CN105243378B - Living body faces detection method and device based on eye information - Google Patents
Living body faces detection method and device based on eye information Download PDFInfo
- Publication number
- CN105243378B CN105243378B CN201510781590.0A CN201510781590A CN105243378B CN 105243378 B CN105243378 B CN 105243378B CN 201510781590 A CN201510781590 A CN 201510781590A CN 105243378 B CN105243378 B CN 105243378B
- Authority
- CN
- China
- Prior art keywords
- living body
- result
- body faces
- image sequence
- human 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.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 83
- 238000000034 method Methods 0.000 claims abstract description 24
- 230000003287 optical effect Effects 0.000 claims abstract description 16
- 210000000887 face Anatomy 0.000 claims description 159
- 210000001508 eye Anatomy 0.000 claims description 107
- 238000012360 testing method Methods 0.000 claims description 22
- 230000008859 change Effects 0.000 claims description 7
- 230000004424 eye movement Effects 0.000 claims description 6
- 230000006870 function Effects 0.000 claims description 6
- 230000002123 temporal effect Effects 0.000 claims description 6
- 210000005252 bulbus oculi Anatomy 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 4
- 235000013399 edible fruits Nutrition 0.000 claims description 3
- 230000001815 facial effect Effects 0.000 claims description 3
- 238000007689 inspection Methods 0.000 claims description 2
- 230000004927 fusion Effects 0.000 abstract 1
- 230000000875 corresponding effect Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000001727 in vivo Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000011835 investigation Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000003565 oculomotor Effects 0.000 description 2
- 230000001766 physiological effect Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 241000209202 Bromus secalinus Species 0.000 description 1
- 241000218691 Cupressaceae Species 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000001574 biopsy Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 210000003205 muscle Anatomy 0.000 description 1
- 210000000056 organ Anatomy 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
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
- G06V40/40—Spoof detection, e.g. liveness detection
- G06V40/45—Detection of the body part being alive
Landscapes
- Engineering & Computer Science (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Image Analysis (AREA)
- Collating Specific Patterns (AREA)
Abstract
The present invention provides the living body faces detection method and device based on eye information, and method includes: to obtain human face image sequence to be detected, and Optical flow estimation is carried out to it and obtains wherein eye areas；It determines each each pixel light stream of frame eye areas in sequence, judges whether each pixel light stream of each frame eye areas is consistent, the first result of detection is living body if otherwise determining, if determining that the first result of detection is prosthese；Hamming distance in the sequence of calculation between each frame eye areas binary image, living body score is determined according to it, determine that the second result of detection is living body if living body score is more than or equal to preset threshold, if living body score is less than preset threshold and determines that the second result of detection is prosthese；Judge to act in sequence with the presence or absence of blink, if then determining that detection third result is living body, if otherwise determining, detection third result is prosthese；Three kinds of result fusions are obtained into final detection result.This method can detect whether face is living body, and complexity is low, is not required to people cooperates on one's own initiative, strong robustness.
Description
Technical field
The present invention relates to biometrics identification technology field more particularly to a kind of living body faces detections based on eye information
Method and device.
Background technique
Currently, biological characteristic has had been applied to access control system, public security organ's investigation, user password etc..In recent years,
In vivo detection technology has seemed extremely important in face identification system, has also started one upsurge in academia.
Living body faces detection, that is, detect a picture or video is true man or false face.The algorithm of recognition of face is close
Year achieves the progress advanced by leaps and bounds, and existing biopsy method mainly judges whether face has physiological activity.Due to video
Cheating also has the various disadvantages such as physiological activity, detection complexity height, and in recognition of face, someone falsely uses photo, video or face
Mask substitutes true man and cheats face identification system.After recognition of face is applied in the fields such as network authentication and social security, bank, false problem is known
The critical issue that must be solved as one, and the face recognition algorithms in existing security system are for living body faces test problems
It is not excessive to consider.
There are various living body faces detection algorithms at present, algorithms of different cuts both ways.According to data source point, have based on individual
The method of picture, the method based on continuous videos and the method based on special installation.
Wherein, the method realizability based on single picture is stronger, can be used for single image or image sequence simultaneously；But it lacks
Point is the information that can not utilize consecutive image sequence, and detection potentiality are slightly worse.In such methods there are three more mature algorithms, point
Be not the feature based on frequency spectrum, the feature based on texture and based on difference of Gaussian (Difference of Gaussian, referred to as
DoG it) is combined with the feature of local binary patterns (Local Binary Patterns, abbreviation LBP).
Method based on video sequence utilizes the correlation information of each frame image, is less susceptible to be spoofed, but algorithm is more multiple
It is miscellaneous, it is sometimes desirable to human-computer interaction.There is more mature algorithm in such methods, algorithm based on optical flow method and be based on light stream direction
The algorithm of histogram, the algorithm based on optical flow method are not introduced into machine learning, and directly according to optical flow computation confidence level, effect is paid no attention to
Think.
It is insufficient efficiently against single picture or video sequence bring based on multimembrane state In vivo detection technology.If single
In vivo detection is carried out by image is obtained purely, potential security risk is more obvious.Since lip can be in pronunciation by people
A series of correlated activation occurs with voice, these active performances in the video sequence, have had paid attention in the prior art
This point, and related muscles are had studied, joint, the activity of tooth.These information can be extracted in multimembrane state In vivo detection, and
Combine and authenticated in conjunction with the information characteristics of voice, considerably increases the difficulty of deception.But it to be accurately tracked by and be partitioned into
The specific location of lip is a quite difficult project, is worth scholar's further investigation.
In consideration of it, how to provide, a kind of complexity is low, living body faces detection method and device of strong robustness become current
The technical issues that need to address.
Summary of the invention
In order to solve the above technical problems, the present invention provides a kind of living body faces detection method and dress based on eye information
It sets, is able to detect whether face is living body, complexity is low, hardware requirement is low, adapts to display environment complicated and changeable, is not required to very important person
It cooperates on one's own initiative, strong robustness.
In a first aspect, the present invention provides a kind of living body faces detection method based on eye information, comprising:
Obtain human face image sequence to be detected；
Optical flow estimation is carried out to the human face image sequence, and obtains the eyes area of face in the human face image sequence
Domain；
The light stream for determining each pixel of eye areas of each frame face in the human face image sequence judges each frame face
Each pixel of eye areas light stream it is whether consistent, if inconsistent, it is determined that living body faces detect the first result be living body, if one
It causes, it is determined that it is prosthese that living body faces, which detect the first result,；
The Hamming distance in the human face image sequence between each frame eye areas image is calculated, according to the Hamming distance
Determine living body score, the degree that the living body score reflection eyes change in the human face image sequence, by the living body point
Number is compared with preset threshold, if the living body score is more than or equal to preset threshold, it is determined that living body faces detect the second knot
Fruit is living body, if the living body score is less than preset threshold, it is determined that it is prosthese that living body faces, which detect the second result,；
Judge to act in the human face image sequence with the presence or absence of blink, if it is dynamic to there is blink in the human face image sequence
Make, it is determined that it is living body that living body faces, which detect third result, if there is no blinks to act in the human face image sequence, it is determined that
It is prosthese that living body faces, which detect third result,；
The living body faces are detected into the first result, the living body faces detect the second result and living body faces detection
Third result is merged, and final living body faces testing result is obtained.
It is optionally, described to obtain human face image sequence to be detected, comprising:
It obtains when the display screen that camera is arranged below the camera shows live image and is acquired in preset duration
Human face image sequence.
Optionally, the eye areas for obtaining face in the human face image sequence, comprising:
Using Gabor characteristic and SVM classifier, the eye areas of face in the human face image sequence is obtained.
Optionally, the Hamming distance calculated in the human face image sequence between each frame eye areas image, comprising:
The human face region image of frame each in the human face image sequence is normalized；
Eye areas image in frame human face region image each after normalization is subjected to binaryzation；
Hamming distance after calculating binaryzation between each frame eye areas image.
Optionally, it is acted in the judgement human face image sequence with the presence or absence of blink, comprising:
Judge that each frame image is to open eyes or close one's eyes, and pass through condition random in the human face image sequence using classifier
Field judges to act in the human face image sequence with the presence or absence of blink.
Optionally, described the living body faces are detected into the first result, the living body faces detect the second result and described
Living body faces detection third result is merged, and final living body faces testing result is obtained, comprising:
Using temporal voting strategy, the living body faces are detected into the first result, the living body faces detect the second result and institute
It states living body faces detection third result to be merged, obtains final living body faces testing result.
Optionally, described the living body faces are detected into the first result, the living body faces detect the second result and described
Living body faces detection third result is merged, and final living body faces testing result is obtained, comprising:
The living body faces are examined by learning a discriminant function using metric learning metric learning strategy
It surveys the first result, the living body faces the second result of detection and living body faces detection third result to be integrated, obtain most
Whole living body faces testing result, and learn to the living body faces to detect the first result, the living body faces the second result of detection
With the different weights of living body faces detection third result.
Second aspect, the present invention provide a kind of living body faces detection device based on eye information, comprising:
Image sequence obtains module, for obtaining human face image sequence to be detected；
Eye areas obtains module, for carrying out Optical flow estimation to the human face image sequence, and obtains the face
The eye areas of face in image sequence；
Intermediate result obtains module, comprising: the first result acquiring unit, the second result acquiring unit and third result obtain
Unit；
The first result acquiring unit, for determining each picture of eye areas of each frame face in the human face image sequence
The light stream of element, judges whether the light stream of each pixel of eye areas of each frame face is consistent, if inconsistent, it is determined that living body people
It is living body that face, which detects the first result, if unanimously, it is determined that it is prosthese that living body faces, which detect the first result,；
The second result acquiring unit, for calculating in the human face image sequence between each frame eye areas image
Hamming distance determines living body score according to the Hamming distance, and the living body score reflection eyes are in the human face image sequence
The living body score is compared by the degree of middle variation with preset threshold, if the living body score is more than or equal to preset threshold,
Then determining that living body faces detect the second result is living body, if the living body score is less than preset threshold, it is determined that living body faces inspection
Surveying the second result is prosthese；
The third result acquiring unit is acted for judging in the human face image sequence with the presence or absence of blink, if institute
State in human face image sequence that there are blink movements, it is determined that it is living body that living body faces, which detect third result, if the facial image
There is no blinks to act in sequence, it is determined that it is prosthese that living body faces, which detect third result,；
Final result obtains module, for the living body faces to be detected the first result, living body faces detection second
As a result it is merged with living body faces detection third result, obtains final living body faces testing result.
Optionally, described image retrieval module, is specifically used for
It obtains when the display screen that camera is arranged below the camera shows live image and is acquired in preset duration
Human face image sequence；
And/or
The eye areas obtains module, is specifically used for
Optical flow estimation is carried out to the human face image sequence, and utilizes Gabor characteristic and SVM classifier, described in acquisition
The eye areas of face in human face image sequence；
And/or
The second result acquiring unit, is specifically used for
The human face region image of frame each in the human face image sequence is normalized, Hou Gezheng face area will be normalized
Eye areas image in area image carries out binaryzation, calculates the Hamming distance after binaryzation between each frame eye areas image,
Living body score, the journey that the living body score reflection eyes change in the human face image sequence are determined according to the Hamming distance
Degree, the living body score is compared with preset threshold, if the living body score is more than or equal to preset threshold, it is determined that living body
The second result of Face datection is living body, if the living body score is less than preset threshold, it is determined that living body faces detect the second result
For prosthese；
And/or
The third result acquiring unit, is specifically used for
Judge that each frame image is to open eyes or close one's eyes, and pass through condition random in the human face image sequence using classifier
Field judge with the presence or absence of blink movement in the human face image sequence, if there are blink movement in the human face image sequence,
Determine that living body faces detection third result is living body, if there is no blinks to act in the human face image sequence, it is determined that living body
Face datection third result is prosthese.
Optionally, the final result obtains module, is specifically used for
Using temporal voting strategy, the living body faces are detected into the first result, the living body faces detect the second result and institute
It states living body faces detection third result to be merged, obtains final living body faces testing result；
And/or
The final result obtains module, is specifically used for
The living body faces are examined by learning a discriminant function using metric learning metric learning strategy
It surveys the first result, the living body faces the second result of detection and living body faces detection third result to be integrated, obtain most
Whole living body faces testing result, and learn to the living body faces to detect the first result, the living body faces the second result of detection
With the different weights of living body faces detection third result.
As shown from the above technical solution, the living body faces detection method and device of the invention based on eye information, can
Detect whether face is living body, complexity is low, hardware requirement is low, adapt to display environment complicated and changeable, is not required to very important person actively matches
It closes, strong robustness.
Detailed description of the invention
Fig. 1 is a kind of process signal for living body faces detection method based on eye information that one embodiment of the invention provides
Figure；
Fig. 2 is a kind of structural representation for living body faces detection device based on eye information that one embodiment of the invention provides
Figure.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, the technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is only
It is only a part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiment of the present invention, ordinary skill people
Member's every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
Fig. 1 shows the process signal of the living body faces detection method based on eye information of one embodiment of the invention offer
Figure, as shown in Figure 1, the living body faces detection method based on eye information of the present embodiment is as described below.
101, human face image sequence to be detected is obtained.
In a particular application, one display screen can be set in the lower section of man face image acquiring camera, it is wide plays public good
The dynamics screens such as announcement, so as to cause the variation of living human eye focus, and can show face and other picture materials, accordingly
Ground, the step 101 can be with specifically: obtain when the display screen that camera is arranged below the camera shows live image
Human face image sequence collected in preset duration.
Further, the preset duration is preferably to guarantee that the preset duration of acquisition human face image sequence was big greater than 5 seconds
In 5 seconds, it is ensured that obtain the blink action sequence of true face.
It will be appreciated that when eye gaze display screen, since display screen shows that live image, human eye can unconsciously add
With note that therefore can detecte oculomotor variation, eyeball generates movement, it is living body when detecting eye movement, does not detect
To when eye movement be prosthese.
102, Optical flow estimation is carried out to the human face image sequence, and obtains the eye of face in the human face image sequence
Eyeball region.
In a particular application, " eye areas for obtaining face in the human face image sequence " in the step 102, can
To include:
Using gal cypress Gabor characteristic and support vector machines (Support Vector Machine, abbreviation SVM) classifier,
Obtain the eye areas of face in the human face image sequence.
103, the light stream for determining each pixel of eye areas of each frame face in the human face image sequence, judges each frame
Whether the light stream of each pixel of the eye areas of face is consistent, if inconsistent, it is determined that and it is living body that living body faces, which detect the first result,
If consistent, it is determined that it is prosthese that living body faces, which detect the first result,.
It further, in a particular application, can be according to each picture of eye areas of each frame face in the step 103
The similarity of the light stream of element carries out the marking between 0 to 1 to the living body of face, and similarity is that 100% corresponding living body obtains
It is divided into 0, similarity is that 0% corresponding living body is scored at 1, and the value range of the living body score is [0,1], and living body obtains
Divide the eyes that each frame face is determined when being more than or equal to preset first threshold value (i.e. similarity is less than or equal to default second threshold)
The light stream of each pixel in region is inconsistent, and determining that living body faces detect the first result is living body, and living body score is less than default first
Threshold value determines that the light stream of each pixel of eye areas of each frame face is consistent when (similarity is greater than default second threshold), determines
It is prosthese that living body faces, which detect the first result,.
It will be appreciated that this step 103 is that the optical flow characteristic that the stereochemical structure based on living human eye generates is different from prosthese
The optical flow characteristic of planar structure.
104, the Hamming distance in the human face image sequence between each frame eye areas image is calculated, according to the Hamming
Distance determines living body score, the degree that the living body score reflection eyes change in the human face image sequence, by the work
Body score is compared with preset threshold, if the living body score is more than or equal to preset threshold, it is determined that living body faces detection the
Two results are living body, if the living body score is less than preset threshold, it is determined that it is prosthese that living body faces, which detect the second result,.
In a particular application, in the step 104 " calculate in the human face image sequence each frame eye areas image it
Between Hamming distance ", may include:
The human face region image of frame each in the human face image sequence is normalized；
Eye areas image in frame human face region image each after normalization is subjected to binaryzation；
Hamming distance after calculating binaryzation between each frame eye areas image.
It will be appreciated that this step 104 is that the reaction of environmental stimuli is caused to goggle based on living body, to generate
Different from the binaryzation feature of prosthese.
105, judge to act in the human face image sequence with the presence or absence of blink, be blinked if existing in the human face image sequence
Eye movement is made, it is determined that and it is living body that living body faces, which detect third result, if there is no blinks to act in the human face image sequence,
Determine that living body faces detection third result is prosthese.
In a particular application, " dynamic with the presence or absence of blink in the judgement human face image sequence in the step 105
Make ", may include:
Judge that each frame image is to open eyes or close one's eyes, and pass through condition random in the human face image sequence using classifier
Field judges to act in the human face image sequence with the presence or absence of blink.
106, the living body faces are detected into the first result, the living body faces detect the second result and the living body faces
Detection third result is merged, and final living body faces testing result is obtained.
In a concrete application, the step 106 can be with specifically:
Using temporal voting strategy, the living body faces are detected into the first result, the living body faces detect the second result and institute
It states living body faces detection third result to be merged, obtains final living body faces testing result.
In another concrete application, the step 106 can be with specifically:
The living body faces are examined by learning a discriminant function using metric learning metric learning strategy
It surveys the first result, the living body faces the second result of detection and living body faces detection third result to be integrated, obtain most
Whole living body faces testing result, and learn to the living body faces to detect the first result, the living body faces the second result of detection
With the different weights of living body faces detection third result.
It will be appreciated that introducing learning strategy, the structure of data can be preferably adapted to.
It in a particular application, can also be to final living after the step 106 obtains final living body faces testing result
Body Face datection result, which provides, correspondingly to be fed back, for example prompt user is false face, it should be noted that etc. prompt informations etc..
The living body faces detection method based on eye information of the present embodiment is able to detect whether face is living body, complicated
It spends that low, hardware requirement is low, adapt to display environment complicated and changeable, is not required to very important person and cooperates on one's own initiative, strong robustness.
Fig. 2 shows a kind of structures for living body faces detection device based on eye information that one embodiment of the invention provides
Schematic diagram, as shown in Fig. 2, the living body faces detection device based on eye information of the present embodiment, comprising: image sequence obtains mould
Block 21, eye areas obtain module 22, intermediate result obtains module 23 and final result obtains module 24；
Image sequence obtains module 21, for obtaining human face image sequence to be detected；
Eye areas obtains module 22, for carrying out Optical flow estimation to the human face image sequence, and obtains the people
The eye areas of face in face image sequence；
Intermediate result obtains module 23, comprising: the first result acquiring unit 23a, the second result acquiring unit 23b and third
As a result acquiring unit 23c；
The first result acquiring unit 23a, for determining the eye areas of each frame face in the human face image sequence
The light stream of each pixel judges whether the light stream of each pixel of eye areas of each frame face is consistent, if inconsistent, it is determined that living
The first result of body Face datection is living body, if unanimously, it is determined that it is prosthese that living body faces, which detect the first result,；
The second result acquiring unit 23b, for calculate in the human face image sequence each frame eye areas image it
Between Hamming distance, living body score is determined according to the Hamming distance, the living body score reflection eyes are in the facial image
The living body score is compared by the degree changed in sequence with preset threshold, is preset if the living body score is more than or equal to
Threshold value, it is determined that it is living body that living body faces, which detect the second result, if the living body score is less than preset threshold, it is determined that living body people
It is prosthese that face, which detects the second result,；
The third result acquiring unit 23c is acted in the human face image sequence with the presence or absence of blink for judging, if
There are blink movements in the human face image sequence, it is determined that it is living body that living body faces, which detect third result, if the face figure
As there is no blinks to act in sequence, it is determined that it is prosthese that living body faces, which detect third result,；
Final result obtains module 24, for the living body faces to be detected the first result, living body faces detection the
Two results and living body faces detection third result are merged, and final living body faces testing result is obtained.
In a particular application, one display screen can be set in the lower section of man face image acquiring camera, it is wide plays public good
The dynamics screens such as announcement, so as to cause the variation of living human eye focus, and can show face and other picture materials, accordingly
Ground, described image retrieval module 21, can be specifically used for
It obtains when the display screen that camera is arranged below the camera shows live image and is acquired in preset duration
Human face image sequence.
Further, the preset duration is preferably to guarantee that the preset duration of acquisition human face image sequence was big greater than 5 seconds
In 5 seconds, it is ensured that obtain the blink action sequence of true face.
It will be appreciated that when eye gaze display screen, since display screen shows that live image, human eye can unconsciously add
With note that therefore can detecte oculomotor variation, eyeball generates movement, it is living body when detecting eye movement, does not detect
To when eye movement be prosthese.
In a particular application, the eye areas obtains module 22, can be specifically used for
Optical flow estimation is carried out to the human face image sequence, and utilizes Gabor characteristic and SVM classifier, described in acquisition
The eye areas of face in human face image sequence.
It in a particular application, can be according to the eyes area of each frame face in the first result acquiring unit 23a
The similarity of the light stream of each pixel in domain carries out the marking between 0 to 1 to the living body of face, and similarity is 100% corresponding work
Body is scored at 0, and similarity is that 0% corresponding living body is scored at 1, and the value range of the living body score is [0,1], living
Body score determines each frame face when being more than or equal to preset first threshold value (i.e. similarity is less than or equal to default second threshold)
Each pixel of eye areas light stream it is inconsistent, determining that living body faces detect the first result is living body, and living body score is less than pre-
If determining the light stream one of each pixel of eye areas of each frame face when first threshold (similarity is greater than default second threshold)
It causes, determining that living body faces detect the first result is prosthese.
It will be appreciated that the first result acquiring unit 23a, the light stream that the stereochemical structure based on living human eye generates is special
Property be different from prosthese planar structure optical flow characteristic.
In a particular application, the second result acquiring unit 23b, can be specifically used for
The human face region image of frame each in the human face image sequence is normalized, Hou Gezheng face area will be normalized
Eye areas image in area image carries out binaryzation, calculates the Hamming distance after binaryzation between each frame eye areas image,
Living body score, the journey that the living body score reflection eyes change in the human face image sequence are determined according to the Hamming distance
Degree, the living body score is compared with preset threshold, if the living body score is more than or equal to preset threshold, it is determined that living body
The second result of Face datection is living body, if the living body score is less than preset threshold, it is determined that living body faces detect the second result
For prosthese.
It will be appreciated that the second result acquiring unit 23b, leads to eyeball for the reaction of environmental stimuli based on living body
Rotation, to generate the binaryzation feature for being different from prosthese.
In a particular application, the third result acquiring unit 23c, can be specifically used for
Judge that each frame image is to open eyes or close one's eyes, and pass through condition random in the human face image sequence using classifier
Field judge with the presence or absence of blink movement in the human face image sequence, if there are blink movement in the human face image sequence,
Determine that living body faces detection third result is living body, if there is no blinks to act in the human face image sequence, it is determined that living body
Face datection third result is prosthese.
In a concrete application, the final result obtains module 24, can be specifically used for
Using temporal voting strategy, the living body faces are detected into the first result, the living body faces detect the second result and institute
It states living body faces detection third result to be merged, obtains final living body faces testing result.
In another concrete application, the final result obtains module 24, can be specifically used for
The living body faces are examined by learning a discriminant function using metric learning metric learning strategy
It surveys the first result, the living body faces the second result of detection and living body faces detection third result to be integrated, obtain most
Whole living body faces testing result, and learn to the living body faces to detect the first result, the living body faces the second result of detection
With the different weights of living body faces detection third result.
It will be appreciated that introducing learning strategy, the structure of data can be preferably adapted to.
In a particular application, after final result acquisition module 24 obtains final living body faces testing result, also
May include not shown in the figure:
Feedback module, for final living body faces testing result provide correspondingly feedback (such as prompt user be false face,
It should be noted that etc. prompt informations etc.).
The living body faces detection device based on eye information of the present embodiment is able to detect whether face is living body, complicated
It spends that low, hardware requirement is low, adapt to display environment complicated and changeable, is not required to very important person and cooperates on one's own initiative, strong robustness.
The living body faces detection device based on eye information of the present embodiment, can be used for executing method shown in earlier figures 1
The technical solution of embodiment, it is similar that the realization principle and technical effect are similar, and details are not described herein again.
" first ", " second " and " third " etc. is not that regulation is made to sequencing in embodiments of the present invention, only
It is to make difference to title, in embodiments of the present invention, does not make any restriction.
Those of ordinary skill in the art will appreciate that: realize that all or part of the steps of above-mentioned each method embodiment can lead to
The relevant hardware of program instruction is crossed to complete.Program above-mentioned can be stored in a computer readable storage medium.The journey
When being executed, execution includes the steps that above-mentioned each method embodiment to sequence；And storage medium above-mentioned include: ROM, RAM, magnetic disk or
The various media that can store program code such as person's CD.
Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention., rather than its limitations；To the greatest extent
Pipe present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: its according to
So be possible to modify the technical solutions described in the foregoing embodiments, or to some or all of the technical features into
Row equivalent replacement；And these are modified or replaceed, various embodiments of the present invention technology that it does not separate the essence of the corresponding technical solution
The range of scheme.
Claims (9)
1. a kind of living body faces detection method based on eye information characterized by comprising
Obtain when the display screen that is arranged below the camera of camera shows live image the people collected in preset duration
Face image sequence；
Optical flow estimation is carried out to the human face image sequence, and obtains the eye areas of face in the human face image sequence；
The light stream for determining each pixel of eye areas of each frame face in the human face image sequence judges the eye of each frame face
Whether the light stream of each pixel in eyeball region is consistent, if inconsistent, it is determined that and it is living body that living body faces, which detect the first result, if unanimously,
Then determining that living body faces detect the first result is prosthese；
The Hamming distance in the human face image sequence between each frame eye areas image is calculated, is determined according to the Hamming distance
Living body score, the degree that living body score reflection eyes change in the human face image sequence, by the living body score with
Preset threshold is compared, if the living body score is more than or equal to preset threshold, it is determined that living body faces detect the second result and are
Living body, if the living body score is less than preset threshold, it is determined that it is prosthese that living body faces, which detect the second result,；
Judge to act in the human face image sequence with the presence or absence of blink, if there are blink movement in the human face image sequence,
Then determine that living body faces detection third result is living body, if there is no blinks to act in the human face image sequence, it is determined that living
Body Face datection third result is prosthese；
The living body faces are detected into the first result, the living body faces detect the second result and the living body faces detect third
As a result it is merged, obtains final living body faces testing result.
2. the method according to claim 1, wherein the eyes for obtaining face in the human face image sequence
Region, comprising:
Using Gabor characteristic and SVM classifier, the eye areas of face in the human face image sequence is obtained.
3. the method according to claim 1, wherein described calculate each frame eyes area in the human face image sequence
Hamming distance between area image, comprising:
The human face region image of frame each in the human face image sequence is normalized；
Eye areas image in frame human face region image each after normalization is subjected to binaryzation；
Hamming distance after calculating binaryzation between each frame eye areas image.
4. being blinked the method according to claim 1, wherein whether there is in the judgement human face image sequence
Eye movement is made, comprising:
Judge that each frame image is to open eyes or close one's eyes, and sentence by condition random field in the human face image sequence using classifier
Break and is acted with the presence or absence of blink in the human face image sequence.
5. the method according to claim 1, wherein described detect the first result, described for the living body faces
Living body faces detect the second result and living body faces detection third result is merged, and obtain final living body faces detection knot
Fruit, comprising:
Using temporal voting strategy, the living body faces are detected into the first result, the living body faces detect the second result and the work
Body Face datection third result is merged, and final living body faces testing result is obtained.
6. the method according to claim 1, wherein described detect the first result, described for the living body faces
Living body faces detect the second result and living body faces detection third result is merged, and obtain final living body faces detection knot
Fruit, comprising:
Using metric learning metric learning strategy, by learning a discriminant function, by living body faces detection the
One result, the living body faces detect the second result and living body faces detection third result is integrated, and are finally lived
Body Face datection is as a result, and learn to the living body faces to detect the first result, the living body faces the second result of detection and institute
State the different weights of living body faces detection third result.
7. a kind of living body faces detection device based on eye information characterized by comprising
Image sequence obtains module, for obtain when the display screen that camera is arranged below the camera shows live image
Human face image sequence collected in preset duration；
Eye areas obtains module, for carrying out Optical flow estimation to the human face image sequence, and obtains the facial image
The eye areas of face in sequence；
Intermediate result obtains module, comprising: the first result acquiring unit, the second result acquiring unit and third result obtain single
Member；
The first result acquiring unit, for determining eye areas each pixel of each frame face in the human face image sequence
Light stream judges whether the light stream of each pixel of eye areas of each frame face is consistent, if inconsistent, it is determined that living body faces inspection
Surveying the first result is living body, if unanimously, it is determined that it is prosthese that living body faces, which detect the first result,；
The second result acquiring unit, for calculating the Hamming in the human face image sequence between each frame eye areas image
Distance determines that living body score, the living body score reflection eyes become in the human face image sequence according to the Hamming distance
The living body score is compared by the degree of change with preset threshold, if the living body score is more than or equal to preset threshold, really
Determine living body faces to detect the second result to be living body, if the living body score is less than preset threshold, it is determined that living body faces detection the
Two results are prosthese；
The third result acquiring unit is acted for judging in the human face image sequence with the presence or absence of blink, if the people
There are blink movements in face image sequence, it is determined that it is living body that living body faces, which detect third result, if the human face image sequence
In there is no blink act, it is determined that living body faces detect third result be prosthese；
Final result obtains module, and for the living body faces to be detected the first result, the living body faces detect the second result
It is merged with living body faces detection third result, obtains final living body faces testing result.
8. device according to claim 7, which is characterized in that the eye areas obtains module, is specifically used for
Optical flow estimation is carried out to the human face image sequence, and utilizes Gabor characteristic and SVM classifier, obtains the face
The eye areas of face in image sequence；
And/or
The second result acquiring unit, is specifically used for
The human face region image of frame each in the human face image sequence is normalized, by frame human face region figure each after normalization
Eye areas image as in carries out binaryzation, calculates the Hamming distance after binaryzation between each frame eye areas image, according to
The Hamming distance determines that living body score, the living body score reflect the degree that eyes change in the human face image sequence,
The living body score is compared with preset threshold, if the living body score is more than or equal to preset threshold, it is determined that living body people
It is living body that face, which detects the second result, if the living body score is less than preset threshold, it is determined that living body faces detect the second result and are
Prosthese；
And/or
The third result acquiring unit, is specifically used for
Judge that each frame image is to open eyes or close one's eyes, and sentence by condition random field in the human face image sequence using classifier
Break and acted with the presence or absence of blink in the human face image sequence, if there are blink movements in the human face image sequence, it is determined that
It is living body that living body faces, which detect third result, if there is no blinks to act in the human face image sequence, it is determined that living body faces
Detection third result is prosthese.
9. device according to claim 7, which is characterized in that the final result obtains module, is specifically used for
Using temporal voting strategy, the living body faces are detected into the first result, the living body faces detect the second result and the work
Body Face datection third result is merged, and final living body faces testing result is obtained；
And/or
The final result obtains module, is specifically used for
Using metric learning metric learning strategy, by learning a discriminant function, by living body faces detection the
One result, the living body faces detect the second result and living body faces detection third result is integrated, and are finally lived
Body Face datection is as a result, and learn to the living body faces to detect the first result, the living body faces the second result of detection and institute
State the different weights of living body faces detection third result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510781590.0A CN105243378B (en) | 2015-11-13 | 2015-11-13 | Living body faces detection method and device based on eye information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510781590.0A CN105243378B (en) | 2015-11-13 | 2015-11-13 | Living body faces detection method and device based on eye information |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105243378A CN105243378A (en) | 2016-01-13 |
| CN105243378B true CN105243378B (en) | 2019-03-01 |
Family
ID=55041019
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510781590.0A Active CN105243378B (en) | 2015-11-13 | 2015-11-13 | Living body faces detection method and device based on eye information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105243378B (en) |
Families Citing this family (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107358152B (en) * | 2017-06-02 | 2020-09-08 | 广州视源电子科技股份有限公司 | Living body identification method and system |
| US20180373380A1 (en) * | 2017-06-27 | 2018-12-27 | Pixart Imaging Inc. | Optical control key, operating method thereof, and image sensor |
| CN110909695B (en) | 2017-07-29 | 2023-08-18 | Oppo广东移动通信有限公司 | Anti-counterfeiting processing method and related product |
| CN107688781A (en) * | 2017-08-22 | 2018-02-13 | 北京小米移动软件有限公司 | Face identification method and device |
| CN107992845A (en) * | 2017-12-14 | 2018-05-04 | 广东工业大学 | A kind of face recognition the method for distinguishing and device, computer equipment |
| CN108537131B (en) * | 2018-03-15 | 2022-04-15 | 中山大学 | Face recognition living body detection method based on face characteristic points and optical flow field |
| CN109191503B (en) * | 2018-08-23 | 2021-08-27 | 河海大学 | Remote sensing image change detection method and system based on conditional random field |
| CN109242832B (en) * | 2018-08-23 | 2021-08-27 | 河海大学 | Multi-temporal multispectral remote sensing image change detection method and system |
| CN111274851A (en) * | 2018-12-05 | 2020-06-12 | 上海中移信息技术有限公司 | Living body detection method and device |
| CN110127468B (en) * | 2019-04-12 | 2023-02-07 | 深圳壹账通智能科技有限公司 | Elevator control method, elevator control device, computer-readable storage medium and computer equipment |
| CN110059624B (en) * | 2019-04-18 | 2021-10-08 | 北京字节跳动网络技术有限公司 | Method and apparatus for detecting living body |
| CN110188715A (en) * | 2019-06-03 | 2019-08-30 | 广州二元科技有限公司 | A kind of video human face biopsy method of multi frame detection ballot |
| CN110598754A (en) * | 2019-08-16 | 2019-12-20 | 泰致德科技发展(深圳)有限公司 | Online monitoring method, device and equipment based on image detection and storage medium |
| CN113283370A (en) * | 2021-06-08 | 2021-08-20 | 深圳市街角电子商务有限公司 | Face living body detection method and device based on double-flow information |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103106397B (en) * | 2013-01-19 | 2016-09-21 | 华南理工大学 | Human face in-vivo detection method based on bright pupil effect |
| CN103440479B (en) * | 2013-08-29 | 2016-12-28 | 湖北微模式科技发展有限公司 | A kind of method and system for detecting living body human face |
| CN103593598B (en) * | 2013-11-25 | 2016-09-21 | 上海骏聿数码科技有限公司 | User's on-line authentication method and system based on In vivo detection and recognition of face |
| CN104794464B (en) * | 2015-05-13 | 2019-06-07 | 上海依图网络科技有限公司 | A kind of biopsy method based on relative priority |
| CN104915649B (en) * | 2015-06-04 | 2018-12-14 | 南京理工大学 | A kind of biopsy method applied to recognition of face |
| CN104966070B (en) * | 2015-06-30 | 2018-04-10 | 北京汉王智远科技有限公司 | Biopsy method and device based on recognition of face |
-
2015
- 2015-11-13 CN CN201510781590.0A patent/CN105243378B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN105243378A (en) | 2016-01-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105243378B (en) | Living body faces detection method and device based on eye information | |
| US10311289B2 (en) | Face recognition method and device and apparatus | |
| Rigas et al. | Biometric identification based on the eye movements and graph matching techniques | |
| US20150044649A1 (en) | Systems and methods for detection of behavior correlated with outside distractions in examinations | |
| EP3680794B1 (en) | Device and method for user authentication on basis of iris recognition | |
| CN109934199A (en) | A kind of Driver Fatigue Detection based on computer vision and system | |
| Rigas et al. | Current research in eye movement biometrics: An analysis based on BioEye 2015 competition | |
| Deravi et al. | Gaze trajectory as a biometric modality | |
| Rigas et al. | Biometric recognition via probabilistic spatial projection of eye movement trajectories in dynamic visual environments | |
| CN101216884A (en) | A method and system for face authentication | |
| US11663853B2 (en) | Iris authentication device, iris authentication method, and recording medium | |
| US11682236B2 (en) | Iris authentication device, iris authentication method and recording medium | |
| Noman et al. | Mobile-based eye-blink detection performance analysis on android platform | |
| KR20200134160A (en) | Living-body detection method and apparatus for face, electronic device ad computer readable medium | |
| Furnari et al. | An experimental analysis of saliency detection with respect to three saliency levels | |
| Stan et al. | Eye-gaze tracking method driven by raspberry PI applicable in automotive traffic safety | |
| Singh et al. | Detection of stress, anxiety and depression (SAD) in video surveillance using ResNet-101 | |
| Ma et al. | VIP: A unifying framework for computational eye-gaze research | |
| Shen et al. | Iritrack: Face presentation attack detection using iris tracking | |
| Adamiak et al. | Liveness detection in remote biometrics based on gaze direction estimation | |
| Zhang et al. | An effective driver fatigue monitoring system | |
| Li et al. | A real-time system for monitoring driver fatigue | |
| Bhardwaj et al. | Liveness detection using eye blink a case study | |
| P Mathai | A New Proposal for Smartphone-Based Drowsiness Detection and Warning System for Automotive Drivers | |
| NAGIT et al. | Facial Emotion Recognition and Detection Application. |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |