CN110532933A - A kind of living body faces detection head pose returns the acquisition methods and detection method of device - Google Patents
A kind of living body faces detection head pose returns the acquisition methods and detection method of device Download PDFInfo
- Publication number
- CN110532933A CN110532933A CN201910788598.8A CN201910788598A CN110532933A CN 110532933 A CN110532933 A CN 110532933A CN 201910788598 A CN201910788598 A CN 201910788598A CN 110532933 A CN110532933 A CN 110532933A
- Authority
- CN
- China
- Prior art keywords
- face
- data
- characteristic point
- human face
- dimensional
- 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
- 238000000034 method Methods 0.000 title claims abstract description 49
- 238000001514 detection method Methods 0.000 title claims abstract description 42
- 239000011159 matrix material Substances 0.000 claims abstract description 71
- 150000001875 compounds Chemical class 0.000 claims abstract description 24
- 238000012545 processing Methods 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims abstract description 4
- 210000003128 head Anatomy 0.000 claims description 64
- 210000000887 face Anatomy 0.000 claims description 23
- 238000004364 calculation method Methods 0.000 claims description 12
- 238000004422 calculation algorithm Methods 0.000 claims description 8
- 238000012937 correction Methods 0.000 claims description 8
- 238000003384 imaging method Methods 0.000 claims description 6
- 230000002441 reversible effect Effects 0.000 claims description 6
- 238000004088 simulation Methods 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 6
- 230000017105 transposition Effects 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 4
- 238000013519 translation Methods 0.000 claims description 4
- 230000008921 facial expression Effects 0.000 claims description 3
- 210000001747 pupil Anatomy 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 230000001815 facial effect Effects 0.000 abstract description 9
- 238000001727 in vivo Methods 0.000 abstract description 6
- 238000007781 pre-processing Methods 0.000 abstract description 4
- 230000006872 improvement Effects 0.000 description 8
- 230000008901 benefit Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000003993 interaction Effects 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/285—Selection of pattern recognition techniques, e.g. of classifiers in a multi-classifier system
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/60—Rotation of whole images or parts thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/64—Three-dimensional objects
-
- 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
- G06V40/166—Detection; Localisation; Normalisation using acquisition arrangements
-
- 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
- G06V40/18—Eye characteristics, e.g. of the iris
-
- 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)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Human Computer Interaction (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Ophthalmology & Optometry (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of living body faces detection head poses to return the acquisition methods and detection method of device, including obtains human face data, formed with compound standardized human face three-dimensional model data;Obtain training sample set;With the two-dimensional coordinate plane set of acquisition being made of characteristic point two-dimensional coordinate, corresponding Eulerian angles and the spin matrix comprising the corresponding relationship between the corresponding Eulerian angles of each two-dimensional coordinate plane are obtained;It obtains head pose and returns device;The present invention returns device by constructing high-precision head pose, and by carrying out corresponding pre-processing to face two dimensional model data, human face three-dimensional model data and iris data, and further progress facial modeling and data processing, it can be improved the accuracy and science of data, the method of the present invention is conducive to effectively carry out recognition of face, the speed and precision of facial modeling is greatly improved, increases the success rate of face In vivo detection.
Description
Technical field
The present invention relates to the acquisition sides that image recognition technology more particularly to a kind of living body faces detection head pose return device
Method and detection method.
Background technique
With the arrival of big data era, personal information security problem is increasingly serious, the recognition of face based on image procossing
It is widely used with detection technique.However, at present human face detection tech both for the lesser facial image of quantity, with
Going deep into for big data concept, image big data processing will propose requirements at the higher level to human face detection tech.Moreover, most of faces
Detection scheme is all based on the direct extraction to human face image information, and no interactions, anti-attack ability is poor, for example, photo, video,
Model camouflage, this just proposes requirement to face In vivo detection, and there has been no mature human face in-vivo detection methods at present, and not yet
It is disclosed in the recurrence device that posture recurrence is carried out in recognition of face, therefore, the present invention proposes a kind of living body faces detection head pose
The acquisition methods and detection method of device are returned, to solve shortcoming in the prior art.
Summary of the invention
In view of the above-mentioned problems, the present invention proposes that a kind of living body faces detection head pose returns acquisition methods and the detection of device
Method returns device by constructing high-precision head pose, and by face two dimensional model data, human face three-dimensional model number
Accordingly and iris data carries out corresponding pre-processing and further progress facial modeling and data processing,
It can be improved the accuracy and science of data, the method for the present invention is conducive to effectively carry out recognition of face, be greatly improved
The speed and precision of facial modeling increases the success rate of face In vivo detection.
The present invention proposes that a kind of living body faces detection head pose returns the acquisition methods of device, comprising the following steps:
A, face two dimensional model data, human face three-dimensional model data and iris data are obtained, then to the face of acquisition
Two dimensional model data, human face three-dimensional model data and iris data carry out screening out processing, remove face two dimensional model data, face
The repeated data of low resolution and appearance in three-dimensional modeling data removes the damaged data in iris data;
B, a certain number of characteristic points are selected in face two dimensional model data, is then selected in human face three-dimensional model data
Identical characteristic point, and characteristic point is mapped, then being formed has compound standardized human face three-dimensional model data;
C, it is pre-processed to compound standardized human face three-dimensional model data, including is filtered denoising, face
Region detection, face are cut and posture correction, then with screen out the iris data after damaged data and merge, as training sample
This collection;
D, selection has the human face characteristic point in compound standardized human face three-dimensional model data;And pass through camera imaging mould
The human face characteristic point is done when arbitrary Random-Rotation and random translation exchange image projecting to camera imaging by type in space
In plane, to obtain the two-dimensional coordinate plane set being made of characteristic point two-dimensional coordinate, and obtains the two-dimensional coordinate and put down
Corresponding Eulerian angles of each two-dimensional coordinate plane and corresponding comprising each two-dimensional coordinate plane in the set of face
The spin matrix of corresponding relationship between Eulerian angles;
E, according to the two-dimensional coordinate of the acquisition and the spin matrix or two-dimensional coordinate and institute according to the acquisition
Eulerian angles are stated, head pose is obtained by least square method and returns device, when the head pose returns device as spin matrix recurrence
When device, in the step E further include:
E1: according to the characteristic point two-dimensional coordinate and the spin matrix, spin matrix is obtained by least square method and is returned
Return device;
Wherein, the spin matrix returns shown in the calculation formula such as formula (1) of device:
W=(STS+λE)-1STR (1)
In formula (1), S indicates the two-dimensional coordinate matrix of characteristic point, STThe two dimension of characteristic point after representing matrix transposition is sat
Matrix is marked, λ indicates that simulation parameter, E indicate that unit matrix, λ E indicate to guarantee that reversible loose item, R indicate the spin moment
Battle array.
Further improvement lies in that: it is formed in the step B after there are compound standardized human face three-dimensional model data, is also wrapped
It includes and is normalized to compound standardized human face three-dimensional model data, there is compound standardized face with described
The distance and axis of orientation benchmark the most of two pupils in three-dimensional modeling data carry out horizontal rotation and transversely and horizontally scaling,
Human face region is cropped to same size, then uses the dense mark for carrying out three-dimensional face to it for method based on plane template
Standardization.
Further improvement lies in that: when human face region detects in the step C, from compound standardized face three-dimensional mould
Detect human face region in type, obtain human face region depth value, then using the face regional depth value by human face region from institute
It states with being split in compound standardized human face three-dimensional model;When face is cut, carried out using the method for contour feature point
Face is cut, and then carries out a subthreshold cutting further according to empirical value;When posture is corrected, rectified using the posture based on plane fitting
Normal operation method carries out posture correction.
Further improvement lies in that: when head pose recurrence device is that Eulerian angles return device in the step E, the step E packet
It includes:
E2: according to the characteristic point two-dimensional coordinate and the Eulerian angles, obtaining Eulerian angles by least square method and return device,
The Eulerian angles return shown in the calculation formula such as formula (2) of device:
W'=(STS+λE)-1STθ (2)
In formula (2), S indicates the two-dimensional coordinate matrix of characteristic point, STThe two dimension of characteristic point after representing matrix transposition is sat
Matrix is marked, λ indicates that simulation parameter, E indicate that unit matrix, λ E indicate to guarantee that reversible loose item, θ indicate Eulerian angles angle
Value.
A kind of living body faces detection method returning device based on head pose, comprising the following steps:
H, the image for the head pose that user makes according to the instruction that terminal issues, including two dimensional image and three-dimensional figure are obtained
Picture recycles the iris data of iris capturing equipment acquisition;
I, according to the two dimensional image and 3-D image and iris data, face frame is obtained by adaboost algorithm;
G, the human face characteristic point coordinate in the face frame is positioned by supervised gradient decent method;
K, the human face characteristic point is subjected to centralization and normalized;
L, according to treated characteristic point data, pass through the head pose and return device and obtain head angle;Work as judgement
When the head angle value of acquisition is within preset threshold, identify successfully.
Further improvement lies in that: when it includes that spin matrix returns device that the head pose, which returns device, obtained described in step L
The step of head angle includes:
L1: device is returned according to treated characteristic point data and the spin matrix and obtains the spin matrix returned;
L2: according to the spin matrix of the recurrence, three Europe of head space posture are obtained by way of mathematical analysis
Draw angle.
Further improvement lies in that: the calculation formula such as public affairs that device obtains the spin matrix returned are returned according to the spin matrix
Shown in formula (3):
hW(Si(x, y))=WSi(x,y)+ξ (3)
In formula (3), hW(Si(x, y)) indicate that the spin matrix after returning, W indicate that spin matrix returns device, ξ indicates to miss
Poor item, SiThe two-dimensional coordinate matrix of (x, y) expression characteristic point.
Further improvement lies in that: when the head pose recurrence device includes that Eulerian angles return device, head is obtained described in step L
The step of angle specifically: according to treated characteristic point data, return device according to the Eulerian angles and obtain head space posture
Three Eulerian angles, the Eulerian angles return the calculation formula such as formula (4) that device obtains three Eulerian angles of head space posture
It is shown:
θ=W'Si(x,y)+ξ (4)
In formula (4), W' indicates that Eulerian angles return device, and ξ indicates error term, Si(x, y) indicates characteristic point coordinates matrix, ξ
Indicate error term.
A kind of living body faces detection method, comprising the following steps:
M, the two dimensional image and 3-D image of the facial expression that user makes according to end command are obtained, and utilizes iris
Acquire the iris data of equipment acquisition;
N, according to the two dimensional image, 3-D image and iris data, face frame is obtained by adaboost algorithm;
O, the human face characteristic point coordinate in the face frame is positioned by supervised gradient decent method;
P, the characteristic point of positioning is subjected to linear transformation;The characteristic information value of characteristic point after judging linear transformation is pre-
If when within the scope of characteristic threshold value, then identifying success.
The invention has the benefit that returning device by constructing high-precision head pose, and by using interaction
Mode, user make corresponding human face posture according to the instruction that terminal issues, and the camera and iris capturing equipment of terminal obtain
User image data and iris data, and by face two dimensional model data, human face three-dimensional model data and iris data
Corresponding pre-processing and further progress facial modeling and data processing are carried out, can be improved data
Accuracy and science are conducive to going on smoothly for finishing operations, if should treated data value in preset data threshold value,
Recognition of face success;The method of the present invention is conducive to effectively carry out recognition of face, and facial modeling is greatly improved
Speed and precision increases the success rate of face In vivo detection.
Detailed description of the invention
Fig. 1 is that a kind of living body faces provided in the embodiment of the present invention one detect head pose recurrence device acquisition methods process
Schematic diagram;
Fig. 2 is a kind of human face three-dimensional model Random-Rotation perspective view provided in the embodiment of the present invention one;
Fig. 3 is a kind of living body faces detection method flow diagram provided by Embodiment 2 of the present invention;
Fig. 4 is a kind of living body faces detection method flow diagram that the embodiment of the present invention three provides.
Specific embodiment
In order to deepen the understanding of the present invention, the present invention is further described below in conjunction with embodiment, the present embodiment
For explaining only the invention, it is not intended to limit the scope of the present invention..
Embodiment one
According to Fig. 1,2, the present embodiment proposes that a kind of living body faces detection head pose returns the acquisition methods of device, packet
Include following steps:
A, face two dimensional model data, human face three-dimensional model data and iris data are obtained, then to the face of acquisition
Two dimensional model data, human face three-dimensional model data and iris data carry out screening out processing, remove face two dimensional model data, face
The repeated data of low resolution and appearance in three-dimensional modeling data removes the damaged data in iris data;
B, a certain number of characteristic points are selected in face two dimensional model data, is then selected in human face three-dimensional model data
Identical characteristic point, and characteristic point is mapped, then being formed has compound standardized human face three-dimensional model data, then right
It is normalized with compound standardized human face three-dimensional model data, has compound standardized face three-dimensional with described
The distance and axis of orientation benchmark the most of two pupils in model data carry out horizontal rotation and transversely and horizontally scaling, by people
Face region is cropped to same size, then uses the dense standard for carrying out three-dimensional face to it for method based on plane template
Change;
C, it is pre-processed to compound standardized human face three-dimensional model data, including is filtered denoising, face
Region detection, face are cut and posture correction is examined from compound standardized human face three-dimensional model when human face region detects
Human face region is measured, human face region depth value is obtained, then has human face region from described using the face regional depth value
It is split in compound standardized human face three-dimensional model;When face is cut, face sanction is carried out using the method for contour feature point
It cuts, then carries out a subthreshold cutting further according to empirical value;When posture is corrected, the posture correction algorithm based on plane fitting is utilized
Carry out posture correction, then with screen out the iris data after damaged data and merge, as training sample set;
D, selection has the human face characteristic point in compound standardized human face three-dimensional model data;And pass through camera imaging mould
The human face characteristic point is done when arbitrary Random-Rotation and random translation exchange image projecting to camera imaging by type in space
In plane, to obtain the two-dimensional coordinate plane set being made of characteristic point two-dimensional coordinate, and obtains the two-dimensional coordinate and put down
Corresponding Eulerian angles of each two-dimensional coordinate plane and corresponding comprising each two-dimensional coordinate plane in the set of face
The spin matrix of corresponding relationship between Eulerian angles;
E, according to the two-dimensional coordinate of the acquisition and the spin matrix or two-dimensional coordinate and institute according to the acquisition
Eulerian angles are stated, head pose is obtained by least square method and returns device, when the head pose returns device as spin matrix recurrence
When device, in the step E further include:
E1: according to the characteristic point two-dimensional coordinate and the spin matrix, spin matrix is obtained by least square method and is returned
Return device;
Wherein, the spin matrix returns shown in the calculation formula such as formula (1) of device:
W=(STS+λE)-1STR (1)
In formula (1), S indicates the two-dimensional coordinate matrix of characteristic point, STThe two dimension of characteristic point after representing matrix transposition is sat
Matrix is marked, λ indicates that simulation parameter, E indicate that unit matrix, λ E indicate to guarantee that reversible loose item, R indicate the spin moment
Battle array;
When head pose recurrence device is that Eulerian angles return device, the step E includes:
E2: according to the characteristic point two-dimensional coordinate and the Eulerian angles, obtaining Eulerian angles by least square method and return device,
The Eulerian angles return shown in the calculation formula such as formula (2) of device:
W'=(STS+λE)-1STθ (2)
In formula (2), S indicates the two-dimensional coordinate matrix of characteristic point, STThe two dimension of characteristic point after representing matrix transposition is sat
Matrix is marked, λ indicates that simulation parameter, E indicate that unit matrix, λ E indicate to guarantee that reversible loose item, θ indicate Eulerian angles angle
Value.
Embodiment two
According to Fig.3, a kind of living body faces detection method that device is returned based on head pose of the present embodiment, including it is following
Step:
H, the image for the head pose that user makes according to the instruction that terminal issues, including two dimensional image and three-dimensional figure are obtained
Picture recycles the iris data of iris capturing equipment acquisition;
I, according to the two dimensional image and 3-D image and iris data, face frame is obtained by adaboost algorithm;
G, the human face characteristic point coordinate in the face frame is positioned by supervised gradient decent method;
K, the human face characteristic point is subjected to centralization and normalized;
L, according to treated characteristic point data, pass through the head pose and return device and obtain head angle;Work as judgement
When the head angle value of acquisition is within preset threshold, identify successfully;
When it includes that spin matrix returns device that the head pose, which returns device, described in step L the step of acquisition head angle
Include:
L1: device is returned according to treated characteristic point data and the spin matrix and obtains the spin matrix returned, is returned
Spin matrix calculation formula such as formula (3) shown in:
hW(Si(x, y))=WSi(x,y)+ξ (3)
In formula (3), hW(Si(x, y)) indicate that the spin matrix after returning, W indicate that spin matrix returns device, ξ indicates to miss
Poor item, SiThe two-dimensional coordinate matrix of (x, y) expression characteristic point;
L2: according to the spin matrix of the recurrence, three Europe of head space posture are obtained by way of mathematical analysis
Draw angle;
The step of head pose returns device when including that Eulerian angles return device, acquisition head angle described in step L is specific
Are as follows: according to treated characteristic point data, three Eulerian angles that device obtains head space posture, institute are returned according to the Eulerian angles
Eulerian angles are stated to return shown in the calculation formula such as formula (4) for three Eulerian angles that device obtains head space posture:
θ=W'Si(x,y)+ξ (4)
In formula (4), W' indicates that Eulerian angles return device, and ξ indicates error term, Si(x, y) indicates characteristic point coordinates matrix, ξ
Indicate error term;
Above content is tested, obtains experimental result and analysis:
The present embodiment method is tested on Biwi Kinect database, Biwi Kinect database includes 20
The RGBD image of people (14 males, 6 women) head rotation different directions;There are 24 sections of video sequence datas in database, this
A few peoples recorded twice in 20 people;Head position and rotation angle in all images are all demarcated;Database
Translation and rotation calibrated error are between 1mm and 1 °;This experiment lists other 5 kinds of head pose estimation methods, with proposition
Two kinds of homing methods compare on the data set, obtain as shown in the results summarized in table 1:
Table 1
As can be seen from the table, the method that the present embodiment proposes is significantly improved in precision, while comparing us can
To find second of method for directly returning angle in the present embodiment, slightly has advantage in precision, can be seen that from above-mentioned experiment
The effect performance on Biwi data set of the present embodiment method is good, and estimated accuracy is compared to existing method and increases, simultaneously
It can be handled in real time in mobile terminal.
Embodiment three
According to Fig.4, a kind of living body faces detection method of the present embodiment, comprising the following steps:
M, the two dimensional image and 3-D image of the facial expression that user makes according to end command are obtained, and utilizes iris
Acquire the iris data of equipment acquisition;
N, according to the two dimensional image, 3-D image and iris data, face frame is obtained by adaboost algorithm;
O, the human face characteristic point coordinate in the face frame is positioned by supervised gradient decent method;
P, the characteristic point of positioning is subjected to linear transformation;The characteristic information value of characteristic point after judging linear transformation is pre-
If when within the scope of characteristic threshold value, then identifying success.
The invention has the benefit that returning device by constructing high-precision head pose, and by using interaction
Mode, user make corresponding human face posture according to the instruction that terminal issues, and the camera and iris capturing equipment of terminal obtain
User image data and iris data, and by face two dimensional model data, human face three-dimensional model data and iris data
Corresponding pre-processing and further progress facial modeling and data processing are carried out, can be improved data
Accuracy and science are conducive to going on smoothly for finishing operations, if should treated data value in preset data threshold value,
Recognition of face success;The method of the present invention is conducive to effectively carry out recognition of face, and facial modeling is greatly improved
Speed and precision increases the success rate of face In vivo detection.
The basic principles, main features and advantages of the invention have been shown and described above.The technical staff of the industry should
Understand, the present invention is not limited to the above embodiments, and the above embodiments and description only describe originals of the invention
Reason, without departing from the spirit and scope of the present invention, various changes and improvements may be made to the invention, these changes and improvements
It all fall within the protetion scope of the claimed invention.The claimed scope of the invention is by appended claims and its equivalent circle
It is fixed.
Claims (9)
1. the acquisition methods that a kind of living body faces detection head pose returns device, which comprises the following steps:
A, face two dimensional model data, human face three-dimensional model data and iris data are obtained, then to the face two dimension of acquisition
Model data, human face three-dimensional model data and iris data carry out screening out processing, and removal face two dimensional model data, face are three-dimensional
The repeated data of low resolution and appearance in model data removes the damaged data in iris data;
B, a certain number of characteristic points are selected in face two dimensional model data, is then selected in human face three-dimensional model data identical
Characteristic point, and characteristic point is mapped, then being formed has compound standardized human face three-dimensional model data;
C, it is pre-processed to compound standardized human face three-dimensional model data, including is filtered denoising, human face region
Detection, face are cut and posture correction, then with screen out the iris data after damaged data and merge, as training sample
Collection;
D, selection has the human face characteristic point in compound standardized human face three-dimensional model data;And it is incited somebody to action by camera imaging model
The human face characteristic point does when arbitrary Random-Rotation and random translation exchange image projecting to camera imaging plane in space
On, to obtain the two-dimensional coordinate plane set being made of characteristic point two-dimensional coordinate, and obtain the two-dimensional coordinate planar set
The corresponding Eulerian angles of each two-dimensional coordinate plane and the Euler corresponding comprising each two-dimensional coordinate plane in conjunction
The spin matrix of corresponding relationship between angle;
E, according to the two-dimensional coordinate of the acquisition and the spin matrix or two-dimensional coordinate and the Europe according to the acquisition
Angle is drawn, head pose is obtained by least square method and returns device, when it is that spin matrix returns device that the head pose, which returns device,
In the step E further include:
E1: according to the characteristic point two-dimensional coordinate and the spin matrix, spin matrix is obtained by least square method and returns device;
Wherein, the spin matrix returns shown in the calculation formula such as formula (1) of device:
W=(STS+λE)-1STR (1)
In formula (1), S indicates the two-dimensional coordinate matrix of characteristic point, STThe two-dimensional coordinate square of characteristic point after representing matrix transposition
Battle array, λ indicate that simulation parameter, E indicate that unit matrix, λ E indicate to guarantee that reversible loose item, R indicate the spin matrix.
2. the acquisition methods that a kind of living body faces detection head pose according to claim 1 returns device, it is characterised in that:
It is formed in the step B after there are compound standardized human face three-dimensional model data, further includes to compound standardized people
Face three-dimensional model data are normalized, with two pupils in compound standardized human face three-dimensional model data
Distance and axis of orientation benchmark the most, rotate horizontally and transversely and horizontally scaling, human face region is cropped to identical big
It is small, then use the dense standardization for carrying out three-dimensional face to it for method based on plane template.
3. the acquisition methods that a kind of living body faces detection head pose according to claim 1 returns device, it is characterised in that:
When human face region detects in the step C, human face region is detected from compound standardized human face three-dimensional model, is obtained
Then human face region is had compound standardized face three from described using the face regional depth value by human face region depth value
It is split in dimension module;When face is cut, face cutting is carried out using the method for contour feature point, then further according to empirical value
Carry out a subthreshold cutting;When posture is corrected, posture correction is carried out using the posture correction algorithm based on plane fitting.
4. the acquisition methods that a kind of living body faces detection head pose according to claim 1 returns device, it is characterised in that:
When head pose recurrence device is that Eulerian angles return device in the step E, the step E includes:
E2: according to the characteristic point two-dimensional coordinate and the Eulerian angles, obtaining Eulerian angles by least square method and return device, described
Eulerian angles return shown in the calculation formula such as formula (2) of device:
W'=(STS+λE)-1STθ (2)
In formula (2), S indicates the two-dimensional coordinate matrix of characteristic point, STThe two-dimensional coordinate square of characteristic point after representing matrix transposition
Battle array, λ indicate that simulation parameter, E indicate that unit matrix, λ E indicate to guarantee that reversible loose item, θ indicate Eulerian angles angle value.
5. a kind of living body faces detection method for being returned device based on any one of the claim 1-4 head pose, feature are existed
In, comprising the following steps:
H, the image for the head pose that user makes according to the instruction that terminal issues, including two dimensional image and 3-D image are obtained,
Recycle the iris data of iris capturing equipment acquisition;
I, according to the two dimensional image and 3-D image and iris data, face frame is obtained by adaboost algorithm;
G, the human face characteristic point coordinate in the face frame is positioned by supervised gradient decent method;
K, the human face characteristic point is subjected to centralization and normalized;
L, according to treated characteristic point data, pass through the head pose and return device and obtain head angle;When judgement obtains
Head angle value within preset threshold when, identify successfully.
6. the living body faces detection method that head pose according to claim 5 returns device, it is characterised in that: when the head
Portion's posture returns device when including that spin matrix returns device, and the step of acquisition head angle described in step L includes:
L1: device is returned according to treated characteristic point data and the spin matrix and obtains the spin matrix returned;
L2: according to the spin matrix of the recurrence, three Eulerian angles of head space posture are obtained by way of mathematical analysis.
7. the living body faces detection method that head pose according to claim 6 returns device, it is characterised in that: according to described
Spin matrix returns device and obtains shown in the calculation formula such as formula (3) of the spin matrix returned:
hW(Si(x, y))=WSi(x,y)+ξ (3)
In formula (3), hW(Si(x, y)) indicate that the spin matrix after returning, W indicate that spin matrix returns device, ξ indicates error term,
SiThe two-dimensional coordinate matrix of (x, y) expression characteristic point.
8. the living body faces detection method that head pose according to claim 6 returns device, it is characterised in that: the head
Posture returns device when including that Eulerian angles return device, described in step L the step of acquisition head angle specifically: special according to treated
Point data is levied, returns three Eulerian angles that device obtains head space posture according to the Eulerian angles, the Eulerian angles return device and obtain
Shown in the calculation formula such as formula (4) for taking three Eulerian angles of head space posture:
θ=W'Si(x,y)+ξ (4)
In formula (4), W' indicates that Eulerian angles return device, and ξ indicates error term, Si(x, y) indicates that characteristic point coordinates matrix, ξ indicate to miss
Poor item.
9. a kind of living body faces detection method, which comprises the following steps:
M, the two dimensional image and 3-D image of the facial expression that user makes according to end command are obtained, and utilizes iris capturing
The iris data of equipment acquisition;
N, according to the two dimensional image, 3-D image and iris data, face frame is obtained by adaboost algorithm;
O, the human face characteristic point coordinate in the face frame is positioned by supervised gradient decent method;
P, the characteristic point of positioning is subjected to linear transformation;The characteristic information value of characteristic point after judging linear transformation is in default spy
When levying within threshold range, then success is identified.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910788598.8A CN110532933A (en) | 2019-08-26 | 2019-08-26 | A kind of living body faces detection head pose returns the acquisition methods and detection method of device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910788598.8A CN110532933A (en) | 2019-08-26 | 2019-08-26 | A kind of living body faces detection head pose returns the acquisition methods and detection method of device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110532933A true CN110532933A (en) | 2019-12-03 |
Family
ID=68662807
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910788598.8A Pending CN110532933A (en) | 2019-08-26 | 2019-08-26 | A kind of living body faces detection head pose returns the acquisition methods and detection method of device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110532933A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111898553A (en) * | 2020-07-31 | 2020-11-06 | 成都新潮传媒集团有限公司 | Method and device for distinguishing virtual image personnel and computer equipment |
CN112416126A (en) * | 2020-11-18 | 2021-02-26 | 青岛海尔科技有限公司 | Page rolling control method and device, storage medium and electronic equipment |
CN113326814A (en) * | 2021-02-22 | 2021-08-31 | 王先峰 | Face recognition equipment based on 5G framework |
CN115147902A (en) * | 2022-06-30 | 2022-10-04 | 北京百度网讯科技有限公司 | Training method and device for human face living body detection model and computer program product |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101438011B1 (en) * | 2013-11-08 | 2014-09-04 | 수원대학교산학협력단 | Three-dimensional face recognition system using 3d scanner |
CN106355147A (en) * | 2016-08-26 | 2017-01-25 | 张艳 | Acquiring method and detecting method of live face head pose detection regression apparatus |
CN108985220A (en) * | 2018-07-11 | 2018-12-11 | 腾讯科技(深圳)有限公司 | A kind of face image processing process, device and storage medium |
CN109325462A (en) * | 2018-10-11 | 2019-02-12 | 深圳斐视沃德科技有限公司 | Recognition of face biopsy method and device based on iris |
-
2019
- 2019-08-26 CN CN201910788598.8A patent/CN110532933A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101438011B1 (en) * | 2013-11-08 | 2014-09-04 | 수원대학교산학협력단 | Three-dimensional face recognition system using 3d scanner |
CN106355147A (en) * | 2016-08-26 | 2017-01-25 | 张艳 | Acquiring method and detecting method of live face head pose detection regression apparatus |
CN108985220A (en) * | 2018-07-11 | 2018-12-11 | 腾讯科技(深圳)有限公司 | A kind of face image processing process, device and storage medium |
CN109325462A (en) * | 2018-10-11 | 2019-02-12 | 深圳斐视沃德科技有限公司 | Recognition of face biopsy method and device based on iris |
Non-Patent Citations (1)
Title |
---|
傅泽华: "基于RGB-D数据的三维人脸建模及标准化", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111898553A (en) * | 2020-07-31 | 2020-11-06 | 成都新潮传媒集团有限公司 | Method and device for distinguishing virtual image personnel and computer equipment |
CN111898553B (en) * | 2020-07-31 | 2022-08-09 | 成都新潮传媒集团有限公司 | Method and device for distinguishing virtual image personnel and computer equipment |
CN112416126A (en) * | 2020-11-18 | 2021-02-26 | 青岛海尔科技有限公司 | Page rolling control method and device, storage medium and electronic equipment |
CN113326814A (en) * | 2021-02-22 | 2021-08-31 | 王先峰 | Face recognition equipment based on 5G framework |
CN115147902A (en) * | 2022-06-30 | 2022-10-04 | 北京百度网讯科技有限公司 | Training method and device for human face living body detection model and computer program product |
CN115147902B (en) * | 2022-06-30 | 2023-11-07 | 北京百度网讯科技有限公司 | Training method, training device and training computer program product for human face living body detection model |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110532933A (en) | A kind of living body faces detection head pose returns the acquisition methods and detection method of device | |
CN107403168B (en) | Face recognition system | |
CN105740780B (en) | Method and device for detecting living human face | |
US8331619B2 (en) | Image processing apparatus and image processing method | |
WO2019056988A1 (en) | Face recognition method and apparatus, and computer device | |
CN105740778B (en) | Improved three-dimensional human face in-vivo detection method and device | |
CN109087261B (en) | Face correction method based on unlimited acquisition scene | |
CN107590430A (en) | Biopsy method, device, equipment and storage medium | |
CN106355147A (en) | Acquiring method and detecting method of live face head pose detection regression apparatus | |
CN109035330A (en) | Cabinet approximating method, equipment and computer readable storage medium | |
CN106372629A (en) | Living body detection method and device | |
CN104318603A (en) | Method and system for generating 3D model by calling picture from mobile phone photo album | |
CN109948400A (en) | It is a kind of to be able to carry out the smart phone and its recognition methods that face characteristic 3D is identified | |
CN111091075A (en) | Face recognition method and device, electronic equipment and storage medium | |
CN105913013A (en) | Binocular vision face recognition algorithm | |
CN113313097B (en) | Face recognition method, terminal and computer readable storage medium | |
CN108268825A (en) | Three-dimensional face tracking and expression recognition system based on mobile holder | |
CN106570447A (en) | Face photo sunglass automatic removing method based on gray histogram matching | |
CN111488943A (en) | Face recognition method and device | |
CN107749084A (en) | A kind of virtual try-in method and system based on 3-dimensional reconstruction technology | |
CN115797451A (en) | Acupuncture point identification method, device and equipment and readable storage medium | |
CN1870005A (en) | Image recognition apparatus and method | |
Fortin et al. | Handling occlusions in real-time augmented reality: dealing with movable real and virtual objects | |
CN110188630A (en) | A kind of face identification method and camera | |
CN111881841B (en) | Face detection and recognition method based on binocular vision |
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: 20191203 |