设计中心 | Human point cloud real-time repairing method based on depth generation model |
| CN113688799A (en) * | 2021-09-30 | 2021-11-23 | 合肥工业大学 | Facial expression recognition method for generating confrontation network based on improved deep convolution |
| CN114025165A (en) * | 2021-11-03 | 2022-02-08 | 杭州海量信息技术有限公司 | Image compression method and system for maintaining face recognition precision |
| CN117218302A (en) * | 2023-11-09 | 2023-12-12 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Doll model generation algorithm based on generation countermeasure network |
| CN117218302B (en) * | 2023-11-09 | 2024-04-23 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Doll model generation algorithm based on generation countermeasure network |
| CN118175238A (en) * | 2024-05-14 | 2024-06-11 | 威海凯思信息科技有限公司 | Image generation method and device based on AIGC |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110222628A (en) | A kind of face restorative procedure based on production confrontation network | |
| CN109377448B (en) | Face image restoration method based on generation countermeasure network | |
| CN108596024B (en) | Portrait generation method based on face structure information | |
| CN109543606B (en) | Human face recognition method with attention mechanism | |
| CN113240691B (en) | Medical image segmentation method based on U-shaped network | |
| CN108399649A (en) | A kind of single picture three-dimensional facial reconstruction method based on cascade Recurrent networks | |
| CN109376582A (en) | A kind of interactive human face cartoon method based on generation confrontation network | |
| CN109785258A (en) | A kind of facial image restorative procedure generating confrontation network based on more arbiters | |
| CN109559287A (en) | A kind of semantic image restorative procedure generating confrontation network based on DenseNet | |
| CN110503601A (en) | Face based on confrontation network generates picture replacement method and system | |
| CN110070483A (en) | A kind of portrait cartooning method based on production confrontation network | |
| CN108197606A (en) | The recognition methods of abnormal cell in a kind of pathological section based on multiple dimensioned expansion convolution | |
| CN107578435B (en) | A kind of picture depth prediction technique and device | |
| CN107680158A (en) | A kind of three-dimensional facial reconstruction method based on convolutional neural networks model | |
| CN112861791B (en) | Lip language identification method combining graph neural network and multi-feature fusion | |
| CN111508069B (en) | Three-dimensional face reconstruction method based on single hand-drawn sketch | |
| CN108154104A (en) | A kind of estimation method of human posture based on depth image super-pixel union feature | |
| CN109919048A (en) | A method of face critical point detection is realized based on cascade MobileNet-V2 | |
| CN111243050A (en) | Portrait simple stroke generation method and system and drawing robot | |
| CN109685724A (en) | A kind of symmetrical perception facial image complementing method based on deep learning | |
| CN110909634A (en) | Visible light and double infrared combined rapid in vivo detection method | |
| CN107977618B (en) | Face alignment method based on double-layer cascade neural network | |
| CN111832517A (en) | Low-definition face key point detection method based on gated convolution | |
| CN112927237A (en) | Honeycomb lung focus segmentation method based on improved SCB-Unet network | |
| CN116012922A (en) | Face image gender identification method suitable for mask wearing state |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190910 |