CN109670491A - Identify method, apparatus, equipment and the storage medium of facial image - Google Patents
Identify method, apparatus, equipment and the storage medium of facial image Download PDFInfo
- Publication number
- CN109670491A CN109670491A CN201910138127.2A CN201910138127A CN109670491A CN 109670491 A CN109670491 A CN 109670491A CN 201910138127 A CN201910138127 A CN 201910138127A CN 109670491 A CN109670491 A CN 109670491A
- Authority
- CN
- China
- Prior art keywords
- facial image
- image
- measured
- sample
- encoding encoder
- 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
- 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
-
- 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
- G06V40/40—Spoof detection, e.g. liveness detection
Landscapes
- Engineering & Computer Science (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Image Processing (AREA)
Abstract
The embodiment of the present invention proposes a kind of method, apparatus, equipment and storage medium for identifying facial image.This method comprises: obtaining the first facial image to be measured;Using denoising self-encoding encoder judge the described first facial image to be measured whether be by image conversion process obtained from image;If it is, obtaining the second facial image to be measured of the denoising self-encoding encoder output;Wherein, the denoising self-encoding encoder is to be obtained based on first sample facial image and the second sample facial image training self-encoding encoder, and the second sample facial image is to carry out image obtained from random image transformation to first facial image.The technical solution of the embodiment of the present invention can be based on a large amount of first sample facial image and the second sample facial image off-line training self-encoding encoder, learn the important feature on face out, and then obtain denoising self-encoding encoder.The denoising self-encoding encoder can remove the noise (including image conversion process) in facial image, to realize the function of identifying and forge facial image and forge face video.
Description
Technical field
The present invention relates to technical field of image processing more particularly to it is a kind of identify the method, apparatus of facial image, equipment and
Storage medium.
Background technique
Presently, there are the tools that can be identified with character facial image in swap image, video.Not sending out molecule can be direct
Using the tool, the face in video is replaced, is mixed the spurious with the genuine.Therefore, it is necessary to carry out to whether face in video has carried out replacement
Identify.Traditional authentication technique is mainly based upon Scale invariant features transform (Scale Invariant Feature
Transform, SIFT) feature, the technology is for the forgery image based on duplication stickup with good identification effect.Go out at present
The face replacement technology based on great amount of samples training is showed, if depth is changed face (Deep fakes) technology, it is non-that face replaces effect
Chang Hao, can cheat naked eyes completely, and the detection technique based on SIFT feature is difficult to identify.
Summary of the invention
The embodiment of the present invention provides a kind of method, apparatus, equipment and storage medium for identifying facial image, existing to solve
One or more technical problems in technology.
In a first aspect, the embodiment of the invention provides a kind of methods for identifying facial image, comprising:
Obtain the first facial image to be measured;
Judge whether the described first facial image to be measured is to obtain by image conversion process using denoising self-encoding encoder
Image;
If it is, obtaining the second facial image to be measured of the denoising self-encoding encoder output;
Wherein, the denoising self-encoding encoder is self-editing based on first sample facial image and the training of the second sample facial image
Code device and obtain, the second sample facial image be to first facial image carry out random image transformation obtained from scheme
Picture.
In one embodiment, the method also includes:
Recognition of face is carried out to the described second facial image to be measured, it is corresponding with the described first facial image to be measured to obtain
People information.
In one embodiment, recognition of face is carried out to the described second facial image to be measured, to obtain and described first
The corresponding people information of facial image to be measured, comprising:
Recognition of face is carried out to the described second facial image to be measured respectively using multiple human face recognition models, it is multiple to obtain
The candidate result of recognition of face;
According to the candidate result of the multiple recognition of face, the corresponding personage's letter of the described first facial image to be measured is determined
Breath.
In one embodiment, the method also includes:
Obtain the first sample facial image;
Random image transformation is carried out to the first sample facial image, to obtain the second sample facial image;
Based on the first sample facial image and the second sample facial image training self-encoding encoder.
In one embodiment, based on the first sample facial image and the second sample facial image training institute
State self-encoding encoder, comprising:
Second sample facial image is inputted into the self-encoding encoder;
Obtain the third sample facial image of the self-encoding encoder output;
According to the difference of the third sample facial image and the first sample facial image, loss function value is calculated;
According to the loss function value, the parameter of the self-encoding encoder is adjusted.
In one embodiment, random image described in the method transformation include brightness of image stochastic transformation, image with
At least one of machine scaling, picture contrast stochastic transformation, image Random-Rotation, the random affine transformation of image.
Second aspect, the embodiment of the present invention provide a kind of device for identifying facial image, comprising:
First obtains module, for obtaining the first facial image to be measured;
Judgment module, for judging whether the described first facial image to be measured is to become by image using denoising self-encoding encoder
Change image obtained from processing;
Second obtains module, for judging the described first facial image to be measured for image obtained from image conversion process
In the case where, obtain the second facial image to be measured of the denoising self-encoding encoder output;
Wherein, the denoising self-encoding encoder is self-editing based on first sample facial image and the training of the second sample facial image
Code device and obtain, the second sample facial image be to first facial image carry out random image transformation obtained from scheme
Picture.
In one embodiment, described device further include:
Face recognition module, for carrying out recognition of face to the described second facial image to be measured, to obtain and described first
The corresponding people information of facial image to be measured.
In one embodiment, the face recognition module includes:
Recognition of face submodule, for being carried out respectively to the described second facial image to be measured using multiple human face recognition models
Recognition of face, to obtain the candidate result of multiple recognitions of face;
It determines submodule, for the candidate result according to the multiple recognition of face, determines the described first face figure to be measured
As corresponding people information.
In one embodiment, described device further include:
Third obtains module, for obtaining the first sample facial image;
Random image conversion module, for carrying out random image transformation to the first sample facial image, to obtain
State the second sample facial image;
Training module, for described certainly based on the first sample facial image and the second sample facial image training
Encoder.
In one embodiment, the training module includes:
Input submodule, for the second sample facial image to be inputted the self-encoding encoder;
Acquisition submodule, for obtaining the third sample facial image of the self-encoding encoder output;
Computational submodule, for the difference according to the third sample facial image and the first sample facial image,
Calculate loss function value;
Adjusting submodule, for adjusting the parameter of the self-encoding encoder according to the loss function value.
In one embodiment, random image described in described device transformation include brightness of image stochastic transformation, image with
At least one of machine scaling, picture contrast stochastic transformation, image Random-Rotation, the random affine transformation of image.
The third aspect, the embodiment of the invention provides a kind of equipment for identifying facial image, the function of the equipment can be with
By hardware realization, corresponding software realization can also be executed by hardware.The hardware or software include it is one or more with
The corresponding module of above-mentioned function.
It include processor and memory in the structure of the equipment in a possible design, the memory is used for
Storage supports described device to execute the program of the above method, the processor is configured to storing in the memory for executing
Program.The equipment can also include communication interface, be used for and other equipment or communication.
Fourth aspect, the embodiment of the invention provides a kind of computer readable storage mediums, identify face figure for storing
Computer software instructions used in the equipment of picture comprising for executing program involved in the above method.
Above-mentioned technical proposal is based on a large amount of first sample facial image and the second sample facial image off-line training is self-editing
Code device, can learn the important feature on face out, and then obtain denoising self-encoding encoder.The denoising self-encoding encoder can remove people
Noise (including image conversion process) in face image, to realize the function for identifying and forging facial image and forging face video
Energy.It is possible to further carry out recognition of face to the facial image after denoising by manual identified and human face recognition model, with
People information into original image.
Above-mentioned general introduction is merely to illustrate that the purpose of book, it is not intended to be limited in any way.Except foregoing description
Schematical aspect, except embodiment and feature, by reference to attached drawing and the following detailed description, the present invention is further
Aspect, embodiment and feature, which will be, to be readily apparent that.
Detailed description of the invention
In the accompanying drawings, unless specified otherwise herein, otherwise indicate the same or similar through the identical appended drawing reference of multiple attached drawings
Component or element.What these attached drawings were not necessarily to scale.It should be understood that these attached drawings depict only according to the present invention
Disclosed some embodiments, and should not serve to limit the scope of the present invention.
Fig. 1 shows the flow chart of the method according to an embodiment of the present invention for identifying facial image.
Fig. 2 shows a kind of flow charts of the method for the identification facial image of embodiment according to embodiments of the present invention.
Fig. 3 shows the flow chart of the method for the identification facial image of another embodiment according to embodiments of the present invention.
Fig. 4 shows the flow chart of the method for the identification facial image of another embodiment according to embodiments of the present invention.
Fig. 5 shows the schematic diagram of digital pattern according to an embodiment of the present invention.
Fig. 6 shows the flow chart of the method for the identification facial image of another embodiment according to embodiments of the present invention.
Fig. 7 shows the training process exemplary diagram of the method according to an embodiment of the present invention for identifying facial image.
Fig. 8 shows the structural block diagram of the device according to an embodiment of the present invention for identifying facial image.
Fig. 9 shows a kind of structural block diagram of the device of the identification facial image of embodiment according to embodiments of the present invention.
Figure 10 shows the structural block diagram of the equipment according to an embodiment of the present invention for identifying facial image.
Specific embodiment
Hereinafter, certain exemplary embodiments are simply just described.As one skilled in the art will recognize that
Like that, without departing from the spirit or scope of the present invention, described embodiment can be modified by various different modes.
Therefore, attached drawing and description are considered essentially illustrative rather than restrictive.
Fig. 1 shows the flow chart of the method according to an embodiment of the present invention for identifying facial image.As shown in Figure 1, this method
It may comprise steps of:
Step S101, the first facial image to be measured is obtained;
Step S102, judge whether the described first facial image to be measured is by image transformation using denoising self-encoding encoder
Image obtained from reason, if it is, entering step S103;
Step S103, the second facial image to be measured of the denoising self-encoding encoder output is obtained.
Wherein, the denoising self-encoding encoder is self-editing based on first sample facial image and the training of the second sample facial image
Code device and obtain, the second sample facial image be to first facial image carry out random image transformation obtained from scheme
Picture.
In one embodiment, the random image transformation scales at random, schemes including brightness of image stochastic transformation, image
Image contrast stochastic transformation, image Random-Rotation, the random affine transformation of image (Affine Transformation or Affine
At least one of Map).
In one example, the second facial image input denoising self-encoding encoder to be measured can be encoded certainly with obtaining denoising
The loss function value of device output may determine that the first face to be measured of input if loss function value is more than the first preset value
Image is image obtained from image conversion process.
In one example, video to be measured is the video of personage A (forging people information).People can be carried out to video to be measured
Face detection, and then obtain first facial image P11, P12, P13 ... to be measured that multiframe includes personage's A face-image.By each
One facial image P11, P12, P13 ... to be measured inputs denoising self-encoding encoder respectively, respectively obtains the multiple second face figures to be measured
As P21, P22, P23 ....By taking the first facial image P11 to be measured as an example, the first facial image P11 to be measured is the side face of personage A
Image, brightness are darker.Denoising self-encoding encoder can remove the image conversion process in the first facial image P11 to be measured, with reduction
First facial image P11 to be measured, obtains the second facial image P21 to be measured, and the second facial image P21 to be measured is the positive face of personage A
Image, brightness are brighter.According to the second facial image P21 to be measured, it can be determined that video to be measured may be forgery video.Similar,
Video to be measured can also be judged respectively according to second facial image P22, P23 to be measured that reduction obtains.In addition, based on more
Whether a second facial image P21, P22, P23 ... to be measured can be to forge video with comprehensive descision video to be measured.
Since during forging video, face can not be always face video camera, it is therefore desirable to the people of personage A
Face image carries out image conversion process, at luminance transformation processing, scale transformation processing, contrast variation's processing, rotation transformation
Reason, affine transformation processing etc., to obtain the facial image of multiframe personage A, then replace the facial image of multiframe personage A respectively
The facial image of personage B in original video, to obtain forging video.
According to the method for the embodiment of the present invention, can be removed at the image transformation in facial image based on denoising self-encoding encoder
Reason restores facial image, and then can judge whether facial image or video are to forge according to the facial image after reduction.
In one embodiment, as shown in Fig. 2, the method for the identification facial image of the embodiment of the present invention can also wrap
It includes:
Step S201, recognition of face is carried out to the described second facial image to be measured, to obtain and the described second face to be measured
The corresponding people information of image.
In one example, the first facial image to be measured may be by the personage C (real person's information) in original image
It is forged into personage A, then carries out the image that image conversion process obtains again.First facial image input denoising to be measured is encoded certainly
Device can restore the first facial image to be measured, remove noise (including image conversion process), obtain clean second to
Survey facial image.Recognition of face is carried out to the second facial image to be measured after denoising, can be improved and obtain the first face figure to be measured
The accuracy rate that people information as in is personage C, that is, improve the accuracy rate for obtaining the people information in original image.
In one embodiment, as shown in figure 3, may include: in step s 201
Step S301, recognition of face is carried out to the described second facial image to be measured respectively using multiple human face recognition models,
To obtain the candidate result of multiple recognitions of face;
Step S302, according to the candidate result of the multiple recognition of face, described and the described first face figure to be measured is determined
As corresponding people information.
In the present embodiment, human face recognition model may include FaceNet, DeepFace, DeepID etc..
In one example, recognition of face can be carried out to the second facial image to be measured using FaceNet, obtains the first time
Select result: the confidence level that the people information in the first facial image to be measured is personage C is 0.9, in the first facial image to be measured
People information is that the confidence level of personage A is 0.1.Recognition of face can be carried out to the second facial image to be measured using DeepFace,
Obtain the second candidate result: the confidence level 0.8 that the people information in the first facial image to be measured is personage C, the first face to be measured
People information in image is that the confidence level of personage A is 0.2.Face can be carried out to the second facial image to be measured using DeepID
Identification, obtains third candidate result: the confidence level 0.5 that the people information in the first facial image to be measured is personage C, people information
Confidence level for personage A is 0.5.Each candidate result of overall merit can determine personage corresponding with the first facial image to be measured
Information is personage C (real person's information).
In one example, candidate result may include carrying out the obtained knot of manual identified to the second facial image to be measured
Fruit.
In one embodiment, the method for the identification facial image of the embodiment of the present invention may include denoising self-encoding encoder
Training method.As shown in figure 4, the training method may include:
Step S401, the first sample facial image is obtained;
Step S402, random image transformation is carried out to the first sample facial image, to obtain the second sample people
Face image;
Step S403, described from coding based on the first sample facial image and the second sample facial image training
Device.
When understanding complexity, primary feature can be first summarized, is then summed up from primary features advanced
Feature.As shown in figure 5, being summarized for identifying digital pattern by study, discovery can " 1,5 " be indicated digital pattern
For several very simple sub-patternsWithCombination.By shining a large amount of black and white landscape
Piece extracts the images fragment analysis of 16*16, it is found that almost all of images fragment can be combined to obtain by 64 kinds of orthogonal sides.
Sound also has same situation, and available 20 kinds of basic structure, most sound are ok in the audio not marked largely
It is obtained by these basic structure linear combinations.Here it is the sparse expressions of feature, lead to too small amount of essential characteristic combination, assembly
Obtain the abstract feature of higher.
Self-encoding encoder (Auto Encoders) is one of deep learning network model, can be used for automatically completing
The process of this feature extraction and expression, and whole process is unsupervised.The denoising self-encoding encoder of the embodiment of the present invention is
On the basis of self-encoding encoder, by then utilizing " the addling " of Noise to its input noise (including random image converts)
Sample go reconstruct without noise it is " clean " input.This Training strategy enables denoising autocoder to learn to more capable of
Reflect the substantive characteristics of input data.
In one embodiment, as shown in fig. 6, may include: in step S403
Step S601, the second sample facial image is inputted into the self-encoding encoder;
Step S602, the third sample facial image of the self-encoding encoder output is obtained;
Step S603, according to the difference of the third sample facial image and the first sample facial image, damage is calculated
Lose functional value;
Step S604, according to the loss function value, the parameter of the self-encoding encoder is adjusted.
Below with reference to an example of the training method of Fig. 7 introduction denoising self-encoding encoder.As shown in fig. 7, self-encoding encoder can
To include convolutional layer (64 cores, size are [3,3]), pond layer and inverse convolutional layer.Originally exemplary method may include:
It step S701, is [224,224,3] by the first sample facial image superposition size that size is [224,224,3]
Gaussian noise;
Step S702, the first sample facial image progress image after superposition Gaussian noise is scaled at random, brightness of image
Stochastic transformation, picture contrast stochastic transformation, image Random-Rotation and image radiate transformation at random, be to obtain size [224,
224,3] the second sample facial image;
Step S703, the second sample facial image is inputted into self-encoding encoder, to generate the third that size is [224,224,3]
Sample facial image;
Step S704, the difference between first sample facial image and third sample facial image is compared, loss letter is calculated
Numerical value;
Step S705, counter give of loss function value is passed into self-encoding encoder, and then adjusts the parameter of self-encoding encoder.
Circulation executes step S701 to S705, is less than preset threshold until damaging functional value, minimizes loss function value,
The training to self-encoding encoder is completed, trained self-encoding encoder can be used as the denoising self-encoding encoder of the embodiment of the present invention.
In one example, loss function can be binary cross entropy (binary_crossentropy), i.e. logarithm loses
Function (log loss).The formula of loss function can be L (Y, P (Y | X))=- logP (Y | X), wherein X indicates first sample
Facial image;Y indicates third sample facial image;P indicates root;L indicates loss function value.The parameter for adjusting self-encoding encoder can be with
Parameter including adjustment convolutional layer, pond layer and inverse convolutional layer.
Based on a large amount of first sample facial image and the second sample facial image off-line training self-encoding encoder, can learn
Important feature on face out, and then obtain denoising self-encoding encoder.The denoising self-encoding encoder can remove making an uproar in facial image
Sound (including image conversion process), to realize the function of identifying and forge facial image and forge face video.
Video is forged based on Deepfakes technology, is substantially to be superimposed noise on the original image, to have cheated people's
Visually.Using denoising self-encoding encoder, remove the noise of Deepfakes superposition, thus what identification was forged by Deepfakes technology
Facial image or video.
It is possible to further carry out face knowledge to the facial image after denoising by manual identified and human face recognition model
Not, to obtain the people information in original image.
Fig. 8 shows the structural block diagram of the device according to an embodiment of the present invention for identifying facial image.As shown in figure 8, the dress
It sets and may include:
First obtains module 801, for obtaining the first facial image to be measured;
Judgment module 802, for judging whether the described first facial image to be measured is by figure using denoising self-encoding encoder
The image as obtained from conversion process;
Second obtains module 803, for judging the described first facial image to be measured for obtained from image conversion process
In the case where image, the second facial image to be measured of the denoising self-encoding encoder output is obtained;
Wherein, the denoising self-encoding encoder is self-editing based on first sample facial image and the training of the second sample facial image
Code device and obtain, the second sample facial image be to first facial image carry out random image transformation obtained from scheme
Picture.
In one embodiment, as shown in figure 9, described device can also include:
Face recognition module 901, for carrying out recognition of face to the described second facial image to be measured, to obtain and described the
The corresponding people information of one facial image to be measured.
In one embodiment, as shown in figure 9, face recognition module 901 may include:
Recognition of face submodule 911, for using multiple human face recognition models respectively to the described second facial image to be measured
Recognition of face is carried out, to obtain the candidate result of multiple recognitions of face;
It determines submodule 912, for the candidate result according to the multiple recognition of face, determines the described first face to be measured
The corresponding people information of image.
In one embodiment, as shown in figure 9, described device can also include:
Third obtains module 902, for obtaining the first sample facial image;
Random image conversion module 903, for carrying out random image transformation to the first sample facial image, to obtain
The second sample facial image;
Training module 904, for based on the first sample facial image and the second sample facial image training institute
State self-encoding encoder.
In one embodiment, as shown in figure 9, training module 904 may include:
Input submodule 941, for the second sample facial image to be inputted the self-encoding encoder;
Acquisition submodule 942, for obtaining the third sample facial image of the self-encoding encoder output;
Computational submodule 943, for the difference according to the third sample facial image and the first sample facial image
Not, loss function value is calculated;
Adjusting submodule 944, for adjusting the parameter of the self-encoding encoder according to the loss function value.
In one embodiment, the random image transformation scales at random, schemes including brightness of image stochastic transformation, image
At least one of image contrast stochastic transformation, image Random-Rotation, the random affine transformation of image.
The function of each module in each device of the embodiment of the present invention may refer to the corresponding description in the above method, herein not
It repeats again.
Figure 10 shows the structural block diagram of the equipment according to an embodiment of the present invention for identifying facial image.As shown in Figure 10, should
Equipment may include: memory 1001 and processor 1002, and being stored in memory 1001 can execute on processor 1002
Computer program.The processor 1002 realizes the identification facial image in above-described embodiment when executing the computer program
Method.The quantity of the memory 1001 and processor 1002 can be one or more.
The equipment can also include:
Communication interface 1003 carries out data interaction for being communicated with external device.
Memory 1001 may include high speed RAM memory, it is also possible to further include nonvolatile memory (non-
Volatile memory), a for example, at least magnetic disk storage.
If memory 1001, processor 1002 and the independent realization of communication interface 1003, memory 1001, processor
1002 and communication interface 1003 can be connected with each other by bus and complete mutual communication.The bus can be industrial mark
Quasi- architecture (ISA, Industry Standard Architecture) bus, external equipment interconnection (PCI,
Peripheral Component Interconnect) bus or extended industry-standard architecture (EISA, Extended
Industry Standard Component) bus etc..The bus can be divided into address bus, data/address bus, control bus
Deng.Only to be indicated with a thick line in Figure 10, it is not intended that an only bus or a type of bus convenient for indicating.
Optionally, in specific implementation, if memory 1001, processor 1002 and communication interface 1003 are integrated in one piece
On chip, then memory 1001, processor 1002 and communication interface 1003 can complete mutual communication by internal interface.
The embodiment of the invention provides a kind of computer readable storage mediums, are stored with computer program, the program quilt
Processor realizes any method in above-described embodiment when executing.
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means specific features, structure, material or spy described in conjunction with this embodiment or example
Point is included at least one embodiment or example of the invention.Moreover, particular features, structures, materials, or characteristics described
It may be combined in any suitable manner in any one or more of the embodiments or examples.In addition, without conflicting with each other, this
The technical staff in field can be by the spy of different embodiments or examples described in this specification and different embodiments or examples
Sign is combined.
In addition, term " first ", " second " are used for descriptive purposes only and cannot be understood as indicating or suggesting relative importance
Or implicitly indicate the quantity of indicated technical characteristic." first " is defined as a result, the feature of " second " can be expressed or hidden
It include at least one this feature containing ground.In the description of the present invention, the meaning of " plurality " is two or more, unless otherwise
Clear specific restriction.
Any process described otherwise above or method description are construed as in flow chart or herein, and expression includes
It is one or more for realizing specific logical function or process the step of executable instruction code module, segment or portion
Point, and the range of the preferred embodiment of the present invention includes other realization, wherein can not press shown or discussed suitable
Sequence, including according to related function by it is basic simultaneously in the way of or in the opposite order, Lai Zhihang function, this should be of the invention
Embodiment person of ordinary skill in the field understood.
Expression or logic and/or step described otherwise above herein in flow charts, for example, being considered use
In the order list for the executable instruction for realizing logic function, may be embodied in any computer-readable medium, for
Instruction execution system, device or equipment (such as computer based system, including the system of processor or other can be held from instruction
The instruction fetch of row system, device or equipment and the system executed instruction) it uses, or combine these instruction execution systems, device or set
It is standby and use.For the purpose of this specification, " computer-readable medium ", which can be, any may include, stores, communicates, propagates or pass
Defeated program is for instruction execution system, device or equipment or the dress used in conjunction with these instruction execution systems, device or equipment
It sets.The more specific example (non-exhaustive list) of computer-readable medium include the following: there is the electricity of one or more wirings
Interconnecting piece (electronic device), portable computer diskette box (magnetic device), random access memory (RAM), read-only memory
(ROM), erasable edit read-only storage (EPROM or flash memory), fiber device and portable read-only memory
(CDROM).In addition, computer-readable medium can even is that the paper that can print described program on it or other suitable Jie
Matter, because can then be edited, be interpreted or when necessary with other for example by carrying out optical scanner to paper or other media
Suitable method is handled electronically to obtain described program, is then stored in computer storage.
It should be appreciated that each section of the invention can be realized with hardware, software, firmware or their combination.Above-mentioned
In embodiment, software that multiple steps or method can be executed in memory and by suitable instruction execution system with storage
Or firmware is realized.It, and in another embodiment, can be under well known in the art for example, if realized with hardware
Any one of column technology or their combination are realized: having a logic gates for realizing logic function to data-signal
Discrete logic, with suitable combinational logic gate circuit specific integrated circuit, programmable gate array (PGA), scene
Programmable gate array (FPGA) etc..
Those skilled in the art are understood that realize all or part of step that above-described embodiment method carries
It suddenly is that relevant hardware can be instructed to complete by program, the program can store in a kind of computer-readable storage medium
In matter, which when being executed, includes the steps that one or a combination set of embodiment of the method.
It, can also be in addition, each functional unit in each embodiment of the present invention can integrate in a processing module
It is that each unit physically exists alone, can also be integrated in two or more units in a module.Above-mentioned integrated mould
Block both can take the form of hardware realization, can also be realized in the form of software function module.The integrated module is such as
Fruit is realized and when sold or used as an independent product in the form of software function module, also can store in a computer
In readable storage medium storing program for executing.The storage medium can be read-only memory, disk or CD etc..
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can readily occur in its various change or replacement,
These should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with the guarantor of the claim
It protects subject to range.
Claims (14)
1. a kind of method for identifying facial image characterized by comprising
Obtain the first facial image to be measured;
Using denoising self-encoding encoder judge the described first facial image to be measured whether be by image conversion process obtained from figure
Picture;
If it is, obtaining the second facial image to be measured of the denoising self-encoding encoder output;
Wherein, the denoising self-encoding encoder is based on first sample facial image and the second sample facial image training self-encoding encoder
And obtain, the second sample facial image is to carry out image obtained from random image transformation to first facial image.
2. the method according to claim 1, wherein further include:
Recognition of face is carried out to the described second facial image to be measured, to obtain personage corresponding with the described first facial image to be measured
Information.
3. according to the method described in claim 2, it is characterized in that, carry out recognition of face to the described second facial image to be measured,
To obtain people information corresponding with the described first facial image to be measured, comprising:
Recognition of face is carried out to the described second facial image to be measured respectively using multiple human face recognition models, to obtain multiple faces
The candidate result of identification;
According to the candidate result of the multiple recognition of face, the corresponding people information of the described first facial image to be measured is determined.
4. the method according to claim 1, wherein further include:
Obtain the first sample facial image;
Random image transformation is carried out to the first sample facial image, to obtain the second sample facial image;
Based on the first sample facial image and the second sample facial image training self-encoding encoder.
5. according to the method described in claim 4, it is characterized in that, being based on the first sample facial image and second sample
This facial image training self-encoding encoder, comprising:
Second sample facial image is inputted into the self-encoding encoder;
Obtain the third sample facial image of the self-encoding encoder output;
According to the difference of the third sample facial image and the first sample facial image, loss function value is calculated;
According to the loss function value, the parameter of the self-encoding encoder is adjusted.
6. method according to any one of claims 1 to 5, which is characterized in that the random image transformation includes that image is bright
Degree stochastic transformation, image scale at random, picture contrast stochastic transformation, image Random-Rotation, in the random affine transformation of image
It is at least one.
7. a kind of device for identifying facial image characterized by comprising
First obtains module, for obtaining the first facial image to be measured;
Judgment module, for judging whether the described first facial image to be measured is by image transformation using denoising self-encoding encoder
Image obtained from reason;
Second obtains module, for judging the described first facial image to be measured for the feelings of image obtained from image conversion process
Under condition, the second facial image to be measured of the denoising self-encoding encoder output is obtained;
Wherein, the denoising self-encoding encoder is based on first sample facial image and the second sample facial image training self-encoding encoder
And obtain, the second sample facial image is to carry out image obtained from random image transformation to first facial image.
8. device according to claim 7, which is characterized in that further include:
Face recognition module, it is to be measured with acquisition and described first for carrying out recognition of face to the described second facial image to be measured
The corresponding people information of facial image.
9. device according to claim 8, which is characterized in that the face recognition module includes:
Recognition of face submodule, for carrying out face to the described second facial image to be measured respectively using multiple human face recognition models
Identification, to obtain the candidate result of multiple recognitions of face;
It determines submodule, for the candidate result according to the multiple recognition of face, determines the described first facial image pair to be measured
The people information answered.
10. device according to claim 7, which is characterized in that further include:
Third obtains module, for obtaining the first sample facial image;
Random image conversion module, for carrying out random image transformation to the first sample facial image, to obtain described the
Two sample facial images;
Training module encodes described in the first sample facial image and the second sample facial image training certainly for being based on
Device.
11. device according to claim 10, which is characterized in that the training module includes:
Input submodule, for the second sample facial image to be inputted the self-encoding encoder;
Acquisition submodule, for obtaining the third sample facial image of the self-encoding encoder output;
Computational submodule is calculated for the difference according to the third sample facial image and the first sample facial image
Loss function value;
Adjusting submodule, for adjusting the parameter of the self-encoding encoder according to the loss function value.
12. according to the described in any item devices of claim 7 to 11, which is characterized in that the random image transformation includes image
Brightness stochastic transformation, image scale at random, picture contrast stochastic transformation, image Random-Rotation, in the random affine transformation of image
At least one.
13. a kind of equipment for identifying facial image characterized by comprising
One or more processors;
Storage device, for storing one or more programs;
When one or more of programs are executed by one or more of processors, so that one or more of processors
Realize such as method described in any one of claims 1 to 6.
14. a kind of computer readable storage medium, is stored with computer program, which is characterized in that the program is held by processor
Such as method described in any one of claims 1 to 6 is realized when row.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910138127.2A CN109670491A (en) | 2019-02-25 | 2019-02-25 | Identify method, apparatus, equipment and the storage medium of facial image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910138127.2A CN109670491A (en) | 2019-02-25 | 2019-02-25 | Identify method, apparatus, equipment and the storage medium of facial image |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109670491A true CN109670491A (en) | 2019-04-23 |
Family
ID=66151376
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910138127.2A Pending CN109670491A (en) | 2019-02-25 | 2019-02-25 | Identify method, apparatus, equipment and the storage medium of facial image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109670491A (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110210393A (en) * | 2019-05-31 | 2019-09-06 | 百度在线网络技术(北京)有限公司 | The detection method and device of facial image |
CN110648393A (en) * | 2019-09-18 | 2020-01-03 | 广州智美科技有限公司 | Glasses customization method and device based on 3D face model and terminal |
CN111739046A (en) * | 2020-06-19 | 2020-10-02 | 百度在线网络技术(北京)有限公司 | Method, apparatus, device and medium for model update and image detection |
CN111783505A (en) * | 2019-05-10 | 2020-10-16 | 北京京东尚科信息技术有限公司 | Method and device for identifying forged faces and computer-readable storage medium |
CN111797931A (en) * | 2020-07-08 | 2020-10-20 | 北京字节跳动网络技术有限公司 | Image processing method, image processing network training method, device and equipment |
CN111860043A (en) * | 2019-04-26 | 2020-10-30 | 北京陌陌信息技术有限公司 | Face image-changing model training method, device, equipment and medium |
CN112668453A (en) * | 2020-12-24 | 2021-04-16 | 平安科技(深圳)有限公司 | Video identification method and related equipment |
CN113449543A (en) * | 2020-03-24 | 2021-09-28 | 百度在线网络技术(北京)有限公司 | Video detection method, device, equipment and storage medium |
KR20230030907A (en) * | 2021-08-26 | 2023-03-07 | 세종대학교산학협력단 | Method for fake video detection and apparatus for executing the method |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104298973A (en) * | 2014-10-09 | 2015-01-21 | 北京工业大学 | Face image rotation method based on autoencoder |
CN108288072A (en) * | 2018-01-26 | 2018-07-17 | 深圳市唯特视科技有限公司 | A kind of facial expression synthetic method based on generation confrontation network |
CN108710831A (en) * | 2018-04-24 | 2018-10-26 | 华南理工大学 | A kind of small data set face recognition algorithms based on machine vision |
AU2018101528A4 (en) * | 2018-10-14 | 2018-11-15 | Li, Junjie Mr | Camouflage image encryption based on variational auto-encoder(VAE) and discriminator |
-
2019
- 2019-02-25 CN CN201910138127.2A patent/CN109670491A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104298973A (en) * | 2014-10-09 | 2015-01-21 | 北京工业大学 | Face image rotation method based on autoencoder |
CN108288072A (en) * | 2018-01-26 | 2018-07-17 | 深圳市唯特视科技有限公司 | A kind of facial expression synthetic method based on generation confrontation network |
CN108710831A (en) * | 2018-04-24 | 2018-10-26 | 华南理工大学 | A kind of small data set face recognition algorithms based on machine vision |
AU2018101528A4 (en) * | 2018-10-14 | 2018-11-15 | Li, Junjie Mr | Camouflage image encryption based on variational auto-encoder(VAE) and discriminator |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111860043A (en) * | 2019-04-26 | 2020-10-30 | 北京陌陌信息技术有限公司 | Face image-changing model training method, device, equipment and medium |
CN111783505A (en) * | 2019-05-10 | 2020-10-16 | 北京京东尚科信息技术有限公司 | Method and device for identifying forged faces and computer-readable storage medium |
CN110210393A (en) * | 2019-05-31 | 2019-09-06 | 百度在线网络技术(北京)有限公司 | The detection method and device of facial image |
CN110648393A (en) * | 2019-09-18 | 2020-01-03 | 广州智美科技有限公司 | Glasses customization method and device based on 3D face model and terminal |
CN113449543A (en) * | 2020-03-24 | 2021-09-28 | 百度在线网络技术(北京)有限公司 | Video detection method, device, equipment and storage medium |
CN111739046A (en) * | 2020-06-19 | 2020-10-02 | 百度在线网络技术(北京)有限公司 | Method, apparatus, device and medium for model update and image detection |
CN111797931A (en) * | 2020-07-08 | 2020-10-20 | 北京字节跳动网络技术有限公司 | Image processing method, image processing network training method, device and equipment |
CN112668453A (en) * | 2020-12-24 | 2021-04-16 | 平安科技(深圳)有限公司 | Video identification method and related equipment |
CN112668453B (en) * | 2020-12-24 | 2023-11-14 | 平安科技(深圳)有限公司 | Video identification method and related equipment |
KR20230030907A (en) * | 2021-08-26 | 2023-03-07 | 세종대학교산학협력단 | Method for fake video detection and apparatus for executing the method |
KR102629848B1 (en) * | 2021-08-26 | 2024-01-25 | 세종대학교산학협력단 | Method for fake video detection and apparatus for executing the method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109670491A (en) | Identify method, apparatus, equipment and the storage medium of facial image | |
Xiao et al. | Image splicing forgery detection combining coarse to refined convolutional neural network and adaptive clustering | |
CN111598761B (en) | Anti-printing shooting image digital watermarking method based on image noise reduction | |
CN108171701B (en) | Significance detection method based on U network and counterstudy | |
CN108604369B (en) | Method, device and equipment for removing image noise and convolutional neural network | |
CN111444881A (en) | Fake face video detection method and device | |
CN109492627B (en) | Scene text erasing method based on depth model of full convolution network | |
CN112818862A (en) | Face tampering detection method and system based on multi-source clues and mixed attention | |
CN110059728B (en) | RGB-D image visual saliency detection method based on attention model | |
CN111292265A (en) | Image restoration method based on generating type antagonistic neural network | |
CN108596003B (en) | Contaminated two-dimensional code repairing method and system based on machine learning | |
CN109948397A (en) | A kind of face image correcting method, system and terminal device | |
CN106408037A (en) | Image recognition method and apparatus | |
JP2021531571A (en) | Certificate image extraction method and terminal equipment | |
CN109740572A (en) | A kind of human face in-vivo detection method based on partial color textural characteristics | |
Singh et al. | Steganalysis of digital images using deep fractal network | |
CN111833360B (en) | Image processing method, device, equipment and computer readable storage medium | |
CN110619334B (en) | Portrait segmentation method based on deep learning, architecture and related device | |
CN105956570B (en) | Smiling face's recognition methods based on lip feature and deep learning | |
CN107832794A (en) | A kind of convolutional neural networks generation method, the recognition methods of car system and computing device | |
CN107256543A (en) | Image processing method, device, electronic equipment and storage medium | |
CN114742985A (en) | Hyperspectral feature extraction method and device and storage medium | |
CN111275070B (en) | Signature verification method and device based on local feature matching | |
CN113111880A (en) | Certificate image correction method and device, electronic equipment and storage medium | |
CN113610071B (en) | Face living body detection method and device, electronic equipment and storage medium |
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 |
Application publication date: 20190423 |
|
RJ01 | Rejection of invention patent application after publication |