CN110059546A - Vivo identification method, device, terminal and readable medium based on spectrum analysis - Google Patents
Vivo identification method, device, terminal and readable medium based on spectrum analysis Download PDFInfo
- Publication number
- CN110059546A CN110059546A CN201910176437.3A CN201910176437A CN110059546A CN 110059546 A CN110059546 A CN 110059546A CN 201910176437 A CN201910176437 A CN 201910176437A CN 110059546 A CN110059546 A CN 110059546A
- Authority
- CN
- China
- Prior art keywords
- facial image
- image region
- images
- recognized
- face
- 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
- G06V40/161—Detection; Localisation; Normalisation
-
- 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)
- General Physics & Mathematics (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
- Collating Specific Patterns (AREA)
Abstract
The embodiment of the invention discloses a kind of vivo identification method based on spectrum analysis, device, terminal and computer-readable mediums, wherein, the described method includes: obtaining images to be recognized, recognition of face is carried out to the images to be recognized, obtains the first facial image region corresponding with the images to be recognized；By preset feature extraction algorithm, the corresponding first frequency response characteristic vector in first facial image region is obtained；The first frequency response characteristic vector is inputted into the object classifiers that training is completed, exports vivo identification result corresponding with the images to be recognized.Using the above-mentioned vivo identification method based on spectrum analysis, device, terminal and computer-readable medium, in recognition of face scene, the difference of living body, non-living body in frequency is extracted by spectrum analysis accordingly detects the corresponding vivo identification of the face in images to be recognized as a result, improving the accuracy rate of the vivo identification in recognition of face scene.
Description
Technical field
The present invention relates to image procossing and technical field of face recognition, and in particular to a kind of living body knowledge based on spectrum analysis
Other method, apparatus, terminal and computer-readable medium.
Background technique
With the development of computer technology and artificial intelligence technology, face recognition technology comparative maturity and has been widely used,
For example, being all made of recognition of face under many application scenarios to identify user identity.Though face recognition technology is with higher
Accuracy, but there is also a problems, i.e., how detection image is confirmed as true man, rather than photo, video or other are aobvious
Show the deception image such as medium or mask zone camouflage；That is, face in user's photo or video substitutes reality
In the case where face, how to identify whether currently detected face is living body.
Currently, the method for vivo identification is mainly include the following types: the first needs user actively to match based on movement
Specified movement is made in conjunction, such as blinks, and shakes the head, and the method can prevent photo from cheating, but video or mask are cheated
Equal behaviors, this method effect are smaller；Second, it is based on optical flow field, by continuous a few frame images, calculates the optical flow field of face location
Variation, to distinguish plane and 3D object, but for deceptive practices such as masks, this method effect is smaller；The third, is based on color
Texture, the difference by analysis true man's color and vein and photo, video, the color of image texture such as mask are true to be confirmed whether it is
People, the method can identify the deceptive practices of various modes in theory, however in the prior art based on the identification of color and vein
Method, due to that cannot be accurately positioned face location, extract LBP (Local Binary Pattern, local binary patterns) feature
When lose colouring information and be not prominent above the fold feature in calculating, there is also the not high problems of recognition accuracy.
That is, in the technical solution of the vivo identification in existing recognition of face scene, for vivo identification
There are certain deficiencies for recognition accuracy.
Summary of the invention
For technical problem present in above-mentioned related art scheme, in the present invention, provide a kind of based on frequency spectrum point
Vivo identification method, device, terminal and the computer-readable medium of analysis, pass through the spectral response to the corresponding image-region of face
The accuracy rate of the vivo identification in recognition of face scene can be improved to determine whether for living body in feature.
One aspect of the present invention provides a kind of vivo identification method based on spectrum analysis, which comprises
Images to be recognized is obtained, recognition of face is carried out to the images to be recognized, is obtained corresponding with the images to be recognized
The first facial image region；
By preset feature extraction algorithm, the corresponding first frequency response characteristic in first facial image region is obtained
Vector；
The first frequency response characteristic vector is inputted into the object classifiers that training is completed, output and the figure to be identified
As corresponding vivo identification result.
Optionally, the method also includes:
Training set is obtained, the training set includes multiple sample images and specimen discerning knot corresponding with the sample image
Fruit；
Recognition of face is carried out to the sample image, obtains the second facial image region of sample image；
By the feature extraction algorithm, the corresponding second frequency response characteristic in the second facial image region is obtained；
The multiple sample images for including according to the training set and its corresponding specimen discerning result, second frequency response spy
Sign is trained preset classifier, obtains the object classifiers that training is completed.
Optionally, described that recognition of face is carried out to the images to be recognized, obtain corresponding with the images to be recognized the
The step of one facial image region, further includes:
By preset face recognition algorithms, the human face region in the images to be recognized is obtained as the first facial image
Region；
Or,
By preset face characteristic recognizer, the human face region obtained in the images to be recognized is cut, is obtained
Take the first facial image region comprising human face five-sense-organ.
Optionally, described that recognition of face is carried out to the images to be recognized, obtain corresponding with the images to be recognized the
After the step of one facial image region, further includes:
First facial image region is normalized, the first facial image area under pre-set dimension is obtained
The length in domain, the pre-set dimension is equal with width.
Optionally, described by preset feature extraction algorithm, obtain first facial image region corresponding first
The step of frequency-response characteristic vector, further includes:
By preset feature extraction algorithm, frequency of first facial image region on designated color channel is obtained
For response characteristic vector as first frequency response characteristic vector, the designated color channel is one or more.
Optionally, described by preset feature extraction algorithm, obtain first facial image region corresponding first
Before the step of frequency-response characteristic vector, further includes:
According to preset illumination normalization algorithm, illumination normalized, institute are carried out to first facial image region
Stating normalized includes the DC component for removing first facial image region and including.
Optionally, described by preset feature extraction algorithm, obtain first facial image region corresponding first
The step of frequency-response characteristic vector, further includes:
By Fourier transformation, the first spectral image corresponding with first facial image is obtained；
It obtains frequency in first spectral image and is greater than preset amplitude threshold 0 to the amplitude between target frequency
Target pixel points calculate the first frequency response characteristic vector according to target pixel points；
The target frequency is determined according to the area size in first facial image region.
In the second aspect of the present invention, a kind of vivo identification device based on spectrum analysis, described device packet are additionally provided
It includes:
Model training module, for obtaining training set, the training set include multiple sample images and with the sample graph
As corresponding specimen discerning result；Recognition of face is carried out to the sample image, obtains the second facial image area of sample image
Domain；By the feature extraction algorithm, the corresponding second frequency response characteristic in the second facial image region is obtained；According to the instruction
Practice collection include multiple sample images and its corresponding specimen discerning result, second frequency response characteristic to preset classifier into
Row training obtains the object classifiers that training is completed；
Face recognition module carries out recognition of face, acquisition and institute to the images to be recognized for obtaining images to be recognized
State images to be recognized corresponding first facial image region；
Characteristic extracting module, for it is corresponding to obtain first facial image region by preset feature extraction algorithm
First frequency response characteristic vector；
Vivo identification module, for the first frequency response characteristic vector to be inputted the object classifiers that training is completed,
Export vivo identification result corresponding with the images to be recognized.
In the third aspect of the present invention, it is also proposed that a kind of computer equipment, including memory and processor, the storage
Device is stored with computer program, when the computer program is executed by the processor, so that the processor executes following step
It is rapid:
Images to be recognized is obtained, recognition of face is carried out to the images to be recognized, is obtained corresponding with the images to be recognized
The first facial image region；
By preset feature extraction algorithm, the corresponding first frequency response characteristic in first facial image region is obtained
Vector；
The first frequency response characteristic vector is inputted into the object classifiers that training is completed, output and the figure to be identified
As corresponding vivo identification result.
In the fourth aspect of the present invention, it is also proposed that a kind of computer readable storage medium is stored with computer program, institute
When stating computer program and being executed by processor, so that the processor executes following steps:
Images to be recognized is obtained, recognition of face is carried out to the images to be recognized, is obtained corresponding with the images to be recognized
The first facial image region；
By preset feature extraction algorithm, the corresponding first frequency response characteristic in first facial image region is obtained
Vector；
The first frequency response characteristic vector is inputted into the object classifiers that training is completed, output and the figure to be identified
As corresponding vivo identification result.
Implement the embodiment of the present invention, will have the following beneficial effects:
After the above-mentioned vivo identification method based on spectrum analysis, device, terminal and computer-readable medium,
In the case where the recognition of face of user identity identification, recognition of face is carried out for the image identified, is obtained corresponding
Then the spectrum signature response of facial image region on different frequency bands is extracted in facial image region, and according to facial image area
The spectrum signature response of domain on different frequency bands carries out vivo identification, also, the process for carrying out vivo identification to feature vector is
The classifier completed by the training of pre-set training set.By the above-mentioned vivo identification method based on spectrum analysis, device,
Whether terminal and computer-readable medium can be that living body carries out to face during the recognition of face of user identity identification
Vivo identification, to improve user identity identification accuracy.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Wherein:
Fig. 1 is the flow diagram of the vivo identification method based on spectrum analysis in one embodiment；
Fig. 2 is the corresponding facial image region of living body and corresponding frequency response schematic diagram in one embodiment；
Fig. 3 is the corresponding facial image region of non-living body and corresponding frequency response schematic diagram in one embodiment；
Fig. 4 is the process signal of object classifiers training in the vivo identification method based on spectrum analysis in one embodiment
Figure；
Fig. 5 is a kind of composition schematic diagram of the vivo identification device based on spectrum analysis in one embodiment；
Fig. 6 is the structure that the computer equipment of the above-mentioned vivo identification method based on spectrum analysis is run in one embodiment
Schematic diagram.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
In the present embodiment, special to propose a kind of vivo identification method based on spectrum analysis, the realization of this method can be according to
Rely in computer program, which can run on the computer system based on von Neumann system, the computer
Program can be the application program of the recognition of face, vivo identification that are identified to user identity.The computer system can be
Run the computer equipment such as smart phone, tablet computer, PC of above-mentioned computer program.
It should be noted that during being identified by image or video image to the identity of user, although can
To identify the identity of user by recognition of face, but it can not determine whether the face recognized corresponds to true man, rather than
Perhaps other display media or mask zone camouflage etc. cheat image for photo, video.Therefore, recognition of face is being carried out to user
During carrying out identification, it is also necessary to vivo identification is carried out, to really improve the accuracy of recognition of face.
As shown in Figure 1, in one embodiment, a kind of vivo identification method based on display medium is provided, it is specific to wrap
Include following steps S102-S106:
Step S102: obtain images to be recognized, to the images to be recognized carry out recognition of face, obtain with it is described to be identified
Image corresponding first facial image region.
Images to be recognized needs to carry out picture frame in the image or video gathered in advance of vivo identification or by taking the photograph
The video image acquired as head.For example, images to be recognized can be the figure acquired in the process of face recognition by camera
Picture.After getting images to be recognized, include by carrying out recognition of face to images to be recognized to detect in images to be recognized
The window of face, as the first facial image region.
In the present embodiment, the face in images to be recognized is obtained by preset face identification method, and obtains identification
The face arrived corresponding first facial image region.For example, using MTCNN (Multi-task convolutional neural
Network, multitask convolutional neural networks) algorithm or other face recognition algorithms carry out recognition of face to images to be recognized.
In another embodiment, it is also possible to carry out images to be recognized by preset face characteristic recognition methods
Identification identifies the face characteristics such as the face in images to be recognized, cuts to face that acquisition only includes the image district of face
Domain as the first facial image region (that is, the first facial image region does not include hair, ear or background image etc.,
The accuracy of recognition of face can be improved).
In a specific embodiment, above-mentioned face characteristic identification method, which can be, treats knowledge by landmark algorithm
Face in other image is cut.
In another optional embodiment, in order to guarantee the size one for the feature vector extracted in subsequent vivo identification
It causes, the first facial image region recognized can also be normalized, that is to say, that the first for what is recognized
Face image region is normalized, and the first facial image region is converted into the figure of the normal size under preset image sizes
Picture or image-region.
It should be noted that in the present embodiment, at the first facial image region of above-mentioned identification acquisition or normalization
In the size in the first facial image region of reason, corresponding length and width are identical.
It in another alternative embodiment, can also include following step after getting the first facial image region
It is rapid: according to preset illumination normalization algorithm, illumination normalized, the normalizing being carried out to first facial image region
Changing processing includes the DC component that removal first facial image region includes.
Because the variation of illumination may will affect the spectral response of facial image, to cause the accuracy of vivo identification not
Foot, therefore, in the present embodiment, after getting the first facial image region, it is also necessary to which illumination is normalized.Also
It is to say, the identification for normalizing (illumination normalization) Lai Tigao vivo identification by superposition illumination is accurate
Rate.
For example, gray scale balance (histogram equalization) the first facial image of the Lai Jinhang area for passing through opencv
The illumination in domain normalizes.It should be noted that in the present embodiment, the normalized purpose of illumination is to carry out normalizing to illumination
Change, reduce or weaken because illumination change influences the accuracy of vivo identification, it can arbitrarily complete the normalized calculation of illumination
Method can be used as the normalized algorithm of illumination in the present embodiment.
In a specific embodiment, the first facial image region is transformed into the space LAB from RGB mode, in the channel L
The facial image region, is then transformed under RGB mode again and continues feature extraction and living body by lower progress illumination normalization
The correlation step of identification.
Step S104: by preset feature extraction algorithm, corresponding first frequency in first facial image region is obtained
Rate response characteristic vector.
In the present embodiment, the extraction that spectrum analysis feature is carried out to the first facial image region for getting obtains the
The corresponding first frequency response characteristic vector in one facial image region.It should be noted that carry out based on spectrum analysis herein
The method of feature extraction can be based on arbitrary preset feature extraction algorithm.
In the present embodiment, in order to further improve spectrum analysis vivo identification accuracy, can also be multiple
The corresponding frequency-response characteristic vector in region is distinguished on Color Channel.
Specifically, in one embodiment, above-mentioned steps S104: by preset feature extraction algorithm, described the is obtained
The corresponding first frequency response characteristic vector in one facial image region further include: by preset feature extraction algorithm, obtain institute
Frequency-response characteristic vector of the first facial image region on designated color channel is stated as first frequency response characteristic vector,
The designated color channel is one or more.
Specifically, the first facial image region is converted into YCrCb mode by RGB mode, the first facial image area is calculated
Frequency-response characteristic vector of the domain under the channel Y, the channel Cr, the channel Cb.
Alternatively, directly acquiring frequency-response characteristic vector of the first facial image region under the channel R, the channel G, channel B.
When determining first frequency response characteristic vector corresponding with the first facial image region, the channel Y, Cr can be led to
The frequency-response characteristic vector under the channel frequency-response characteristic vector sum R, the channel G, channel B under road, the channel Cb uses simultaneously
Or be applied in combination, it can also be rung simultaneously using the corresponding frequency-response characteristic vector in the first facial image region as first frequency
Answer a part of feature vector.
Further, in a specific embodiment, the acquisition process of above-mentioned first frequency response characteristic vector is: logical
Fourier transformation is crossed, the first spectral image corresponding with first facial image is obtained；It obtains in first spectral image
The target pixel points that frequency is greater than preset amplitude threshold 0 to the amplitude between target frequency, calculate according to target pixel points
The first frequency response characteristic vector；The target frequency is determined according to the area size in first facial image region.
Specifically, the frequency-response characteristic vector of facial image is rung from amplitude of the image after discrete fourier changes
It answers.This Fourier transformation is shifted (shift) operation, and 0 frequency response is made to be located at the centre bit of two-dimensional frequency response
It sets.From the central position outward, different frequencies is respectively represented, same frequency is located on a concentric circles centered on 0 frequency
Face.We have only used the response from 0 to N/2 frequency range, and wherein N is the width and height of facial image, and frequency does not have higher than N/2's
There is use.The frequency for arriving N/2 to 0, wherein amplitude is greater than preset width to the point for being 0 to N/2 frequency in two-dimensional frequency response diagram
The pixel for spending threshold value T is added, and obtains the overall response in this frequency.
It can be expressed with following formula
Wherein, F (u, v) is that frequency is u in the x direction, and frequency is the Frequency and Amplitude response at v in y-direction.
It should be noted that in the present embodiment, the recognition performance of response is improved by preset amplitude threshold T.?
In an optional embodiment, one or more amplitude threshold T can be used1、T2、……、TkTo make first frequency response characteristic
Vector is elongated, improves the Generalization Capability of identification.
As previously mentioned, during features described above is extracted with vivo identification, it is also necessary to be carried out to the first facial image region
Illumination normalization, this is because direct current (frequency 0) component occupies most energy in general pattern, this direct current point
Amount cannot be included among feature vector and normalized parameter, it is therefore desirable to be carried out illumination normalization, be removed in direct current (frequency
For the frequency response on 0) component, avoid value in other frequency responses due to classify than direct current value it is much smaller and very not
Sensitivity, thus identification living body is not achieved and obtains purpose.
In a specific embodiment, the above-mentioned normalized parameter used is that the summation of all frequency responses subtracts 0 frequency
The response of rate can be expressed with following formula:
D=∑u∑v| F (u, v) |-F (0,0),
D (f)=d0(f)/D。
In a specific embodiment, by taking N=64 as an example, final first frequency response characteristic vector are as follows:
D (T)=[d (1), d (2) ..., d (31)],
If using multiple threshold value T1, T2..., Tk, then this first frequency response characteristic vector are as follows:
D (C)=[d (T1) d (T2)...d(Tk)],
Wherein C is a Color Channel.
If using multiple Color Channels, such as Y, Cb, Cr, then this first frequency response characteristic vector are as follows:
D=[d (Y) d (Cb) d (Cr)].
As shown in Fig. 2, giving the corresponding facial image region of living body in one embodiment and corresponding frequency response
Schematic diagram；And Fig. 3 gives the corresponding facial image region of non-living body in one embodiment and corresponding frequency response signal
Figure.According to shown in Fig. 2, Fig. 3, the corresponding frequency-response characteristic vector in living body facial image region corresponding from non-living body is different, according to
This can carry out vivo identification.
Step S106: the first frequency response characteristic vector is inputted into the object classifiers that training is completed, output and institute
State the corresponding vivo identification result of images to be recognized.
In the present embodiment, object classifiers are the classification being trained after completing by pre-set training set
Device, for example, MLP (Multi-Layer Perceptron, multilayer perceptron) classifier or SVM (support vector machines, Support
Vector Machine) classifier.The first frequency extracted in step S104 can be responded by the object classifiers special
It levies vector and carries out vivo identification, obtain corresponding vivo identification result.
In the present embodiment, it is calculated in step S104 after first frequency response characteristic vector, by described first
Frequency-response characteristic vector inputs the object classifiers, to obtain the vivo identification result of object classifiers output.
In a specific embodiment, above-mentioned MLP classifier or SVM classifier are neural network based point a kind of
Class device after training, can carry out vivo identification to the corresponding face of corresponding sample image by training set.It needs
Bright, in the present embodiment, above-mentioned MLP classifier or SVM classifier can also be other neural network models or artificial
Model of mind.
Further, as shown in figure 4, in the present embodiment, additionally providing the above-mentioned vivo identification side based on spectrum analysis
The training method of object classifiers in method.Specifically, including the steps that S202-S208 as shown in Figure 5:
Step S202: obtaining training set, and the training set includes multiple sample images and corresponding with the sample image
Specimen discerning result；
Step S204: recognition of face is carried out to the sample image, obtains the second facial image region of sample image；
Step S206: by the feature extraction algorithm, the corresponding second frequency response in the second facial image region is obtained
Feature；
Step S208: the multiple sample images and its corresponding specimen discerning result, second for including according to the training set
Frequency-response characteristic is trained preset classifier, obtains the object classifiers that training is completed.
Above-mentioned sample set contains multiple sample images, each sample image has corresponded to collected user identity identification
Process acquisition facial image, and in each sample image corresponding face whether be living body vivo identification result (sample
Recognition result) it is also included.In the process and step S102-S106 of the second frequency response characteristic of said extracted sample image
The process for acquiring the first frequency response characteristic of images to be recognized is consistent, and the side of frequency-response characteristic is extracted during the two
Method must be consistent, and just can guarantee the accuracy of subsequent vivo identification result.
During being trained to object classifiers, using the corresponding second frequency response characteristic of sample image as defeated
Enter, using the corresponding specimen discerning result of sample image as output, object classifiers are trained.Further, in this reality
Apply in example, can also a part to sample image as training sample, another part is as verifying sample, in verifying sample
In the case that vivo identification rate reaches certain threshold value, just regards as object classifiers training and complete.
Further, as shown in figure 5, in the present embodiment, it is also proposed that a kind of vivo identification dress based on spectrum analysis
It sets, described device includes:
Model training module 502, for obtaining training set, the training set include multiple sample images and with the sample
The corresponding specimen discerning result of image；Recognition of face is carried out to the sample image, obtains the second facial image of sample image
Region；By the feature extraction algorithm, the corresponding second frequency response characteristic in the second facial image region is obtained；According to described
The multiple sample images and its corresponding specimen discerning result, second frequency response characteristic that training set includes are to preset classifier
It is trained, obtains the object classifiers that training is completed；
Face recognition module 504 carries out recognition of face to the images to be recognized, obtains for obtaining images to be recognized
The first facial image region corresponding with the images to be recognized；
Characteristic extracting module 506, for obtaining first facial image region pair by preset feature extraction algorithm
The first frequency response characteristic vector answered；
Vivo identification module 508, for the first frequency response characteristic vector to be inputted the target classification that training is completed
Device exports vivo identification result corresponding with the images to be recognized.
In an alternative embodiment, face recognition module 504 is also used to obtain by preset face recognition algorithms
Human face region in the images to be recognized is as the first facial image region；Or, face recognition module 504 is also used to by pre-
If face characteristic recognizer, the human face region obtained in the images to be recognized cut, and obtaining includes human face five-sense-organ
The first facial image region.
In an alternative embodiment, characteristic extracting module 506 is also used to carry out first facial image region
Normalized, obtains the first facial image region under pre-set dimension, and the length of the pre-set dimension is equal with width.
In an alternative embodiment, characteristic extracting module 506 is also used to obtain by preset feature extraction algorithm
Frequency-response characteristic vector of first facial image region on designated color channel as first frequency response characteristic to
Amount, the designated color channel are one or more.
In an alternative embodiment, characteristic extracting module 506 is also used to according to preset illumination normalization algorithm, right
First facial image region carries out illumination normalized, and the normalized includes removing first facial image
The DC component that region includes.
In an alternative embodiment, characteristic extracting module 506 is also used to through Fourier transformation, is obtained and described the
Corresponding first spectral image of one facial image；
It obtains frequency in first spectral image and is greater than preset amplitude threshold 0 to the amplitude between target frequency
Target pixel points calculate the first frequency response characteristic vector according to target pixel points；
The target frequency is determined according to the area size in first facial image region.
Fig. 5 shows the internal structure chart of computer equipment in one embodiment.The computer equipment specifically can be clothes
Business device.As shown in figure 5, the computer equipment includes processor, memory and the network interface connected by system bus.Its
In, memory includes non-volatile memory medium and built-in storage.The non-volatile memory medium of the computer equipment is stored with
Operating system can also be stored with computer program, when which is executed by processor, processor realization may make to be based on
The vivo identification method of spectrum analysis.Computer program can also be stored in the built-in storage, the computer program is by processor
When execution, processor may make to execute the vivo identification method based on spectrum analysis.Network interface is for communication with the outside.
It will be understood by those skilled in the art that structure shown in Fig. 5, the only frame of part-structure relevant to application scheme
Figure, does not constitute the restriction for the computer equipment being applied thereon to application scheme, specific computer equipment can wrap
It includes than more or fewer components as shown in the figure, perhaps combines certain components or with different component layouts.
In one embodiment, the vivo identification method provided by the present application by spectrum analysis can be implemented as it is a kind of based on
The form of calculation machine program, computer program can be run in computer equipment as shown in Figure 5.In the memory of computer equipment
Each process template of composition short text filter device can be stored.For example, model training module 502, face recognition module 504,
Characteristic extracting module 506, vivo identification module 508.
A kind of computer equipment, including memory and processor, the memory are stored with computer program, the calculating
When machine program is executed by the processor, so that the processor executes following steps:
Above-mentioned computer equipment, when above-mentioned computer program is executed by the processor in one of the embodiments, also
For executing following steps:
Images to be recognized is obtained, recognition of face is carried out to the images to be recognized, is obtained corresponding with the images to be recognized
The first facial image region；
By preset feature extraction algorithm, the corresponding first frequency response characteristic in first facial image region is obtained
Vector；
The first frequency response characteristic vector is inputted into the object classifiers that training is completed, output and the figure to be identified
As corresponding vivo identification result.
When above-mentioned computer program is executed by the processor in one of the embodiments, it is also used to execute following step
It is rapid:
Training set is obtained, the training set includes multiple sample images and specimen discerning knot corresponding with the sample image
Fruit；
Recognition of face is carried out to the sample image, obtains the second facial image region of sample image；
By the feature extraction algorithm, the corresponding second frequency response characteristic in the second facial image region is obtained；
The multiple sample images for including according to the training set and its corresponding specimen discerning result, second frequency response spy
Sign is trained preset classifier, obtains the object classifiers that training is completed.
A kind of computer readable storage medium is stored with computer program, when the computer program is executed by processor,
So that the processor executes following steps:
Images to be recognized is obtained, recognition of face is carried out to the images to be recognized, is obtained corresponding with the images to be recognized
The first facial image region；
By preset feature extraction algorithm, the corresponding first frequency response characteristic in first facial image region is obtained
Vector；
The first frequency response characteristic vector is inputted into the object classifiers that training is completed, output and the figure to be identified
As corresponding vivo identification result.
When above-mentioned computer program is executed by the processor in one of the embodiments, it is also used to execute following step
It is rapid:
Training set is obtained, the training set includes multiple sample images and specimen discerning knot corresponding with the sample image
Fruit；
Recognition of face is carried out to the sample image, obtains the second facial image region of sample image；
By the feature extraction algorithm, the corresponding second frequency response characteristic in the second facial image region is obtained；
The multiple sample images for including according to the training set and its corresponding specimen discerning result, second frequency response spy
Sign is trained preset classifier, obtains the object classifiers that training is completed.
It should be noted that the above-mentioned vivo identification method based on spectrum analysis, the vivo identification dress based on spectrum analysis
It sets, computer equipment and computer readable storage medium belong to the same inventive concept, vivo identification, base based on spectrum analysis
The content involved in the vivo identification device of spectrum analysis, computer equipment and computer readable storage medium can be fitted mutually
With.
Implement the embodiment of the present invention, will have the following beneficial effects:
After the above-mentioned vivo identification method based on spectrum analysis, device, terminal and computer-readable medium,
In the case where the recognition of face of user identity identification, recognition of face is carried out for the image identified, is obtained corresponding
Then the spectrum signature response of facial image region on different frequency bands is extracted in facial image region, and according to facial image area
The spectrum signature response of domain on different frequency bands carries out vivo identification, also, the process for carrying out vivo identification to feature vector is
The classifier completed by the training of pre-set training set.By the above-mentioned vivo identification method based on spectrum analysis, device,
Whether terminal and computer-readable medium can be that living body carries out to face during the recognition of face of user identity identification
Vivo identification, to improve user identity identification accuracy.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the program can be stored in a non-volatile computer and can be read
In storage medium, the program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein, provided herein
Each embodiment used in any reference to memory, storage, database or other media, may each comprise non-volatile
And/or volatile memory.Nonvolatile memory may include that read-only memory (ROM), programming ROM (PROM), electricity can be compiled
Journey ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include random access memory
(RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms, such as static state RAM
(SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhanced SDRAM
(ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) directly RAM (RDRAM), straight
Connect memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
Each technical characteristic of above embodiments can be combined arbitrarily, for simplicity of description, not to above-described embodiment
In each technical characteristic it is all possible combination be all described, as long as however, the combination of these technical characteristics be not present lance
Shield all should be considered as described in this specification.
The several embodiments of the application above described embodiment only expresses, the description thereof is more specific and detailed, but simultaneously
The limitation to the application the scope of the patents therefore cannot be interpreted as.It should be pointed out that for those of ordinary skill in the art
For, without departing from the concept of this application, various modifications and improvements can be made, these belong to the guarantor of the application
Protect range.Therefore, the scope of protection shall be subject to the appended claims for the application patent.
Claims (10)
1. a kind of vivo identification method based on spectrum analysis, which is characterized in that the described method includes:
Images to be recognized is obtained, recognition of face is carried out to the images to be recognized, obtains corresponding with the images to be recognized the
One facial image region；
By preset feature extraction algorithm, obtain the corresponding first frequency response characteristic in first facial image region to
Amount；
The first frequency response characteristic vector is inputted into the object classifiers that training is completed, output and the images to be recognized pair
The vivo identification result answered.
2. the vivo identification method according to claim 1 based on spectrum analysis, which is characterized in that the method is also wrapped
It includes:
Training set is obtained, the training set includes multiple sample images and specimen discerning result corresponding with the sample image；
Recognition of face is carried out to the sample image, obtains the second facial image region of sample image；
By the feature extraction algorithm, the corresponding second frequency response characteristic in the second facial image region is obtained；
The multiple sample images and its corresponding specimen discerning result, second frequency response characteristic pair for including according to the training set
Preset classifier is trained, and obtains the object classifiers that training is completed.
3. the vivo identification method according to claim 1 based on spectrum analysis, which is characterized in that it is described to described wait know
The step of other image carries out recognition of face, obtains the first facial image region corresponding with the images to be recognized, further includes:
By preset face recognition algorithms, the human face region in the images to be recognized is obtained as the first facial image area
Domain；
Or,
By preset face characteristic recognizer, the human face region obtained in the images to be recognized is cut, and obtains packet
The first facial image region containing human face five-sense-organ.
4. the vivo identification method according to claim 1 based on spectrum analysis, which is characterized in that it is described to described wait know
After the step of other image carries out recognition of face, obtains the first facial image region corresponding with the images to be recognized, also wrap
It includes:
First facial image region is normalized, the first facial image region under pre-set dimension, institute are obtained
The length for stating pre-set dimension is equal with width.
5. the vivo identification method according to claim 1 based on spectrum analysis, which is characterized in that described by preset
Feature extraction algorithm, the step of obtaining the corresponding first frequency response characteristic vector in first facial image region, further includes:
By preset feature extraction algorithm, frequency response of first facial image region on designated color channel is obtained
For feature vector as first frequency response characteristic vector, the designated color channel is one or more.
6. the vivo identification method according to claim 1 based on spectrum analysis, which is characterized in that described by preset
Feature extraction algorithm, before the step of obtaining the corresponding first frequency response characteristic vector in first facial image region, also
Include:
According to preset illumination normalization algorithm, illumination normalized is carried out to first facial image region, it is described to return
One changes the DC component that processing includes including removal first facial image region.
7. the vivo identification method according to claim 1 based on spectrum analysis, which is characterized in that described by preset
Feature extraction algorithm, the step of obtaining the corresponding first frequency response characteristic vector in first facial image region, further includes:
By Fourier transformation, the first spectral image corresponding with first facial image is obtained；
Obtain the target that frequency is greater than preset amplitude threshold 0 to the amplitude between target frequency in first spectral image
Pixel calculates the first frequency response characteristic vector according to target pixel points；
The target frequency is determined according to the area size in first facial image region.
8. a kind of vivo identification device based on spectrum analysis, which is characterized in that described device includes:
Model training module, for obtaining training set, the training set include multiple sample images and with the sample image pair
The specimen discerning result answered；Recognition of face is carried out to the sample image, obtains the second facial image region of sample image；It is logical
The feature extraction algorithm is crossed, the corresponding second frequency response characteristic in the second facial image region is obtained；According to the training set
Including multiple sample images and its corresponding specimen discerning result, second frequency response characteristic preset classifier is instructed
Practice, obtains the object classifiers that training is completed；
Face recognition module, for obtaining images to be recognized, to the images to be recognized carry out recognition of face, obtain with it is described to
Identify image corresponding first facial image region；
Characteristic extracting module, for by preset feature extraction algorithm, obtaining first facial image region corresponding the
One frequency-response characteristic vector；
Vivo identification module, for the first frequency response characteristic vector to be inputted the object classifiers that training is completed, output
Vivo identification result corresponding with the images to be recognized.
9. a kind of terminal, including memory and processor, the memory is stored with computer program, the computer program quilt
When the processor executes, so that the processor is executed such as the step of any one of claims 1 to 7 the method.
10. a kind of computer readable storage medium is stored with computer program, when the computer program is executed by processor,
So that the processor is executed such as the step of any one of claims 1 to 7 the method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910176437.3A CN110059546A (en) | 2019-03-08 | 2019-03-08 | Vivo identification method, device, terminal and readable medium based on spectrum analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910176437.3A CN110059546A (en) | 2019-03-08 | 2019-03-08 | Vivo identification method, device, terminal and readable medium based on spectrum analysis |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110059546A true CN110059546A (en) | 2019-07-26 |
Family
ID=67316689
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910176437.3A Pending CN110059546A (en) | 2019-03-08 | 2019-03-08 | Vivo identification method, device, terminal and readable medium based on spectrum analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110059546A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111178137A (en) * | 2019-12-04 | 2020-05-19 | 百度在线网络技术（北京）有限公司 | Method, device, electronic equipment and computer readable storage medium for detecting real human face |
| CN111460944A (en) * | 2020-03-24 | 2020-07-28 | 深圳数联天下智能科技有限公司 | Live body identification method, device and equipment based on thermal imaging and storage medium |
| CN111553202A (en) * | 2020-04-08 | 2020-08-18 | 浙江大华技术股份有限公司 | Training method, detection method and device of neural network for detecting living body |
| CN111680544A (en) * | 2020-04-24 | 2020-09-18 | 北京迈格威科技有限公司 | Face recognition method, device, system, equipment and medium |
| CN111695515A (en) * | 2020-06-12 | 2020-09-22 | 百度在线网络技术（北京）有限公司 | Video image detection method and device, electronic equipment and storage medium |
| CN112101121A (en) * | 2020-08-19 | 2020-12-18 | 深圳数联天下智能科技有限公司 | Face sensitivity identification method and device, storage medium and computer equipment |
| WO2021080815A1 (en) | 2019-10-21 | 2021-04-29 | Sony Interactive Entertainment Inc. | Fake video detection |
| CN112801013A (en) * | 2021-02-08 | 2021-05-14 | 的卢技术有限公司 | Face recognition method, system and device based on key point recognition and verification |
| CN112884720A (en) * | 2021-02-01 | 2021-06-01 | 广东电网有限责任公司广州供电局 | Distribution line pollution flashover insulator detection method and system |
| CN112926515A (en) * | 2021-03-26 | 2021-06-08 | 支付宝(杭州)信息技术有限公司 | Living body model training method and device |
| CN113657197A (en) * | 2021-07-27 | 2021-11-16 | 浙江大华技术股份有限公司 | Image recognition method, training method of image recognition model and related device |
| CN111695515B (en) * | 2020-06-12 | 2024-06-28 | 百度在线网络技术（北京）有限公司 | Video image detection method and device, electronic equipment and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101999900A (en) * | 2009-08-28 | 2011-04-06 | 南京壹进制信息技术有限公司 | Living body detecting method and system applied to human face recognition |
| CN107958235A (en) * | 2017-12-28 | 2018-04-24 | 泰康保险集团股份有限公司 | A kind of facial image detection method, device, medium and electronic equipment |
| CN108846924A (en) * | 2018-05-31 | 2018-11-20 | 上海商汤智能科技有限公司 | Vehicle and car door solution lock control method, device and car door system for unlocking |
| CN108875544A (en) * | 2018-04-11 | 2018-11-23 | 北京旷视科技有限公司 | Face identification method, device, system and storage medium |
| CN109344716A (en) * | 2018-08-31 | 2019-02-15 | 深圳前海达闼云端智能科技有限公司 | Training method, detection method, device, medium and equipment of living body detection model |
-
2019
- 2019-03-08 CN CN201910176437.3A patent/CN110059546A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101999900A (en) * | 2009-08-28 | 2011-04-06 | 南京壹进制信息技术有限公司 | Living body detecting method and system applied to human face recognition |
| CN107958235A (en) * | 2017-12-28 | 2018-04-24 | 泰康保险集团股份有限公司 | A kind of facial image detection method, device, medium and electronic equipment |
| CN108875544A (en) * | 2018-04-11 | 2018-11-23 | 北京旷视科技有限公司 | Face identification method, device, system and storage medium |
| CN108846924A (en) * | 2018-05-31 | 2018-11-20 | 上海商汤智能科技有限公司 | Vehicle and car door solution lock control method, device and car door system for unlocking |
| CN109344716A (en) * | 2018-08-31 | 2019-02-15 | 深圳前海达闼云端智能科技有限公司 | Training method, detection method, device, medium and equipment of living body detection model |
Non-Patent Citations (2)
| Title |
|---|
| JIANGWEI LI 等: "Live Face Detection Based on the Analysis of Fourier Spectra", 《BIOMETRIC TECHNOLOGY FOR HUMAN IDENTIFICATION》 * |
| 曹瑜: "活体人脸检测技术研究", 《中国优秀硕士学位论文全文数据库 信息科技辑（月刊）计算机软件及计算机应用》 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP4049174A4 (en) * | 2019-10-21 | 2024-01-03 | Sony Interactive Entertainment Inc | Fake video detection |
| WO2021080815A1 (en) | 2019-10-21 | 2021-04-29 | Sony Interactive Entertainment Inc. | Fake video detection |
| CN111178137A (en) * | 2019-12-04 | 2020-05-19 | 百度在线网络技术（北京）有限公司 | Method, device, electronic equipment and computer readable storage medium for detecting real human face |
| CN111178137B (en) * | 2019-12-04 | 2023-05-26 | 百度在线网络技术（北京）有限公司 | Method, device, electronic equipment and computer readable storage medium for detecting real face |
| CN111460944A (en) * | 2020-03-24 | 2020-07-28 | 深圳数联天下智能科技有限公司 | Live body identification method, device and equipment based on thermal imaging and storage medium |
| CN111460944B (en) * | 2020-03-24 | 2023-08-04 | 深圳数联天下智能科技有限公司 | Living body identification method, device and equipment based on thermal imaging and storage medium |
| CN111553202B (en) * | 2020-04-08 | 2023-05-16 | 浙江大华技术股份有限公司 | Training method, detection method and device for neural network for living body detection |
| CN111553202A (en) * | 2020-04-08 | 2020-08-18 | 浙江大华技术股份有限公司 | Training method, detection method and device of neural network for detecting living body |
| CN111680544A (en) * | 2020-04-24 | 2020-09-18 | 北京迈格威科技有限公司 | Face recognition method, device, system, equipment and medium |
| CN111680544B (en) * | 2020-04-24 | 2023-07-21 | 北京迈格威科技有限公司 | Face recognition method, device, system, equipment and medium |
| CN111695515A (en) * | 2020-06-12 | 2020-09-22 | 百度在线网络技术（北京）有限公司 | Video image detection method and device, electronic equipment and storage medium |
| CN111695515B (en) * | 2020-06-12 | 2024-06-28 | 百度在线网络技术（北京）有限公司 | Video image detection method and device, electronic equipment and storage medium |
| CN112101121A (en) * | 2020-08-19 | 2020-12-18 | 深圳数联天下智能科技有限公司 | Face sensitivity identification method and device, storage medium and computer equipment |
| CN112101121B (en) * | 2020-08-19 | 2024-04-30 | 深圳数联天下智能科技有限公司 | Face sensitive identification method and device, storage medium and computer equipment |
| CN112884720A (en) * | 2021-02-01 | 2021-06-01 | 广东电网有限责任公司广州供电局 | Distribution line pollution flashover insulator detection method and system |
| CN112801013A (en) * | 2021-02-08 | 2021-05-14 | 的卢技术有限公司 | Face recognition method, system and device based on key point recognition and verification |
| CN112801013B (en) * | 2021-02-08 | 2024-04-09 | 的卢技术有限公司 | Face recognition method, system and device based on key point recognition verification |
| CN112926515A (en) * | 2021-03-26 | 2021-06-08 | 支付宝(杭州)信息技术有限公司 | Living body model training method and device |
| CN113657197A (en) * | 2021-07-27 | 2021-11-16 | 浙江大华技术股份有限公司 | Image recognition method, training method of image recognition model and related device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110059546A (en) | Vivo identification method, device, terminal and readable medium based on spectrum analysis | |
| Chen et al. | A multi-task convolutional neural network for joint iris detection and presentation attack detection | |
| CN112215180B (en) | Living body detection method and device | |
| US20190095701A1 (en) | Living-body detection method, device and storage medium | |
| CN103914676B (en) | A kind of method and apparatus used in recognition of face | |
| CN106446872A (en) | Detection and recognition method of human face in video under low-light conditions | |
| CN110069983A (en) | Vivo identification method, device, terminal and readable medium based on display medium | |
| KR102286468B1 (en) | Method and system for integrity verification of fake video created through deep learning | |
| Kashem et al. | Face recognition system based on principal component analysis (PCA) with back propagation neural networks (BPNN) | |
| CN110046652A (en) | Face method for evaluating quality, device, terminal and readable medium | |
| WO2022206319A1 (en) | Image processing method and apparatus, and device, storage medium and computer program product | |
| CN109086723B (en) | Method, device and equipment for detecting human face based on transfer learning | |
| CN108829900A (en) | A kind of Research on face image retrieval based on deep learning, device and terminal | |
| WO2016084072A1 (en) | Anti-spoofing system and methods useful in conjunction therewith | |
| Laibacher et al. | M2U-Net: Effective and efficient retinal vessel segmentation for resource-constrained environments | |
| CN108647625A (en) | A kind of expression recognition method and device | |
| CN106446754A (en) | Image identification method, metric learning method, image source identification method and devices | |
| CN109299658B (en) | Face detection method, face image rendering device and storage medium | |
| CN108182714A (en) | Image processing method and device, storage medium | |
| CN104021384B (en) | A kind of face identification method and device | |
| Nishiyama et al. | Facial deblur inference to improve recognition of blurred faces | |
| CN109255319A (en) | For the recognition of face payment information method for anti-counterfeit of still photo | |
| CN110838119A (en) | Human face image quality evaluation method, computer device and computer readable storage medium | |
| CN108369644A (en) | A kind of method and intelligent terminal quantitatively detecting face wrinkles on one's forehead | |
| CN111178130A (en) | Face recognition method, system and readable storage medium based on deep learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190726 |