CN106599872A - Method and equipment for verifying living face images - Google Patents
Method and equipment for verifying living face images Download PDFInfo
- Publication number
- CN106599872A CN106599872A CN201611204264.4A CN201611204264A CN106599872A CN 106599872 A CN106599872 A CN 106599872A CN 201611204264 A CN201611204264 A CN 201611204264A CN 106599872 A CN106599872 A CN 106599872A
- Authority
- CN
- China
- Prior art keywords
- image
- living body
- body faces
- attack signature
- characteristic area
- 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
- G06V40/168—Feature extraction; Face representation
-
- 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
- G06V40/45—Detection of the body part being alive
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Human Computer Interaction (AREA)
- Oral & Maxillofacial Surgery (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Collating Specific Patterns (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a method and equipment for verifying living face images. The method comprises steps that an image of at least one characteristic region of face images is acquired; whether the image of each characteristic region comprises attack characteristics is determined; statistics of the quantity of the images of the characteristic regions comprising attack characteristics is carried out; when the quantity of the images of the characteristic regions comprising the attack characteristics is greater than a preset quantity, a face image is determined to be a non-living face image, when the quantity of the images of the characteristic regions comprising the attack characteristics is not greater than the preset quantity, the face image is determined to be a living face image. The method is advantaged in that special actions of a user are not required, user experience is improved, and thereby attack of masks, photographs, copying or computer synthesis aiming at living face identification can be effectively prevented.
Description
Technical field
Embodiment of the disclosure is related to a kind of method and apparatus for verifying living body faces image.
Background technology
The authentication system for being currently based on living body faces is used widely in such as safety-security area.With being based on
The popularization of the authentication system of living body faces, the method for having derived some malicious attack living body faces authentications.
In a kind of authentication system based on living body faces, can based on the facial image of current shooting with deposit in advance
Comparison between the human face photo of storage, carries out authentication.However, working as this being based on will be placed in by the photo of counterfeiter
When before the photographic head in the authentication system of living body faces, this authentication system based on living body faces can be by using
Family authentication.In other words, malicious user can using malicious attack (that is, photo attack) is carried out by the photo of counterfeiter,
This authentication system based on living body faces can not resist photo attack.
Attack for above-mentioned photo, the above-mentioned authentication system based on living body faces is improved.Changing
In the authentication system based on living body faces for entering, above-mentioned photograph is effectively coped with by whether inspection face has fine movement
Piece is attacked.Further, can require that user carries out required movement, so as to strengthen the authentication system based on living body faces
Attack tolerant.It is, however, required that user makes required movement coordinates authentication so that Consumer's Experience is poor and lose time.More
Further, there is the software that can simulate human face expression, therefore, traditional authentication system based on living body faces is easy
It is under attack.
The content of the invention
Embodiment of the disclosure provides a kind of method for verifying living body faces image, including:Obtain in facial image
The image of at least one characteristic area;Whether attack signature is included in judging the image of each characteristic area;Statistics includes
The quantity of the image of the characteristic area of the attack signature;When including the attack signature characteristic area image quantity it is big
When predetermined number, judge that the facial image is not living body faces image, when the characteristic area including the attack signature
When the quantity of image is less than or equal to predetermined number, judge that the facial image is living body faces image.
For example, in the method for verifying living body faces image that the embodiment of the present disclosure is provided, the attack signature bag
Include pore disappearance, wrinkle disappearance, the muscle movement that shade is lacked, non-natural reflective, living body faces cannot be made, mosaic, saw
One kind or its combination in tooth and moire fringes.
For example, in the method for verifying living body faces image that the embodiment of the present disclosure is provided, in the facial image
The image of at least one characteristic area is chosen and/or is randomly selected according to priori.
For example, in the method for verifying living body faces image that the embodiment of the present disclosure is provided, described each institute of judgement
Whether include including attack signature in stating the image of characteristic area:The image regulation of each characteristic area is encoded to into spy
Levy vector;Judge in the characteristic vector, whether include attack signature vector using the neutral net trained.
For example, in the method for verifying living body faces image that the embodiment of the present disclosure is provided, the god for having trained
Jing networks are obtained by following operation:Acquisition includes pore disappearance, wrinkle disappearance, shade disappearance, non-natural reflective, living body faces
The training image of one kind or its combination in muscle movement, mosaic, sawtooth and the moire fringes that cannot make;The training is schemed
As regularization is encoded to training feature vector;The neutral net is trained using the training feature vector.
For example, in the method for verifying living body faces image that the embodiment of the present disclosure is provided, described utilization is trained
Neutral net judge the characteristic vector in whether include including attack signature vector:Using the neutral net trained
Determine that the characteristic vector includes the probability of the attack signature vector;When the probability is more than predetermined probabilities, institute is judged
Stating characteristic vector includes the attack signature vector.
For example, in the method for verifying living body faces image that the embodiment of the present disclosure is provided, obtaining facial image
In at least one characteristic area image before, also including obtaining the facial image.
For example, the embodiment of the present disclosure provide for the method for verifying living body faces image, also including output living body faces
Image authentication result.
Embodiment of the disclosure also provides a kind of equipment for verifying living body faces image, including:At one or more
Reason device;One or more memorizeies;And storage computer program instructions in which memory, in the computer program
Instruction performs following steps when being run by the processor:Obtain the image of at least one characteristic area in facial image;Judge
Whether attack signature is included in the image of each characteristic area;Statistics includes the image of the characteristic area of the attack signature
Quantity;When the quantity of the image of the characteristic area including the attack signature is more than predetermined number, the face figure is judged
As being not living body faces image, when the quantity of the image of the characteristic area including the attack signature is less than or equal to predetermined number
When, judge that the facial image is living body faces image.
For example, in the equipment for verifying living body faces image that the embodiment of the present disclosure is provided, the attack signature bag
Include pore disappearance, wrinkle disappearance, the muscle movement that shade is lacked, non-natural reflective, living body faces cannot be made, mosaic, saw
One kind or its combination in tooth and moire fringes.
For example, in the equipment for verifying living body faces image that the embodiment of the present disclosure is provided, in the facial image
The image of at least one characteristic area is chosen and/or is randomly selected according to priori.
For example, in the equipment for verifying living body faces image that the embodiment of the present disclosure is provided, described each institute of judgement
Whether include including attack signature in stating the image of characteristic area:By the image regulation of characteristic area coding be characterized to
Amount;Judge in the characteristic vector, whether include attack signature vector using the neutral net trained.
For example, in the equipment for verifying living body faces image that the embodiment of the present disclosure is provided, the god for having trained
Jing networks are obtained by following operation:Acquisition includes pore disappearance, wrinkle disappearance, shade disappearance, non-natural reflective, living body faces
The training image of one kind or its combination in muscle movement, mosaic, sawtooth and the moire fringes that cannot make;The training is schemed
As regularization is encoded to training feature vector;The neutral net is trained using the training feature vector.
For example, in the equipment for verifying living body faces image that the embodiment of the present disclosure is provided, described utilization is trained
Neutral net judge the characteristic vector in whether include including attack signature vector:Using the neutral net trained
Determine that the characteristic vector includes the probability of the attack signature vector;When the probability is more than predetermined probabilities, institute is judged
Stating characteristic vector includes the attack signature vector.
For example, the equipment for verifying living body faces image that the embodiment of the present disclosure is provided, also including image acquiring device,
It is configured to, in facial image is obtained before the image of at least one characteristic area, obtain the facial image.
For example, the equipment for verifying living body faces image that the embodiment of the present disclosure is provided, also including output device, is matched somebody with somebody
It is set to output living body faces image authentication result.
Embodiment of the disclosure also provides a kind of equipment for verifying living body faces image, including:Characteristic area is extracted
Device, is configured to obtain the image of at least one characteristic area in facial image;Attack signature judgment means, are configured to sentence
Whether attack signature is included in the image of disconnected each characteristic area;And statistics and judgment means, it is configured to statistics bag
The quantity of the image of the characteristic area of the attack signature is included, when the quantity of the image of the characteristic area including the attack signature
During more than predetermined number, judge that the facial image is not living body faces image, when the characteristic area including the attack signature
Image quantity be less than or equal to predetermined number when, judge that the facial image is living body faces image.
For example, the method and apparatus for verifying living body faces image that the embodiment of the present disclosure is provided does not need user to make
Specific action, improves Consumer's Experience, and can more effectively defend mask, photo, reproduction or computer synthesis etc. to be directed to
The attack of living body faces identification.
Description of the drawings
In order to be illustrated more clearly that the technical scheme of the embodiment of the present disclosure, below will be in embodiment or description of Related Art
The required accompanying drawing for using is briefly described, it should be apparent that, drawings in the following description merely relate to some of the disclosure
Embodiment, restriction not of this disclosure.
Fig. 1 is shown for realizing the example for verifying the method and apparatus of living body faces image of the embodiment of the present disclosure
The schematic block diagram of property electronic equipment;
Fig. 2 is the indicative flowchart for verifying the method for living body faces image according to the embodiment of the present disclosure;
Fig. 3 is
The no indicative flowchart including attack signature;
Fig. 4 is whether to include attack signature vector in the neutral net judging characteristic vector that the utilization shown in Fig. 3 has been trained
Indicative flowchart;
Fig. 5 is the schematic block diagram for verifying the equipment of living body faces image according to the embodiment of the present disclosure;And
Fig. 6 is the schematic block diagram of attack signature judgment means in Fig. 5.
Specific embodiment
Below in conjunction with accompanying drawing, the technical scheme in the embodiment of the present disclosure is clearly and completely described with reference to attached
The non-limiting example embodiment for illustrating in figure and describing in detail in the following description, the example for being more fully described below the disclosure are implemented
Example and their various features and Advantageous details.It should be noted that the feature illustrated in figure is not required to be drawn to scale.This
The open description for eliminating known materials, component and Technology, so as to the example embodiment for not making the disclosure is obscured.It is given
Example be only intended to the enforcement for being conducive to understanding disclosure example embodiment, and further enable those skilled in the art real
Apply example embodiment.Thus, these examples are understood not to the restriction of the scope of embodiment of this disclosure.
Unless otherwise specifically defined, the disclosure using technical term or scientific terminology should be disclosure art
The ordinary meaning understood by the interior personage with general technical ability." first ", " second " and similar word used in the disclosure
Language is not offered as any order, quantity or importance, and is used only to distinguish different ingredients.Additionally, in the disclosure
In each embodiment, same or similar reference number represents same or similar component.
The method and apparatus for verifying living body faces image that the embodiment of the present disclosure is provided does not need user to make specific
Action, improve Consumer's Experience, and can more effectively defend mask, photo, reproduction or computer synthesis etc. for live body
The attack of recognition of face.
First, with reference to Fig. 1 come describe for realize the embodiment of the present disclosure for realize checking living body faces image side
The example electronic device 100 of method and equipment.
For example, as shown in figure 1, electronic equipment 100 includes one or more processors 102, one or more storage devices
104th, image acquiring device 106 and output device 108, these components pass through the connection of bus system 110 and/or other forms
Mechanism's (such as wireless telecommunication system, Fig. 1 not shown in) interconnection.It should be noted that the component of electronic equipment 100 shown in Fig. 1 and
Structure is illustrative, and not restrictive, and as needed, electronic equipment can also have other assemblies and structure.
For example, processor 102 can be CPU (CPU) or there is data-handling capacity and/or instruction to hold
The processing unit of the other forms of row ability, and can be with other components in control electronics 100 performing desired work(
Energy;The CPU can for example be monokaryon or multi-core CPU etc., for example, can be based on X86-based or ARM frameworks etc., for example, can be
Parallel computation mode or serial computing mode etc., embodiment of the disclosure not limited to this.
For example, storage device 104 can include one or more computer programs, and computer program can be wrapped
Include various forms of computer-readable recording mediums, such as volatile memory and/or nonvolatile memory.Volatile storage
Device can for example include random access memory (RAM) and/or cache memory (cache) etc..Nonvolatile memory
Read only memory (ROM), hard disk, flash memory etc. can for example be included.Can store on computer-readable recording medium one or
Multiple computer program instructions, processor 102 can be instructed with operation program, realizing in the embodiment of the present disclosure hereafter (by
Reason device realize) function and/or other desired functions.Various answering can also be stored in a computer-readable storage medium
With program and various data, such as various data of facial image and application program use and/or generation etc..
For example, image acquiring device 106 can shoot the desired image of user (such as photo, video etc.), and by institute
The image of shooting is stored in storage device 104 and uses for other components.Image acquiring device 106 can be visible ray or red
Outer image acquiring device, the image which obtains include but is not limited to gray level image, coloured image or infrared image etc..For example,
Image acquiring device 106 can include one or more imageing sensors, for example, when image acquiring device 106 includes two figures
During as sensor, one can be used for obtaining coloured image and another can obtain black white image, or one is used to collect
Image color information and another is responsible for collecting brightness, the detailed information such as profile etc..Image acquiring device 106 is, for example, high definition figure
As acquisition device, more detailed information in acquired image, can be included.
Image acquiring device 106 can be arranged together (such as same for constituting with processor 102 and storage device 104
One equipment, such as mobile terminal), or can discretely arrange with processor 102 and storage device 104, such as processor
102 and storage device 104 can be implemented as server, such as centralized server or distributed server, or Cloud Server.
For example, output device 108 can export various information (such as image or sound) to outside (such as user), and
Can be including one or more in display, speaker etc..
For example, electronic equipment 100 can also include input equipment (not shown in figure 1), and input equipment can be that user uses
Carry out the device of input instruction, and can be including one or more in keyboard, mouse, mike and touch screen etc..Instruction example
Referred to using the instruction of 106 shooting image of image acquiring device, or the request for carrying out authentication using electronic equipment 100 in this way
Order.
For example, the electronic equipment 100 for the method and apparatus of realization checking living body faces image may be implemented as all
Such as the mobile terminal such as smart mobile phone, panel computer, wearable device, can be applied to the gate control system or finance neck of safety-security area
The payment system in domain.However, embodiment of the disclosure not limited to this, the members in electronic equipment 100 can be it is fixed,
Such as image acquiring device 106 can be fixed on predetermined position, processor 102, storage device 104 and output device 108
Can be mounted together and can include desk-top calculating as independent fixed terminal or mobile terminal, fixed terminal
The electronic equipment of the usual not shift position such as machine.Image acquiring device 106 and independent fixed terminal or mobile terminal can lead to
Cross wired or wireless way communication.
For example, for realize the embodiment of the present disclosure for realize checking living body faces image method and apparatus electronics
Equipment 100 can not also include image acquiring device, but receive the image of miscellaneous equipment collection and carry out subsequent treatment.
Below, with reference to Fig. 2 describing according to the embodiment of the present disclosure for the method for realizing checking living body faces image.
For example, embodiment of the disclosure provides a kind of method for verifying living body faces image, as shown in Fig. 2 the party
Method comprises the steps:
Step S220:Obtain the image of at least one characteristic area in facial image;
Step S230:Whether attack signature is included in judging the image of each characteristic area;
Step S240:Statistics includes the quantity of the image of the characteristic area of attack signature;
Step S250:When the quantity of the image of the characteristic area including attack signature is more than predetermined number, step is proceeded to
S260, when the quantity of the image of the characteristic area including attack signature is less than or equal to predetermined number, proceeds to step S260 ';
Step S260:Judge that facial image is not living body faces image;
Step S260 ':Judge that facial image is living body faces image.
For example, as shown in Fig. 2 embodiment of the disclosure provide for the method for verifying living body faces image, obtaining
In facial image before the image of at least one characteristic area (step S220), step S210 can also be included:Obtain face figure
Picture.
For example, the facial image for obtaining in step S210 is not limited to the face image of a people, or multiple
The face image of people.
For example, as shown in Fig. 2 embodiment of the disclosure provide for the method for verifying living body faces image, can be with
Including step S270:Output living body faces image authentication result.
For example, the dotted line frame in Fig. 2 represents that step S210 and step S270 are optional step.
For example, the step of the method for verifying living body faces image that the embodiment of the present disclosure is provided in S220, face
In image, the image of at least one characteristic area can be chosen and/or be randomly selected according to priori.For example, characteristic area
Image includes but is not limited to forehead image, eye image, buccal image, nose image, lip image etc..For example, priori is known
Knowledge can be obtained by pre-stage test, can improve discrimination, for example, nose according to the image in priori selected characteristic region
Image is used to judge whether that the discrimination of shade disappearance is higher.For example, the image of a part of characteristic area can be according to elder generation
Knowledge selection is tested, the image of another part characteristic area can be randomly selected, so can be prevented for selecting according to priori
The attack in the characteristic image region for taking.
For example, attack in S230 the step of the method for verifying living body faces image that the embodiment of the present disclosure is provided
Feature includes pore disappearance, wrinkle disappearance, the muscle movement that shade is lacked, non-natural reflective, living body faces cannot be made, Marseille
Gram, the one kind in sawtooth and moire fringes or its combination.It should be noted that the attack signature of the embodiment of the present disclosure includes but not office
It is limited to features described above.
For example, as shown in figure 3, in the method for verifying living body faces image that the embodiment of the present disclosure is provided, judging
Whether comprise the steps including attack signature (step S230 in Fig. 2) in the image of each characteristic area:
Step S231:The image regulation of each characteristic area is encoded to into characteristic vector;
Step S232:Using whether vectorial including attack signature in the neutral net judging characteristic vector trained.
For example, in step S231, unessential characteristics of image can be weakened using regularization coding, it is special from multiple images
Levy the middle data volume for generating important characteristic vector, reducing characteristic vector.
For example, in step S232, the neutral net trained can be obtained by following operation:Acquisition includes that pore lacks
Lose, wrinkle disappearance, shade disappearance, non-natural reflective, living body faces cannot make muscle movement, mosaic, sawtooth and mole
The training image of one kind or its combination in stricture of vagina;Training image regularization is encoded to into training feature vector;And using training
Characteristic vector trains neutral net.
For example, neutral net is, for example, deep neural network.Again for example, neutral net is convolutional neural networks.The nerve
Network can have but be not limited to following structure:Input layer, receives characteristic vector;Convolutional layer, the characteristic vector to being input into are carried out
Smooth and extraction feature;Feedback layer, the feature to being input into further are extracted;Full articulamentum, it is abstract for carrying out to feature;
Output layer, for exporting final judged result.
For example, for known attack face and real human face, using neutral net to wherein believing comprising a large amount of details
The region of breath is learnt.For position same on face, the detailed information of real human face is truly enriched very much, and attacks people
Can there is excalation so as to embody attack signature in the detailed information of face.Neutral net can be carried out effectively to this difference
Catch, these information are carried out into the description of profound level using complicated mathematical formulae, and is constantly repaiied when new picture is received
Just this mathematical formulae is strengthening which for the ability to express of details area.The mathematical formulae for learning can be while express all
Attack face minutia, so as to obtain the neural network classifier for real human face/attacks facial detail, with judgement
Whether attack signature vector is included in the corresponding characteristic vector of image of each characteristic area.
For example, as shown in figure 4, using whether vectorial including attack signature in the neutral net judging characteristic vector trained
(step S232 in Fig. 3) comprises the steps:
Step S2321:Determine that characteristic vector includes the probability of attack signature vector using the neutral net trained;
Step S2322:When probability is more than predetermined probabilities, step S2323 is proceeded to, when probability is less than or equal to predetermined probabilities
When, proceed to step S2323 ';
Step S2323:Judging characteristic vector includes that (i.e. the image in this feature region includes attacking special attack signature vector
Levy);
Step S2323 ':Do not include in judging characteristic vector that (i.e. the image in this feature region does not include attack signature vector
Attack signature).
For example, in step S2321, characteristic vector application nonlinear transformation is built using the neutral net trained
Mould and identification, to determine that characteristic vector includes the probability of attack signature vector.
For example, predetermined probabilities can be actually needed flexible according to safe class, so that false alarm rate and discrimination
By equilibrium with meet user's request, improve Consumer's Experience.
For example, Fig. 2 is returned to, in following step S240, S250, S260 and S260 ', statistics includes attack signature
The quantity of the image of characteristic area, when the quantity of the image of the characteristic area including attack signature is more than predetermined number, judges
Facial image is not living body faces image (not verified by live body);When the number of the image of the characteristic area including attack signature
When amount is less than or equal to predetermined number (threshold value), judge that facial image is living body faces image (verifying by live body).
For example, the predetermined number in step S240 and S250 can be actually needed flexible according to safe class, make
False alarm rate and discrimination are by equilibrium meeting user's request, improve Consumer's Experience.
For example, the result of living body faces image authentication in step S270, can be exported, that is to say, that by step S260 or
The result output of S260 ', the form of output can include but is not limited to image and sound, and " checking is logical for such as image or voice
Cross " or " checking does not pass through ".Can so make user know the feedback information of checking, improve Consumer's Experience.
For example, embodiment of the disclosure provide for the method for verifying living body faces image, be not limited to a people
Face is verified, it is also possible to which multiple faces are verified together.That is, multiple faces can be obtained in step S220
The image of the multiple characteristic areas in image is used for the process of subsequent step.When being verified to multiple faces together, in order to
Improve the accuracy of checking, the quantity of the characteristic area chosen in can properly increasing step S220, while in step s 250
It is corresponding to improve predetermined number.
Below, the equipment for verifying living body faces image according to the embodiment of the present disclosure will be described with reference to Fig. 5.For testing
The equipment of card living body faces image can perform said method.Due to for verify living body faces image equipment perform each
The details of operation is essentially identical with the method being described above, therefore in order to avoid repeating, hereinafter only to being used to verify
The equipment of living body faces image carries out brief description, and omits the description to same detail.
For example, as shown in figure 5, including spy according to the equipment 500 for verifying living body faces image of the embodiment of the present disclosure
Levy region extracting device 520, attack signature judgment means 530, statistics and judgment means 540 and storage device 550.For example, such as
Shown in Fig. 5, in one example, for verify living body faces image equipment 500 can also include image acquiring device 510 with
And output device 560.For example, the image acquiring device 106 that image acquiring device 510 can be as shown in Figure 1 is realized;Characteristic area
Extraction element 520, attack signature judgment means 530, statistics and judgment means 540 can be as shown in Figure 1 processor 102 it is real
It is existing;Storage device 550 can be as shown in Figure 1 storage device 104 realize;Output device 560 can be as shown in Figure 1 output dress
Put 108 realizations.The concrete structure and implementation of each device can adopt existing various feasible schemes, here not to carry out
Limit.
For example, image acquiring device 510 is configured to perform step S210 in Fig. 2, i.e., in facial image is obtained extremely
Before the image of a few characteristic area, facial image is obtained, image acquiring device can be that visible ray or infrared image are obtained
Device, the image which obtains include but is not limited to gray level image, coloured image or infrared image.
For example, characteristic region extraction device 520 is configured to perform step S220 in Fig. 2, that is, obtain in facial image
The image of at least one characteristic area.
For example, in the equipment for verifying living body faces image that the embodiment of the present disclosure is provided, in facial image at least
The image of one characteristic area is chosen and/or is randomly selected according to priori.
For example, attack signature judgment means 530 are configured to perform step S230 in Fig. 2, that is, judge each characteristic area
Whether attack signature is included in the image in domain.
For example, in the equipment for verifying living body faces image that the embodiment of the present disclosure is provided, attack signature includes hair
Hole disappearance, wrinkle disappearance, shade disappearance, non-natural reflective, living body faces cannot make muscle movement, mosaic, sawtooth and
One kind or its combination in moire fringes.
For example, as shown in fig. 6, attack signature judgment means 530 include that regularization coding module 531 and neutral net judge
Module 532.For example, regularization coding module 531 is configured to perform step S231 in Fig. 3, will each characteristic area
Image regulation is encoded to characteristic vector;Neutral net judge module 532 is configured to perform step S232 in Fig. 3, i.e. profit
Whether attack signature vector is included with the neutral net judging characteristic vector trained.
For example, in the equipment for verifying living body faces image that the embodiment of the present disclosure is provided, the nerve net trained
Network is obtained by following operation:Acquisition includes pore disappearance, wrinkle disappearance, shade is lacked, non-natural reflective, living body faces cannot
The training image of one kind or its combination in muscle movement, mosaic, sawtooth and the moire fringes made;By training image regularization
It is encoded to training feature vector;Neutral net is trained using training feature vector.
For example, can be training feature vector and the corresponding attack signature (example of this feature vector when neutral net is trained
Such as pore disappearance, wrinkle disappearance, the muscle movement that shade is lacked, non-natural reflective, living body faces cannot be made, mosaic, saw
One kind or its combination in tooth and moire fringes) as the input of neutral net, make neutral net (such as thousands of by a large amount of study
Or tens thousand of groups of data) improve discrimination.
For example, in the equipment for verifying living body faces image that the embodiment of the present disclosure is provided, using the god for having trained
Whether include including attack signature vector in Jing networks judging characteristic vector:Characteristic vector is determined using the neutral net trained
Include the probability of attack signature vector;When probability is more than predetermined probabilities, judging characteristic vector includes attack signature vector.
For example, as shown in figure 5, statistics and judgment means 540 be configured to perform Fig. 2 in step S240, S250,
S260, S260 ', i.e., statistics includes the quantity of the image of the characteristic area of attack signature, when the characteristic area including attack signature
Image quantity be more than predetermined number when, judge that facial image is not living body faces image, when include attack signature feature
When the quantity of the image in region is less than or equal to predetermined number, judge that facial image is living body faces image.
For example, storage device 550 is configured to store in facial image, the image of characteristic area and image processing process
Intermediate data.Additionally, storage device 550 is additionally configured to storage for realizing according to the embodiment of the present disclosure for verifying work
The computer program code of the method for body facial image.
For example, output device 560 is configured to perform step S270 in Fig. 2, that is, export living body faces image authentication knot
Really.
Additionally, according to the embodiment of the present disclosure, additionally providing a kind of computer program, which includes that computer-readable is stored
Medium, stores computer program instructions on the computer-readable recording medium.The computer program instructions are being counted
Calculation machine can be realized when running according to the embodiment of the present disclosure for the method for verifying living body faces image, and/or can be with
Realize according to the characteristic region extraction device in the equipment for verifying living body faces image of the embodiment of the present disclosure, attack signature
All or part of function of judgment means and statistics and judgment means.
The method and apparatus for verifying living body faces image that the embodiment of the present disclosure is provided does not need user to make specific
Action, improve Consumer's Experience, and can more effectively defend mask, photo, reproduction or computer synthesis etc. for live body
The attack of recognition of face.
Although above having used general explanation and specific embodiment, make detailed description to the disclosure,
On the basis of the embodiment of the present disclosure, it can be made some modifications or improvements, this is apparent to those skilled in the art
's.Therefore, the these modifications or improvements on the basis of without departing from disclosure spirit, belong to what the disclosure was claimed
Scope.
Claims (19)
1. a kind of method for verifying living body faces image, including:
Obtain the image of at least one characteristic area in facial image;
Whether attack signature is included in judging the image of each characteristic area;
Statistics includes the quantity of the image of the characteristic area of the attack signature;
When the quantity of the image of the characteristic area including the attack signature is more than predetermined number, the facial image is judged not
It is living body faces image, when the quantity of the image of the characteristic area including the attack signature is less than or equal to predetermined number, sentences
The facial image that breaks is living body faces image.
2. the method for verifying living body faces image according to claim 1, wherein, the attack signature includes pore
Muscle movement that disappearance, wrinkle disappearance, shade disappearance, non-natural reflective, living body faces cannot be made, mosaic, sawtooth and rub
One kind or its combination in your stricture of vagina.
3. the method for verifying living body faces image according to claim 1, wherein, at least one in the facial image
The image of individual characteristic area is chosen and/or is randomly selected according to priori.
4. according to any one of claim 1-3 for the method for verifying living body faces image, wherein, it is described to judge each
Whether include including attack signature in the image of the characteristic area:
The image regulation of each characteristic area is encoded to into characteristic vector;
Judge in the characteristic vector, whether include attack signature vector using the neutral net trained.
5. the method for verifying living body faces image according to claim 4, wherein, the neutral net trained
Obtained by following operation:
Acquisition include pore disappearance, wrinkle disappearance, shade lack, non-natural reflective, living body faces cannot make muscle movement,
The training image of one kind or its combination in mosaic, sawtooth and moire fringes;
The training image regularization is encoded to into training feature vector;
The neutral net is trained using the training feature vector.
6. the method for verifying living body faces image according to claim 4, wherein, it is described using the nerve trained
Whether network includes including attack signature vector in judging the characteristic vector:
Determine that the characteristic vector includes the probability of the attack signature vector using the neutral net trained;
When the probability is more than predetermined probabilities, judge that the characteristic vector includes the attack signature vector.
7. according to any one of claim 1-3 for the method for verifying living body faces image, in facial image is obtained
Before the image of at least one characteristic area, also including the acquisition facial image.
8. according to any one of claim 1-3 for the method for verifying living body faces image, also including output live body people
Face image the result.
9. a kind of equipment for verifying living body faces image, including:
One or more processors;
One or more memorizeies;And
Storage computer program instructions in which memory, when the computer program instructions are run by the processor
Perform following steps:
Obtain the image of at least one characteristic area in facial image;
Whether attack signature is included in judging the image of each characteristic area;
Statistics includes the quantity of the image of the characteristic area of the attack signature;
When the quantity of the image of the characteristic area including the attack signature is more than predetermined number, the facial image is judged not
It is living body faces image, when the quantity of the image of the characteristic area including the attack signature is less than or equal to predetermined number, sentences
The facial image that breaks is living body faces image.
10. the equipment for verifying living body faces image according to claim 9, wherein, the attack signature includes hair
Hole disappearance, wrinkle disappearance, shade disappearance, non-natural reflective, living body faces cannot make muscle movement, mosaic, sawtooth and
One kind or its combination in moire fringes.
11. equipment for verifying living body faces image according to claim 9, wherein, in the facial image at least
The image of one characteristic area is chosen and/or is randomly selected according to priori.
12. equipment for verifying living body faces image according to any one of claim 9-11, wherein, it is described to judge every
Whether include including attack signature in the image of the individual characteristic area:
The image regulation of the characteristic area is encoded to into characteristic vector;
Judge in the characteristic vector, whether include attack signature vector using the neutral net trained.
13. equipment for verifying living body faces image according to claim 12, wherein, the nerve net trained
Network is obtained by following operation:
Acquisition include pore disappearance, wrinkle disappearance, shade lack, non-natural reflective, living body faces cannot make muscle movement,
The training image of one kind or its combination in mosaic, sawtooth and moire fringes;
The training image regularization is encoded to into training feature vector;
The neutral net is trained using the training feature vector.
14. equipment for verifying living body faces image according to claim 12, wherein, it is described using the god for having trained
Whether Jing networks include including attack signature vector in judging the characteristic vector:
Determine that the characteristic vector includes the probability of the attack signature vector using the neutral net trained;
When the probability is more than predetermined probabilities, judge that the characteristic vector includes the attack signature vector.
15. equipment for verifying living body faces image according to any one of claim 9-11, also obtain including image
Device, is configured to, in facial image is obtained before the image of at least one characteristic area, obtain the facial image.
16. equipment for verifying living body faces image according to any one of claim 9-11, also including output device,
It is configured to export living body faces image authentication result.
A kind of 17. equipment for verifying living body faces image, including:
Characteristic region extraction device, is configured to obtain the image of at least one characteristic area in facial image;
Whether attack signature judgment means, include attack signature in being configured to the image for judge each characteristic area;With
And
Statistics and judgment means, the quantity of the image for being configured to count the characteristic area for including the attack signature, when including
When the quantity of the image of the characteristic area of the attack signature is more than predetermined number, judge that the facial image is not living body faces
Image, when the quantity of the image of the characteristic area including the attack signature is less than or equal to predetermined number, judges the face
Image is living body faces image.
18. equipment for verifying living body faces image according to claim 17, also include:Image acquiring device, quilt
It is configured to, in facial image is obtained before the image of at least one characteristic area, obtain the facial image.
19. equipment for verifying living body faces image according to claim 17 or 18, also include:Output device, quilt
It is configured as output to living body faces image authentication result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611204264.4A CN106599872A (en) | 2016-12-23 | 2016-12-23 | Method and equipment for verifying living face images |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611204264.4A CN106599872A (en) | 2016-12-23 | 2016-12-23 | Method and equipment for verifying living face images |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106599872A true CN106599872A (en) | 2017-04-26 |
Family
ID=58601216
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611204264.4A Pending CN106599872A (en) | 2016-12-23 | 2016-12-23 | Method and equipment for verifying living face images |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106599872A (en) |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107403166A (en) * | 2017-08-02 | 2017-11-28 | 广东工业大学 | A kind of method and apparatus for extracting facial image pore feature |
CN107886070A (en) * | 2017-11-10 | 2018-04-06 | 北京小米移动软件有限公司 | Verification method, device and the equipment of facial image |
CN108009493A (en) * | 2017-11-30 | 2018-05-08 | 电子科技大学 | Face anti-fraud recognition methods based on action enhancing |
CN108875331A (en) * | 2017-08-01 | 2018-11-23 | 北京旷视科技有限公司 | Face unlocking method, device and system and storage medium |
CN108875467A (en) * | 2017-06-05 | 2018-11-23 | 北京旷视科技有限公司 | The method, apparatus and computer storage medium of In vivo detection |
CN109255319A (en) * | 2018-09-02 | 2019-01-22 | 珠海横琴现联盛科技发展有限公司 | For the recognition of face payment information method for anti-counterfeit of still photo |
CN109271978A (en) * | 2018-11-23 | 2019-01-25 | 四川长虹电器股份有限公司 | Recognition of face anti-fraud method |
WO2020077866A1 (en) * | 2018-10-17 | 2020-04-23 | 平安科技(深圳)有限公司 | Moire-based image recognition method and apparatus, and device and storage medium |
CN111178112A (en) * | 2018-11-09 | 2020-05-19 | 株式会社理光 | Real face recognition device |
CN111356935A (en) * | 2017-09-18 | 2020-06-30 | 埃利蒙特公司 | Method, system, and medium for detecting spoofing in mobile authentication |
CN111382646A (en) * | 2018-12-29 | 2020-07-07 | Tcl集团股份有限公司 | Living body identification method, storage medium and terminal equipment |
CN111723761A (en) * | 2020-06-28 | 2020-09-29 | 杭州海康威视系统技术有限公司 | Method and device for determining abnormal face image and storage medium |
CN113128255A (en) * | 2019-12-30 | 2021-07-16 | 上海依图网络科技有限公司 | Living body detection device method, device, chip, and computer-readable storage medium |
JP2021174529A (en) * | 2020-04-17 | 2021-11-01 | ベイジン バイドゥ ネットコム サイエンス アンド テクノロジー カンパニー リミテッド | Method and device for biometric detection |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105117712A (en) * | 2015-09-15 | 2015-12-02 | 北京天创征腾信息科技有限公司 | Single-sample human face recognition method compatible for human face aging recognition |
CN105260731A (en) * | 2015-11-25 | 2016-01-20 | 商汤集团有限公司 | Human face living body detection system and method based on optical pulses |
CN105518711A (en) * | 2015-06-29 | 2016-04-20 | 北京旷视科技有限公司 | In-vivo detection method, in-vivo detection system, and computer program product |
CN105912986A (en) * | 2016-04-01 | 2016-08-31 | 北京旷视科技有限公司 | In vivo detection method, in vivo detection system and computer program product |
-
2016
- 2016-12-23 CN CN201611204264.4A patent/CN106599872A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105518711A (en) * | 2015-06-29 | 2016-04-20 | 北京旷视科技有限公司 | In-vivo detection method, in-vivo detection system, and computer program product |
CN105117712A (en) * | 2015-09-15 | 2015-12-02 | 北京天创征腾信息科技有限公司 | Single-sample human face recognition method compatible for human face aging recognition |
CN105260731A (en) * | 2015-11-25 | 2016-01-20 | 商汤集团有限公司 | Human face living body detection system and method based on optical pulses |
CN105912986A (en) * | 2016-04-01 | 2016-08-31 | 北京旷视科技有限公司 | In vivo detection method, in vivo detection system and computer program product |
Cited By (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108875467B (en) * | 2017-06-05 | 2020-12-25 | 北京旷视科技有限公司 | Living body detection method, living body detection device and computer storage medium |
CN108875467A (en) * | 2017-06-05 | 2018-11-23 | 北京旷视科技有限公司 | The method, apparatus and computer storage medium of In vivo detection |
CN108875331B (en) * | 2017-08-01 | 2022-08-19 | 北京旷视科技有限公司 | Face unlocking method, device and system and storage medium |
CN108875331A (en) * | 2017-08-01 | 2018-11-23 | 北京旷视科技有限公司 | Face unlocking method, device and system and storage medium |
CN107403166A (en) * | 2017-08-02 | 2017-11-28 | 广东工业大学 | A kind of method and apparatus for extracting facial image pore feature |
CN111356935A (en) * | 2017-09-18 | 2020-06-30 | 埃利蒙特公司 | Method, system, and medium for detecting spoofing in mobile authentication |
CN107886070A (en) * | 2017-11-10 | 2018-04-06 | 北京小米移动软件有限公司 | Verification method, device and the equipment of facial image |
CN108009493A (en) * | 2017-11-30 | 2018-05-08 | 电子科技大学 | Face anti-fraud recognition methods based on action enhancing |
CN108009493B (en) * | 2017-11-30 | 2021-07-06 | 电子科技大学 | Human face anti-cheating recognition method based on motion enhancement |
CN109255319A (en) * | 2018-09-02 | 2019-01-22 | 珠海横琴现联盛科技发展有限公司 | For the recognition of face payment information method for anti-counterfeit of still photo |
WO2020077866A1 (en) * | 2018-10-17 | 2020-04-23 | 平安科技(深圳)有限公司 | Moire-based image recognition method and apparatus, and device and storage medium |
CN111178112A (en) * | 2018-11-09 | 2020-05-19 | 株式会社理光 | Real face recognition device |
CN111178112B (en) * | 2018-11-09 | 2023-06-16 | 株式会社理光 | Face recognition device |
CN109271978A (en) * | 2018-11-23 | 2019-01-25 | 四川长虹电器股份有限公司 | Recognition of face anti-fraud method |
CN111382646A (en) * | 2018-12-29 | 2020-07-07 | Tcl集团股份有限公司 | Living body identification method, storage medium and terminal equipment |
CN111382646B (en) * | 2018-12-29 | 2023-09-05 | Tcl科技集团股份有限公司 | Living body identification method, storage medium and terminal equipment |
CN113128255A (en) * | 2019-12-30 | 2021-07-16 | 上海依图网络科技有限公司 | Living body detection device method, device, chip, and computer-readable storage medium |
JP2021174529A (en) * | 2020-04-17 | 2021-11-01 | ベイジン バイドゥ ネットコム サイエンス アンド テクノロジー カンパニー リミテッド | Method and device for biometric detection |
JP7191139B2 (en) | 2020-04-17 | 2022-12-16 | ベイジン バイドゥ ネットコム サイエンス テクノロジー カンパニー リミテッド | Method and apparatus for detecting living organisms |
CN111723761A (en) * | 2020-06-28 | 2020-09-29 | 杭州海康威视系统技术有限公司 | Method and device for determining abnormal face image and storage medium |
CN111723761B (en) * | 2020-06-28 | 2023-08-11 | 杭州海康威视系统技术有限公司 | Method, device and storage medium for determining abnormal face image |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106599872A (en) | Method and equipment for verifying living face images | |
Zhang et al. | Casia-surf: A large-scale multi-modal benchmark for face anti-spoofing | |
US10275672B2 (en) | Method and apparatus for authenticating liveness face, and computer program product thereof | |
WO2020199611A1 (en) | Liveness detection method and apparatus, electronic device, and storage medium | |
WO2020248780A1 (en) | Living body testing method and apparatus, electronic device and readable storage medium | |
CN106503655B (en) | A kind of electric endorsement method and sign test method based on face recognition technology | |
CN109657554A (en) | A kind of image-recognizing method based on micro- expression, device and relevant device | |
CN108124486A (en) | Face living body detection method based on cloud, electronic device and program product | |
CN109840467A (en) | A kind of in-vivo detection method and system | |
CN115050064A (en) | Face living body detection method, device, equipment and medium | |
CN109635021A (en) | A kind of data information input method, device and equipment based on human testing | |
Al-Dhabi et al. | Deepfake video detection by combining convolutional neural network (cnn) and recurrent neural network (rnn) | |
CN111178130A (en) | Face recognition method, system and readable storage medium based on deep learning | |
CN114241587B (en) | Evaluation method and device for human face living body detection confrontation robustness | |
WO2021042544A1 (en) | Facial verification method and apparatus based on mesh removal model, and computer device and storage medium | |
CN110532891A (en) | Target object state identification method, device, medium and equipment | |
WO2022156214A1 (en) | Liveness detection method and apparatus | |
CN108154070A (en) | Face identification method and device | |
CN109636867A (en) | Image processing method, device and electronic equipment | |
CN113869219A (en) | Face living body detection method, device, equipment and storage medium | |
CN109670285A (en) | Face recognition login method, device, computer equipment and storage medium | |
CN108921080A (en) | Image-recognizing method, device and electronic equipment | |
CN108460811B (en) | Face image processing method and device and computer equipment | |
Grinchuk et al. | 3d mask presentation attack detection via high resolution face parts | |
CN113762969A (en) | Information processing method, information processing device, computer 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 | ||
CB02 | Change of applicant information |
Address after: 100190 room 313, block A, 2 South Road, Haidian District Academy of Sciences, Beijing, China Applicant after: MEGVII INC. Applicant after: Beijing maigewei Technology Co., Ltd. Address before: 100080 room A, block, No. 2 South Road, Haidian District Academy of Sciences, Beijing, China Applicant before: MEGVII INC. Applicant before: Beijing aperture Science and Technology Ltd. |
|
CB02 | Change of applicant information | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170426 |
|
RJ01 | Rejection of invention patent application after publication |