CN107423700B - Method and device for verifying testimony of a witness - Google Patents
Method and device for verifying testimony of a witness Download PDFInfo
- Publication number
- CN107423700B CN107423700B CN201710581244.7A CN201710581244A CN107423700B CN 107423700 B CN107423700 B CN 107423700B CN 201710581244 A CN201710581244 A CN 201710581244A CN 107423700 B CN107423700 B CN 107423700B
- Authority
- CN
- China
- Prior art keywords
- face image
- network
- certificate
- natural light
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000012795 verification Methods 0.000 claims abstract description 53
- 238000012549 training Methods 0.000 claims description 72
- 230000006870 function Effects 0.000 claims description 44
- 238000005286 illumination Methods 0.000 claims description 10
- 230000008447 perception Effects 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 claims description 3
- 239000000523 sample Substances 0.000 claims 27
- 239000013074 reference sample Substances 0.000 claims 4
- 230000007123 defense Effects 0.000 claims 1
- 230000001815 facial effect Effects 0.000 claims 1
- 230000008569 process Effects 0.000 description 9
- 238000010586 diagram Methods 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 239000004576 sand Substances 0.000 description 2
- 208000009119 Giant Axonal Neuropathy Diseases 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 201000003382 giant axonal neuropathy 1 Diseases 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 210000000697 sensory organ Anatomy 0.000 description 1
Images
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
- G06V40/161—Detection; Localisation; Normalisation
-
- 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
- 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)
- Human Computer Interaction (AREA)
- Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Multimedia (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a method and a device for verifying testimony. The method comprises the following steps: acquiring a certificate face image in a certificate photo, and acquiring a natural light face image of a user; inputting the certificate face image into a pre-trained generation countermeasure network, and obtaining a reconstructed face image corresponding to the certificate face image according to the output of the generation countermeasure network; the generation countermeasure network is used for adding preset natural light attribute information to the input certificate face image, and the resolution of the output reconstructed face image is higher than that of the certificate face image; and comparing the reconstructed face image with the natural light face image, and verifying the testimony according to the comparison result. The invention can effectively improve the accuracy of authentication and verification.
Description
Technical Field
The invention relates to the technical field of face recognition, in particular to a method, a device and the like for verifying a testimony of a witness.
Background
With the rapid development of face recognition technology, people's identity card verification systems are also receiving wide attention, and the demands on practical application scenarios such as security and banking are sharply increased. In the heterogeneous face recognition technology, the difference of the modes makes the face images have great difference, which is a main reason for difficulty in accurate discrimination. The testimony verification problem belongs to heterogeneous face recognition and is to judge whether a certificate photo face image with lower resolution is matched with a natural illumination face image with higher resolution.
The biggest problem of the testimonial verification system is that the verification accuracy is low, and the traditional heterogeneous face recognition method can only eliminate modal difference from two aspects of feature extraction and similarity measurement for verification, and cannot be well suitable for testimonial verification.
Disclosure of Invention
Based on this, the embodiment of the invention provides a method and a device for testimony verification, which can effectively solve the problem of low testimony verification accuracy caused by modal heterogeneity.
In one aspect, the present invention provides a method for testimonial verification, comprising:
acquiring a certificate face image in a certificate photo, and acquiring a natural light face image of a user;
inputting the certificate face image into a pre-trained generation countermeasure network, and obtaining a reconstructed face image corresponding to the certificate face image according to the output of the generation countermeasure network; the generation countermeasure network is used for adding preset natural light attribute information to the input certificate face image, and the resolution of the output reconstructed face image is higher than that of the certificate face image;
and comparing the reconstructed face image with the natural light face image, and verifying the testimony according to the comparison result.
One aspect of the present invention provides an apparatus for testimony verification, comprising:
the face image acquisition module is used for acquiring a certificate face image in the certificate photo and acquiring a natural light face image of a user;
the face image reconstruction module is used for inputting the certificate face image into a pre-trained generation countermeasure network and obtaining a reconstructed face image corresponding to the certificate face image according to the output of the generation countermeasure network; the generation countermeasure network is used for adding preset natural light attribute information to the input certificate face image, and the resolution of the output reconstructed face image is higher than that of the certificate face image;
and the testimony verification module is used for comparing the reconstructed face image with the natural light face image and verifying the testimony according to the comparison result.
One aspect of the present invention provides a computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the method when executing the program.
According to the technical scheme, a certificate face image in a certificate photo is acquired, and a natural light face image of a user is acquired; inputting the certificate face image into a pre-trained generation countermeasure network, and obtaining a reconstructed face image corresponding to the certificate face image according to the output of the generation countermeasure network; the generation countermeasure network is used for adding preset natural light attribute information to the input certificate face image, and the resolution of the output reconstructed face image is higher than that of the certificate face image; and finally, verifying the testimony by comparing the reconstructed face image with the natural light face image. Therefore, on the premise of keeping the information of the original certificate photo image, super-resolution reconstruction of the certificate photo face image (converting a low-resolution image into a high-resolution image) can be realized, the certificate photo is converted into a natural illumination mode, the heterogeneous human identity card verification problem is converted into a general face recognition problem, and the accuracy of human identity card verification is effectively improved.
Drawings
FIG. 1 is a schematic flow chart diagram of a method of witness verification of an embodiment;
FIG. 2 is a schematic diagram of a generator network according to an embodiment;
FIG. 3 is a schematic flow chart diagram of generating a resistance network retrain of an embodiment;
fig. 4 is a schematic configuration diagram of an apparatus for testimonial verification according to an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
FIG. 1 is a schematic flow chart diagram of a method of witness verification of an embodiment; as shown in fig. 1, the method for testimony verification in the present embodiment includes the steps of:
s11, acquiring a certificate face image in the certificate photo, and collecting a natural light face image of a user;
it will be appreciated that the document referred to in embodiments of the invention may be any document with an image of a person's face, for example: identity cards, passports, drivers licenses, passports, or business cards, and the like. The identification photo can be a photo pasted in the identification card or a photo printed on the identification card, such as a second generation identification card. In addition, the identification photo can be black and white or can be colored.
It should be noted that the natural light face image refers to an image acquired in real time in the current environment, and includes both an image acquired in real time in an outdoor natural light environment and an image acquired in real time in an indoor light environment.
S12, inputting the certificate face image into a pre-trained generation countermeasure network, and obtaining a reconstructed face image corresponding to the certificate face image according to the output of the generation countermeasure network;
the generation of a countermeasure network (generic adaptive Networks, referred to as GANs for short) is a generation model, and the basic idea is to obtain many training samples from a training library so as to learn the probability distribution generated by the training cases. The realization method is to make two network models compete with each other to play a game. One of them, called the generator network (or generator, generating network), is used to generate images that look 'natural', requiring as much as possible correspondence with the original data distribution; the other is called a network of discriminators (or discriminator, discriminating network) for determining whether a given image appears 'natural', in other words, as if it was artificially (machine) generated. The generator network is targeted to 'spoof' the arbiter network, and the arbiter network is targeted to not go up the generator network. When the two groups of network models are trained continuously, the generator network continuously generates new results to try, and the capacities of the two groups of network models are mutually improved until the artificial samples generated by the generator network are not different from the original samples in appearance.
The resolution ratio of the certificate face image collected from the certificate is lower, the generation countermeasure network in the step can be used for carrying out high-resolution reconstruction on the certificate face image with low resolution ratio, preset natural light attribute information is added, and missing pixel parts in the original image can be compensated.
In the embodiment of the invention, the resolution of the reconstructed face image output by the generation countermeasure network is higher than that of the certificate face image. Ideally, the modality of the reconstructed face image obtained by generating the countermeasure network is consistent with the modality of the corresponding natural light face image.
And S13, comparing the reconstructed face image with the natural light face image, and verifying the testimony according to the comparison result.
Based on the above steps, the certificate face image has been reconstructed (i.e. mode conversion), so step S13 is ordinary face image recognition, rather than heterogeneous face image recognition, which on one hand reduces the complexity of recognition and on the other hand is also beneficial to improving the accuracy of recognition.
By the method for verifying the testimony of the person, the testimony can be converted into a natural illumination mode on the premise of keeping the image information of the original testimony, the problem of verifying the testimony of the person is further converted into the problem of identifying a general face image, and the accuracy of verifying the testimony of the person is effectively improved.
In an alternative embodiment, step S11 includes: and after the certificate photo in the evidence is collected, face detection and alignment are carried out, and a face area of the face five sense organ area is cut out to obtain a certificate face image. The natural light face image is preferably a front high-definition image.
In an alternative embodiment, the generation countermeasure network employed in the above step S12 includes a generator network and a discriminator network. The generator network comprises a 6-layer residual error convolution network structure, as shown in fig. 2, wherein the first 3 layers are convolution layers, the second 3 layers are reverse convolution layers, and the last reverse convolution layer outputs a reconstructed face image. The specific network configuration is, for example: in the 6-layer residual convolutional network structure, a batch normalization layer (BN layer for short) and a ReLU activation function layer are connected behind each 3 × 3 convolutional layer, so that the input of each convolutional layer keeps the same distribution, and the overall training speed of the network is high. From the first five layers, 64 feature maps are extracted per layer, stride is set to 1, pad is set to 0, and the last reverse convolution layer is used to reconstruct the image. The network structure not only can make the network easier to train, but also can keep and utilize the image information of the input certificate face image to reconstruct a high-resolution image. Preferably, the pixel size of the reconstructed image output by the generator network is consistent with the pixel size of the natural light face image collected in step S11. The discriminator network comprises a Light CNN residual network structure, which is beneficial to enhancing the feature robustness and reducing the network parameters. Preferably, the Max Feature Map operation is used as the activation function using the networkkin Network architecture in Light CNN. The advantages of using a Network in Network architecture include: better local abstraction, smaller global overriding, and fewer network parameters.
In an optional embodiment, the method further comprises the step of pre-training the generation of the countermeasure network, including training the generator network and training the discriminator network. The way of training may be: firstly, generating a countermeasure network based on an ImageNet database and pre-training the countermeasure network; and then retraining the generated confrontation network after the pre-training based on a preset testimony sample library until the generated confrontation network meeting the preset conditions is obtained. The ImageNet database is the largest database for image recognition in the world at present; the testimony sample library comprises a plurality of testimony photo samples and natural light face image samples corresponding to the testimony photo samples. Through the training of different databases in two stages, a generation countermeasure network meeting the requirements of certificate verification can be obtained, so that a certificate face image with low resolution is converted into a reconstructed face image with high resolution.
In an optional embodiment, retraining the pre-trained generated confrontation network based on the pre-set testimonial sample library comprises: the two training processes are mutually influenced for the training of a generator network in the generation countermeasure network and the training of a discriminator network in the generation countermeasure network, so that the two networks are gradually perfected to meet the network model of the testimony verification requirement.
Wherein the process of generating generator network training in the countermeasure network comprises: acquiring a certificate photo sample and a natural light face image sample corresponding to the certificate photo sample from a certificate sample library, acquiring the certificate face image sample from the certificate photo sample, taking the certificate face image sample as the input of a generator network, and training the network parameters of the generator network based on a square loss function until the square loss function is minimized; the square loss function is a function of the pixel-based squared difference of the natural light face image samples and the reconstructed face image output by the generator network.
The process of training the arbiter network in the countermeasure network comprises the following steps: taking the natural light face image sample and a reconstructed face image output by a generator network as the input of a discriminator network, and training the network parameters of the discriminator network and the network parameters of the generator network based on a perception loss function; the perception loss function is a function of the probability of distinguishing the reconstructed face image output by the generator network into a real natural light face image.
In a preferred embodiment, the training process for generating the countermeasure network is illustrated with reference to FIG. 3. Let IyFor high resolution natural light face image samples, IxFor low resolution certificate face image samples, IsTo reconstruct a face image; the reconstructed face image and the natural light face image sample have the same pixel size. Generating a countermeasure network from a generator network
And arbiter network
Composition, θ denotes Generation of countermeasure networkThe network parameters to be trained.
Inputting certificate face image sample I with lower resolution ratio into generator networkxReconstructed face image I output by generator networksTo a generator network. The generator network corresponds to a function model of
In an alternative embodiment, the objective function of the network parameters of the generator network is trained
Comprises the following steps:
θGnetwork parameters representing the generator network,/sAs a function of the square loss, N is the total number of reference photographs taken in training, IyRepresenting natural light face image samples, IxRepresenting a sample of a face image of a document, IsAnd representing a reconstructed face image corresponding to the certificate face image sample.
Preferably, used in the generator network training is a loss function based on the squared loss of pixels (MSE), i.e.:
wherein r represents the size ratio of the natural light face image to the certificate face image; w represents the pixel of the certificate face image in the width direction, and H represents the pixel of the certificate face image in the length direction.
Reconstructing a face image I by network output of a generatorsAnd high-resolution natural light face image sample IyInputting the discriminator network and training the discriminator network. The function model corresponding to the discriminator network is expressed as
θDRepresenting network parameters of a network of discriminators, the task of which is to discriminate the generator network
Resulting reconstructed image
(i.e. I)s) If true, the "infinitesimal max game" problem is solved, and the objective function can be expressed as:
wherein, Ε represents a mathematical expectation, Iy～pdata(Iy) Representing natural light face image samples IyProbability distribution satisfying high resolution image is pdata(Iy)；Ix～pG(Ix) Representing a sample of a document face image IxSatisfies the probability distribution of the generator as pG(Iy) (ii) a log represents a logarithmic operation;
is to use a natural light face image sample IyJudging the probability of the face image with real natural light;
representation arbiter network
Generator network
Output reconstructed image
And judging the probability of the face image as real natural light.
Training discriminator
Maximizing the probability of correct labels for input instances and generated samples while training the generator
Minimization
When globally optimal, there is pdata＝pGI.e. the generator network can fit the probability distribution of the high resolution image perfectly.
Preferably, the reconstructed image output by the generator network is generated if the discriminator network is connected to the generator network
And (4) judging the probability of the real natural light face image, outputting 1 by the discriminator network, and otherwise, outputting-1. And optimizing the discriminator network according to the discrimination result of the discriminator network. I.e. the objective function of the optimized arbiter network is a function related to the decision result of the arbiter network.
As the MSE loss function adopted by the generator network training can lose the high-frequency information of the input image to a certain extent and cause a fuzzy phenomenon, the perception loss function is added into the discrimination network, and the discriminator network is optimized from the perception angle. I.e. using the perceptual loss function
Optimizing the objective function of the arbiter network:
wherein,
representation arbiter network
Generator network
Output reconstructed image
And judging the probability of the face image as real natural light.
Network of generators
And arbiter network
And when the training is finished, finishing the training of the generation countermeasure network. When the certificate verification is carried out, the acquired original certificate face image can be reconstructed based on the trained generation countermeasure network, and then the authentication verification is carried out based on the reconstructed face image and the natural light face image acquired in real time, so that the verification accuracy is improved.
Therefore, the whole task of generating the countermeasure network is like a game, the generator network ensures that the discriminator network confuses the authenticity of the reconstructed image, and the target of the discriminator network is to distinguish the authenticity of the image as far as possible. Therefore, the training mode of the generation countermeasure network is different from the conventional mode of minimizing pixel errors, and the generated countermeasure network not only can effectively retain the high-frequency information of the input original image, but also can generate a reconstructed image with high similarity through a perception optimization mode.
The method for verifying the human identity card can effectively solve the problem of image heterogeneity in the human identity card verification, by taking the human identity card verification of the second generation identity card as an example, a generated countermeasure network aiming at the super-resolution reconstruction of the second generation identity card can be obtained through deep training, natural light attribute information (such as the brightness, illumination, color and the like of each area of a human face) is compensated under the condition of keeping original identity card illumination information, and a high-resolution human face reconstructed image of the second generation identity card is output. And then, the face image is compared with the collected natural illumination face image, and the verification of the second-generation identity card can be effectively carried out by utilizing the existing face recognition technology.
It should be noted that, for the sake of simplicity, the foregoing method embodiments are described as a series of acts or combinations, but those skilled in the art should understand that the present invention is not limited by the described order of acts, as some steps may be performed in other orders or simultaneously according to the present invention.
The present invention also provides a device for testimonial verification, which can be used to execute the method for testimonial verification described above, based on the same idea as the method for testimonial verification in the above-described embodiments. For convenience of explanation, only the parts related to the embodiments of the present invention are shown in the schematic structural diagrams of the embodiments of the apparatus for testimony verification, and it will be understood by those skilled in the art that the illustrated structure does not constitute a limitation of the apparatus, and may include more or less components than those illustrated, or combine some components, or arrange different components.
Fig. 4 is a schematic structural diagram of a credential verification apparatus according to an embodiment of the present invention, and as shown in fig. 4, the credential verification apparatus according to the embodiment includes: a face image acquisition module 410, a face image reconstruction module 420, and a testimony verification module 430, each of which is described in detail as follows:
the face image collecting module 410 is used for acquiring a certificate face image in a certificate photo and collecting a natural light face image of a user.
The face image reconstruction module 420 is configured to input the certificate face image into a pre-trained generation countermeasure network, and obtain a reconstructed face image corresponding to the certificate face image according to an output of the generation countermeasure network; the generation countermeasure network is used for adding preset natural light attribute information to the input certificate face image, and the resolution of the output reconstructed face image is higher than that of the certificate face image.
The testimony verification module 430 is configured to compare the reconstructed face image with the natural light face image, and perform testimony verification according to a comparison result.
In an optional embodiment, the system further comprises a network training module for training generation of the countermeasure network. The network training module is specifically used for pre-training the generation countermeasure network based on the ImageNet database; and retraining the generated confrontation network after the pre-training based on a preset testimony sample library until the generated confrontation network meeting the preset condition is obtained.
Preferably, the generating countermeasure network comprises a generator network and a discriminator network; the generator network comprises 6 layers of residual error convolution network structures, wherein the first 3 layers are convolution layers, the second 3 layers are reverse convolution layers, and the last reverse convolution layer outputs a reconstructed face image. The discriminator network comprises a Light CNN residual network structure. The network structure can effectively accelerate the training process of generating the confrontation network and shorten the training time.
In an optional embodiment, the network training module comprises: a first training unit and a second training unit.
The first training unit is used for training a generator network in the generation countermeasure network, and the specific training mode comprises the following steps: acquiring a certificate photo sample and a natural light face image sample corresponding to the certificate photo sample from a certificate sample library, taking the certificate photo sample as the input of a generator network, and training the network parameters of the generator network based on a square loss function; the square loss function is a function of the pixel-based squared difference of the natural light face image samples and the reconstructed face image output by the generator network.
The second training unit is used for training a discriminator network in a generation countermeasure network, and the specific training mode comprises the following steps: taking the natural light face image sample and a reconstructed face image output by a generator network as the input of a discriminator network, and training the network parameters of the discriminator network and the network parameters of the generator network based on a perception loss function; the perceptual loss function is a function of the probability of discriminating the reconstructed face image output by the generator network into a real high-resolution image.
Preferably, the training process for generating the countermeasure network can be described with reference to the specific process shown in fig. 3 with reference to the above method embodiment.
It should be noted that, in the embodiment of the apparatus for testimony verification in the above example, because the contents of information interaction, execution process, and the like between the modules/units are based on the same concept as the foregoing method embodiment of the present invention, the technical effect brought by the contents is the same as the foregoing method embodiment of the present invention, and specific contents may refer to the description in the method embodiment of the present invention, and are not described again here.
In addition, in the above exemplary embodiments of the witness verification apparatus, the logical division of the functional modules is only an example, and in practical applications, the above functions may be distributed by different functional modules according to needs, for example, due to configuration requirements of corresponding hardware or due to convenience of implementation of software, that is, the internal structure of the witness verification apparatus is divided into different functional modules to complete all or part of the above described functions. The functional modules can be realized in a hardware mode or a software functional module mode.
It will be understood by those skilled in the art that all or part of the processes of the methods of the above embodiments may be implemented by a computer program, which is stored in a computer readable storage medium and sold or used as a stand-alone product. The program, when executed, may perform all or a portion of the steps of the embodiments of the methods described above. In addition, the storage medium may be provided in a computer device, and the computer device further includes a processor, and when the processor executes the program in the storage medium, all or part of the steps of the embodiments of the methods described above can be implemented. The storage medium may be a magnetic disk, an optical disk, a Read-only Memory (ROM), a Random Access Memory (RAM), or the like.
In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.
The above-described examples merely represent several embodiments of the present invention and should not be construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (10)
1. A method of witness verification, comprising:
acquiring a certificate face image in a certificate photo, and acquiring a natural light face image of a user; the natural light face image is a high-definition image on the front side;
inputting the certificate face image into a pre-trained generation countermeasure network, and obtaining a reconstructed face image corresponding to the certificate face image according to the output of the generation countermeasure network; the generation countermeasure network is used for adding preset natural light attribute information to the input certificate face image, filling up missing pixel parts of the certificate face image, and the resolution of the output reconstructed face image is higher than that of the certificate face image; the modality of the reconstructed face image obtained by the generation of the countermeasure network is consistent with the modality of the natural light face image, and the pixel size of the reconstructed face image is consistent with the pixel size of the natural light face image; the natural illumination attribute comprises brightness, illumination and/or color;
comparing the reconstructed face image with the natural light face image, and verifying the testimony according to the comparison result;
the certificate face image in the certificate photo is acquired, and the method comprises the following steps:
acquiring a certificate photo, carrying out face detection and alignment on the certificate photo, and cutting out a face area of a face five-sense-organ area to obtain a certificate face image;
comparing the reconstructed face image with the natural light face image, and verifying the testimony according to the comparison result, wherein the method comprises the following steps:
if the matching degree of the reconstructed face image and the natural light face image is greater than a set threshold value, judging that the testimony verification is passed; otherwise, the verification of the human identity card is judged to fail.
2. The method of witness verification according to claim 1, further comprising: training a generation of a countermeasure network, the step comprising:
pre-training a pairing defense network based on an ImageNet database;
retraining the generated pre-trained confrontation network based on a preset testimony sample library; the testimony sample library comprises a plurality of testimony photo samples and natural light face image samples corresponding to the testimony photo samples.
3. The method of witness verification of claim 2, wherein said generating a countermeasure network comprises a generator network and a discriminator network;
the generator network comprises 6 layers of residual error convolution network structures, wherein the first 3 layers are convolution layers, the second 3 layers are reverse convolution layers, and the last reverse convolution layer outputs a reconstructed face image;
the discriminator network comprises a Light CNN residual network structure.
4. The method of witness verification according to claim 2, wherein retraining the pre-trained generated confrontation network based on a pre-set witness sample library comprises:
generating generator network training in the countermeasure network based on a preset testimony sample library, and specifically comprising the following steps:
acquiring a certificate photo sample and a natural light face image sample corresponding to the certificate photo sample from a certificate sample library, acquiring the certificate face image sample from the certificate photo sample, taking the certificate face image sample as the input of a generator network, and training the network parameters of the generator network based on a square loss function; the square loss function is a function of the square difference between the natural light face image sample and the reconstructed face image output by the generator network based on the pixel;
the generation of the discriminant network training in the countermeasure network based on the preset testimony sample library specifically comprises the following steps:
taking the natural light face image sample and a reconstructed face image output by a generator network as the input of a discriminator network, and training the network parameters of the discriminator network and the network parameters of the generator network based on a perception loss function; the perceptual loss function is a function of the probability that the reconstructed face image output by the generator network is discriminated as a real natural light face image by the discriminator network.
5. The method of witness verification as claimed in claim 4 wherein the generator network corresponds to a functional model of
Training an objective function of network parameters of the generator network
Comprises the following steps:
theta denotes a network parameter for generating a countermeasure network, thetaGNetwork parameters representing the generator network,/sAs a function of the squared loss, N is the total number of samples participating in the training certificate photo,
representing the natural light face image sample I corresponding to the nth certificate reference sample participating in the trainingy，
Certificate face image sample I in nth certificate photo sample for representing trainingx。
6. The method of witness verification as claimed in claim 5 wherein the function model corresponding to the network of discriminators is
Training the discriminatorThe objective function of the network is:
wherein, thetaDNetwork parameters representing the arbiter network, Ε being a mathematical expectation, Iy～pdata(Iy) Representing natural light face image samples IyProbability distribution satisfying high resolution image is pdata(Iy)；Ix～pG(Ix) Representing a sample of a document face image IxSatisfies the probability distribution of the generator as pG(Iy) (ii) a log represents a logarithmic operation;
is to use a natural light face image sample IyThe probability of distinguishing as a real natural light face image,
representation arbiter network
Generator network
Output reconstructed image
And judging the probability of the face image as real natural light.
7. An apparatus for witness verification, comprising:
the face image acquisition module is used for acquiring a certificate face image in the certificate photo and acquiring a natural light face image of a user; the natural light face image is a high-definition image on the front side;
the face image reconstruction module is used for inputting the certificate face image into a pre-trained generation countermeasure network and obtaining a reconstructed face image corresponding to the certificate face image according to the output of the generation countermeasure network; the generation countermeasure network is used for adding preset natural light attribute information to the input certificate face image, filling up missing pixel parts of the certificate face image, and the resolution of the output reconstructed face image is higher than that of the certificate face image; the modality of the reconstructed face image obtained by the generation of the countermeasure network is consistent with the modality of the natural light face image, and the pixel size of the reconstructed face image is consistent with the pixel size of the natural light face image; the natural illumination attribute comprises brightness, illumination and/or color;
the testimony verification module is used for comparing the reconstructed face image with the natural light face image and verifying the testimony according to the comparison result;
the face image acquisition module is also used for acquiring a certificate photo, carrying out face detection and alignment on the certificate photo, and cutting out a face area of a face facial feature area to obtain a certificate face image;
the testimony verification module is also used for judging that the testimony verification is passed if the matching degree of the reconstructed face image and the natural light face image is greater than a set threshold value; otherwise, the verification of the human identity card is judged to fail.
8. The apparatus for testimony verification according to claim 7, further comprising a network training module for pre-training the generation countermeasure network based on the ImageNet database; retraining the generated confrontation network after the pre-training based on a preset testimony sample library until the generated confrontation network meeting the preset conditions is obtained; the testimony sample library comprises a plurality of testimony photo samples and natural light face image samples corresponding to the testimony photo samples.
9. The apparatus for witness verification as claimed in claim 8, wherein said network training module comprises:
the system comprises a first training unit, a second training unit and a third training unit, wherein the first training unit is used for training a generator network in the generation countermeasure network, and specifically comprises the steps of acquiring a certificate reference sample and a natural light face image sample corresponding to the certificate reference sample from a certificate sample library, obtaining the certificate face image sample from the certificate reference sample, using the certificate face image sample as the input of the generator network, and training the network parameters of the generator network based on a square loss function; the square loss function is a function of the square difference between the natural light face image sample and the reconstructed face image output by the generator network based on the pixel;
the second training unit is used for training a discriminator network in a generation countermeasure network, and specifically comprises: taking the natural light face image sample and a reconstructed face image output by a generator network as the input of a discriminator network, and training the network parameters of the discriminator network and the network parameters of the generator network based on a perception loss function; the perceptual loss function is a function of the probability that the reconstructed face image output by the generator network is discriminated as a real natural light face image by the discriminator network.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 6.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710581244.7A CN107423700B (en) | 2017-07-17 | 2017-07-17 | Method and device for verifying testimony of a witness |
| PCT/CN2018/093784 WO2019015466A1 (en) | 2017-07-17 | 2018-06-29 | Method and apparatus for verifying person and certificate |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710581244.7A CN107423700B (en) | 2017-07-17 | 2017-07-17 | Method and device for verifying testimony of a witness |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107423700A CN107423700A (en) | 2017-12-01 |
| CN107423700B true CN107423700B (en) | 2020-10-20 |
Family
ID=60429889
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710581244.7A Active CN107423700B (en) | 2017-07-17 | 2017-07-17 | Method and device for verifying testimony of a witness |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN107423700B (en) |
| WO (1) | WO2019015466A1 (en) |
Families Citing this family (48)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107423700B (en) * | 2017-07-17 | 2020-10-20 | 广州广电卓识智能科技有限公司 | Method and device for verifying testimony of a witness |
| CN107578017B (en) * | 2017-09-08 | 2020-11-17 | 百度在线网络技术（北京）有限公司 | Method and apparatus for generating image |
| US11250329B2 (en) | 2017-10-26 | 2022-02-15 | Nvidia Corporation | Progressive modification of generative adversarial neural networks |
| US11263525B2 (en) | 2017-10-26 | 2022-03-01 | Nvidia Corporation | Progressive modification of neural networks |
| CN108090905B (en) * | 2018-01-05 | 2019-02-05 | 清华大学 | The judgment method and system of producing line exception |
| CN108280413B (en) * | 2018-01-17 | 2022-04-19 | 百度在线网络技术（北京）有限公司 | Face recognition method and device |
| CN108416326B (en) * | 2018-03-27 | 2021-07-16 | 百度在线网络技术（北京）有限公司 | Face recognition method and device |
| CN110263603B (en) * | 2018-05-14 | 2021-08-06 | 桂林远望智能通信科技有限公司 | Face recognition method and device based on central loss and residual error visual simulation network |
| CN109711364A (en) * | 2018-12-29 | 2019-05-03 | 成都视观天下科技有限公司 | A kind of facial image super-resolution reconstruction method, device and computer equipment |
| CN111583096B (en) * | 2019-02-15 | 2024-07-16 | 北京京东乾石科技有限公司 | Picture processing method and device, electronic equipment and computer readable medium |
| CN111753595A (en) * | 2019-03-29 | 2020-10-09 | 北京市商汤科技开发有限公司 | Living body detection method and apparatus, device, and storage medium |
| CN110097609B (en) * | 2019-04-04 | 2022-11-29 | 浙江凌迪数字科技有限公司 | Sample domain-based refined embroidery texture migration method |
| CN110163114B (en) * | 2019-04-25 | 2022-02-15 | 厦门瑞为信息技术有限公司 | Method and system for analyzing face angle and face blurriness and computer equipment |
| CN110097543B (en) * | 2019-04-25 | 2023-01-13 | 东北大学 | Hot-rolled strip steel surface defect detection method based on generation type countermeasure network |
| CN110119746B (en) * | 2019-05-08 | 2021-11-30 | 北京市商汤科技开发有限公司 | Feature recognition method and device and computer readable storage medium |
| CN110288537A (en) * | 2019-05-20 | 2019-09-27 | 湖南大学 | Facial image complementing method based on the depth production confrontation network from attention |
| CN110457994B (en) * | 2019-06-26 | 2024-05-10 | 平安科技（深圳）有限公司 | Face image generation method and device, storage medium and computer equipment |
| CN110443746B (en) * | 2019-07-25 | 2023-07-14 | 创新先进技术有限公司 | Picture processing method and device based on generation countermeasure network and electronic equipment |
| CN110517195B (en) * | 2019-07-26 | 2022-12-06 | 西安电子科技大学 | Unsupervised SAR image denoising method |
| CN112330526B (en) * | 2019-08-05 | 2024-02-09 | 深圳Tcl新技术有限公司 | Training method of face conversion model, storage medium and terminal equipment |
| CN110580682A (en) * | 2019-09-16 | 2019-12-17 | 电子科技大学 | Countermeasure network seismic data super-resolution reconstruction method based on optimization generation |
| CN111062290B (en) * | 2019-12-10 | 2023-04-07 | 西北大学 | Method and device for constructing Chinese calligraphy style conversion model based on generation confrontation network |
| CN111161200A (en) * | 2019-12-22 | 2020-05-15 | 天津大学 | Human body posture migration method based on attention mechanism |
| CN111476749B (en) * | 2020-04-03 | 2023-02-28 | 陕西师范大学 | Face repairing method for generating confrontation network in guiding mode based on face key points |
| CN111476717B (en) * | 2020-04-07 | 2023-03-24 | 西安电子科技大学 | Face image super-resolution reconstruction method based on self-attention generation countermeasure network |
| CN111553208A (en) * | 2020-04-15 | 2020-08-18 | 上海携程国际旅行社有限公司 | Identity recognition method, system, device and medium based on image of people and certificate integration |
| CN111597978B (en) * | 2020-05-14 | 2023-04-07 | 公安部第三研究所 | Method for automatically generating pedestrian re-identification picture based on StarGAN network model |
| CN111754478A (en) * | 2020-06-22 | 2020-10-09 | 怀光智能科技（武汉）有限公司 | Unsupervised domain adaptation system and unsupervised domain adaptation method based on generation countermeasure network |
| CN113761997B (en) * | 2020-08-27 | 2024-04-09 | 北京沃东天骏信息技术有限公司 | Method and device for generating semi-occlusion face recognition device |
| CN112070145B (en) * | 2020-09-04 | 2024-05-28 | 世纪易联(北京)科技有限公司 | Freshness attribute migration method of fruit image based on countermeasure network |
| CN112102186B (en) * | 2020-09-07 | 2024-04-05 | 河海大学 | Real-time enhancement method for underwater video image |
| CN112053303B (en) * | 2020-09-08 | 2024-04-05 | 河海大学 | Video image real-time enhancement method for underwater AUV |
| CN112508782B (en) * | 2020-09-10 | 2024-04-26 | 浙江大华技术股份有限公司 | Training method of network model, and super-resolution reconstruction method and device of face image |
| CN112233017B (en) * | 2020-10-28 | 2023-09-26 | 中国科学院合肥物质科学研究院 | Method for enhancing pathological face data based on generation countermeasure network |
| CN112598125B (en) * | 2020-11-25 | 2024-04-30 | 西安科技大学 | Handwriting digital generation method based on dual-discriminant weighting generation countermeasure network |
| CN114596236B (en) * | 2020-12-04 | 2024-09-10 | 国网智能科技股份有限公司 | Method and system for enhancing low-illumination image of closed cavity |
| CN112598611A (en) * | 2020-12-16 | 2021-04-02 | 阳光保险集团股份有限公司 | Method and device for synthesizing and identifying embossed bank card number image |
| CN112668623B (en) * | 2020-12-22 | 2024-04-16 | 中国铁道科学研究院集团有限公司 | Method and device for generating binaural pin defect sample based on generation countermeasure network |
| CN112818764B (en) * | 2021-01-15 | 2023-05-02 | 西安交通大学 | Low-resolution image facial expression recognition method based on feature reconstruction model |
| CN113034393A (en) * | 2021-03-25 | 2021-06-25 | 北京百度网讯科技有限公司 | Photo repairing method, device, equipment and storage medium |
| CN113378721B (en) * | 2021-06-11 | 2023-08-18 | 西安电子科技大学 | Symmetrical and local discrimination-based face correction method and system for generating countermeasure |
| CN113723174B (en) * | 2021-07-09 | 2024-06-28 | 北京工业大学 | Face image super-resolution restoration and reconstruction method and system based on generation countermeasure network |
| CN113705400B (en) * | 2021-08-18 | 2023-08-15 | 中山大学 | Single-mode face living body detection method based on multi-mode face training |
| CN113762180B (en) * | 2021-09-13 | 2023-09-01 | 中国科学技术大学 | Training method and system for human body activity imaging based on millimeter wave radar signals |
| CN113780534B (en) * | 2021-09-24 | 2023-08-22 | 北京字跳网络技术有限公司 | Compression method, image generation method, device, equipment and medium of network model |
| CN114067399B (en) * | 2021-11-16 | 2024-03-15 | 桂林电子科技大学 | Face reconstruction and recognition method for non-matching scene |
| CN114913086B (en) * | 2022-05-05 | 2023-05-02 | 上海云思智慧信息技术有限公司 | Face image quality enhancement method based on generation countermeasure network |
| CN115661002B (en) * | 2022-12-14 | 2023-04-21 | 北京数慧时空信息技术有限公司 | Multi-temporal remote sensing data restoration method based on GAN |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106683048A (en) * | 2016-11-30 | 2017-05-17 | 浙江宇视科技有限公司 | Image super-resolution method and image super-resolution equipment |
| CN106951867A (en) * | 2017-03-22 | 2017-07-14 | 成都擎天树科技有限公司 | Face identification method, device, system and equipment based on convolutional neural networks |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7127087B2 (en) * | 2000-03-27 | 2006-10-24 | Microsoft Corporation | Pose-invariant face recognition system and process |
| CN101694691A (en) * | 2009-07-07 | 2010-04-14 | 北京中星微电子有限公司 | Method and device for synthesizing facial images |
| CN105844202A (en) * | 2015-01-12 | 2016-08-10 | 芋头科技（杭州）有限公司 | Image recognition system and method |
| CN106845471A (en) * | 2017-02-20 | 2017-06-13 | 深圳市唯特视科技有限公司 | A kind of vision significance Forecasting Methodology based on generation confrontation network |
| CN106845449A (en) * | 2017-02-22 | 2017-06-13 | 浙江维尔科技有限公司 | A kind of image processing apparatus, method and face identification system |
| CN106952229A (en) * | 2017-03-15 | 2017-07-14 | 桂林电子科技大学 | Image super-resolution rebuilding method based on the enhanced modified convolutional network of data |
| CN107423700B (en) * | 2017-07-17 | 2020-10-20 | 广州广电卓识智能科技有限公司 | Method and device for verifying testimony of a witness |
-
2017
- 2017-07-17 CN CN201710581244.7A patent/CN107423700B/en active Active
-
2018
- 2018-06-29 WO PCT/CN2018/093784 patent/WO2019015466A1/en active Application Filing
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106683048A (en) * | 2016-11-30 | 2017-05-17 | 浙江宇视科技有限公司 | Image super-resolution method and image super-resolution equipment |
| CN106951867A (en) * | 2017-03-22 | 2017-07-14 | 成都擎天树科技有限公司 | Face identification method, device, system and equipment based on convolutional neural networks |
Non-Patent Citations (1)
| Title |
|---|
| Generative Face Completion;LI Yi Jun等;《Computer Vision and Pattern Recognition》;20170419;第3914页公式（1） * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107423700A (en) | 2017-12-01 |
| WO2019015466A1 (en) | 2019-01-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107423700B (en) | Method and device for verifying testimony of a witness | |
| Quan et al. | Distinguishing between natural and computer-generated images using convolutional neural networks | |
| CN110348319B (en) | Face anti-counterfeiting method based on face depth information and edge image fusion | |
| Cui et al. | Identifying materials of photographic images and photorealistic computer generated graphics based on deep CNNs. | |
| CN109376603A (en) | A kind of video frequency identifying method, device, computer equipment and storage medium | |
| CN113128271A (en) | Counterfeit detection of face images | |
| CN110414350A (en) | The face false-proof detection method of two-way convolutional neural networks based on attention model | |
| CN114842524B (en) | Face false distinguishing method based on irregular significant pixel cluster | |
| CN109670559A (en) | Recognition methods, device, equipment and the storage medium of handwritten Chinese character | |
| CN112949469A (en) | Image recognition method, system and equipment for face tampered image characteristic distribution | |
| CN116340887B (en) | Multi-mode false news detection method and system | |
| CN106529395A (en) | Signature image recognition method based on deep brief network and k-means clustering | |
| CN112580445A (en) | Human body gait image visual angle conversion method based on generation of confrontation network | |
| WO2022156214A1 (en) | Liveness detection method and apparatus | |
| CN113628297A (en) | COVID-19 deep learning diagnosis system based on attention mechanism and transfer learning | |
| CN109117810A (en) | Fatigue driving behavioral value method, apparatus, computer equipment and storage medium | |
| CN116229528A (en) | Living body palm vein detection method, device, equipment and storage medium | |
| CN115294638A (en) | Iris identification system deployment method based on FPGA, iris identification method and system | |
| CN113095158A (en) | Handwriting generation method and device based on countermeasure generation network | |
| Levine et al. | Face recognition subject to variations in facial expression, illumination and pose using correlation filters | |
| Zhao et al. | A transferable anti-forensic attack on forensic CNNs using a generative adversarial network | |
| CN112287724A (en) | Finger vein identity recognition system based on transfer learning and GAN network | |
| CN103942545A (en) | Method and device for identifying faces based on bidirectional compressed data space dimension reduction | |
| CN117935375A (en) | Anti-fake method for handwriting electronic signature | |
| CN116469177A (en) | Living body target detection method with mixed precision and training method of living body detection model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP03 | Change of name, title or address |
Address after: No. 001-030, Yuntong space, office building, No. 9, Kelin Road, Science City, Guangzhou hi tech Industrial Development Zone, Guangzhou, Guangdong 510000 Patentee after: GRG TALLY-VISION I.T. Co.,Ltd. Country or region after: China Patentee after: Guangdian Yuntong Group Co.,Ltd. Address before: No. 001-030, Yuntong space, office building, No. 9, Kelin Road, Science City, Guangzhou hi tech Industrial Development Zone, Guangzhou, Guangdong 510000 Patentee before: GRG TALLY-VISION I.T. Co.,Ltd. Country or region before: China Patentee before: GRG BANKING EQUIPMENT Co.,Ltd. |
|
| CP03 | Change of name, title or address | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240201 Address after: No. 001-030, Yuntong space, office building, No. 9, Kelin Road, Science City, Guangzhou hi tech Industrial Development Zone, Guangzhou, Guangdong 510000 Patentee after: GRG TALLY-VISION I.T. Co.,Ltd. Country or region after: China Address before: No. 001-030, Yuntong space, office building, No. 9, Kelin Road, Science City, Guangzhou hi tech Industrial Development Zone, Guangzhou, Guangdong 510000 Patentee before: GRG TALLY-VISION I.T. Co.,Ltd. Country or region before: China Patentee before: Guangdian Yuntong Group Co.,Ltd. |
|
| TR01 | Transfer of patent right |