CN110096992A - A kind of face identification method indicating non-linear fusion Pasteur coefficient based on collaboration - Google Patents
A kind of face identification method indicating non-linear fusion Pasteur coefficient based on collaboration Download PDFInfo
- Publication number
- CN110096992A CN110096992A CN201910342740.6A CN201910342740A CN110096992A CN 110096992 A CN110096992 A CN 110096992A CN 201910342740 A CN201910342740 A CN 201910342740A CN 110096992 A CN110096992 A CN 110096992A
- Authority
- CN
- China
- Prior art keywords
- sample
- virtual
- test sample
- formula
- training sample
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of face identification methods that non-linear fusion Pasteur coefficient is indicated based on collaboration, this method, which introduces, reinforces the more low and higher equicohesive pixel of original image, and reduce other pixels, new virtual sample is generated, then with virtual training sample and original training sample difference linear expression virtual test sample and test sample；And calculate Pasteur's coefficient similarity between test sample and training sample；By Pasteur's coefficient similarity and Euclidean distance non-linear fusion, test sample is finally determined as to the classification of the smallest training sample of residual values.The invention has the benefit that the histogram information of Pasteur's coefficient similarity, which is introduced into collaboration, to be indicated to play the role of a kind of supplement to Euclidean distance in algorithm；And the merging of virtual sample and original sample, merging for Euclidean distance and the histogram information of Pasteur's coefficient similarity are allowed between two kinds of information by the way of non-linear fusion and can preferably be combined, so that the accuracy of image classification is higher.
Description
Technical field
The present invention relates to technical field of image processing, indicate non-linear fusion bar based on collaboration in particular to one kind
The face identification method of family name's coefficient.
Background technique
With the fast development of modern information technologies, the technology for carrying out authentication has gone to biological characteristic level.It is modern
Biological identification technology mainly passes through computer and is intimately associated with high-tech means, utilizes the intrinsic physiological property of human body and row
It is characterized to carry out the identification of personal identification.Wherein face is the set of a mode for including abundant information, is that the mankind are mutual
One of mutually dialectical and outstanding feature of identification, compared with other human body biological characteristics such as fingerprint, iris, voice, recognition of face is more
Add it is direct, friendly, without interfering the normal behaviour of people that can admirably achieve recognition effect.
Recognition of face identification, access control, in terms of have a wide range of applications, be present mode
One research hotspot of identification and artificial intelligence field.And due to can arbitrarily be placed using the equipment of face recognition technology,
The placement concealment of equipment is very good, can non-contact quick lock in target identification object, therefore face recognition technology quilt at a distance
Foreign countries are widely applied in public's security system, and application is in large scale.But in practical applications, we can not often obtain greatly
The training sample of amount extracts for training for classifying and knowing another characteristic.This aspect is because of face identification system
Memory space it is limited, a large amount of training sample can not be accommodated；On the other hand be because in a short time for same target without
Method obtains its multiple face sample photo and is used as training.And limited training sample can not comprehensively express face in illumination item
Expression shape change and change in location under part, so being difficult to improve the accuracy of recognition of face.Therefore, it solves to have in training sample
In the case where limit, improving the problem of quickly identifying the discrimination of facial image in the short time is particularly important.
Summary of the invention
To solve the above problems, the purpose of the present invention is to provide one kind to indicate non-linear fusion Pasteur system based on collaboration
Several face identification method, for quickly identifying facial image, it is right under real scene to meet under training sample limited circumstances
The requirement for the accuracy that small sample is quickly identified.
To achieve the above object, the present invention provides a kind of faces that non-linear fusion Pasteur coefficient is indicated based on collaboration
Recognition methods, this method comprises: the following steps are included:
Step 1: the intensity by enhancing the pixel of the moderate strength of original image reduces the intensity of other pixels, generates
Virtual sample then by reinforcing the more low and higher equicohesive pixel of original image, and reduces other pixels, generates
New virtual sample, selected part original image is as original training sample, remaining original image is as test sample；
Step 2: algorithm being indicated by collaboration, with virtual training sample, new virtual training sample and original training sample
Linear expression virtual test sample, new virtual test sample and test sample respectively；
Step 3: calculating virtual test sample, new virtual test sample, original test sample and virtual training sample,
Pasteur's coefficient similarity between new virtual training sample, original training sample；
Step 4: the Pasteur's coefficient similarity and Euclidean distance non-linear fusion that step 3 is calculated；
Step 5: being determined according to the calculated result of step 4 to virtual test sample, new virtual test sample and original survey
Sample is originally classified, and is merged, is differentiated according to fused residual error, residual values by class to all obtained residual errors
The smallest differentiation classification is original test sample, virtual test sample and new virtual test sample generic.
It is further improved as of the invention, in step 1, the image of new virtual sample is expressed as,
Wherein, IijIndicate the pixel value of original image ith row and jth column, JijIndicate new the i-th row of virtual sample image
With the pixel value of jth column.
As further improvement of the invention, step 2 is specifically included:
Step 201: image array is become into column vector,Indicate test sample Z-direction amount form,Indicate original training
Sample xiVector form, and meet formula (1)
Wherein, aiIndicate coefficient；
If withThen formula (1) can be write as formula (2)
Step 202: passing throughA is calculated, wherein λ indicates one
Positive integer (λ=0.01), I are a unit matrix；The similarity of class training sample and test sample is calculated by formula (3),
Indicate residual error:
Wherein, dkFor the expression residual error of every a kind of training sample and test sample,For test sample Z-direction amount form,
Indicate original training sample xiVector form.
Step 203: by virtual training sample X '1Carrying out collaboration indicates that algorithm, linear list show virtual test sample
And expression residual error is calculated by formula (4):
Wherein, dlFor the expression residual error of class virtual training sample and virtual test sample,For virtual test sample vector
Form,Indicate the vector form of virtual training sample,For the expression of virtual training sample linear expression virtual test sample
Coefficient.
Step 204: by new virtual training sample X '2Carrying out collaboration indicates that algorithm, linear list show new virtual test
SampleAnd expression residual error is calculated by formula (5),
Wherein, dqFor the expression residual error of every a new class of virtual training sample and new virtual test sample,It is virtual
Test sample vector form,Indicate the vector form of virtual training sample,It is new for new virtual training sample linear expression
Virtual test sample expression coefficient.
As further improvement of the invention, step 3 is specifically included:
Step 301: calculating original test sampleThe i class sample x concentrated with original training sampleiHistogram it is similar
Degree, Pasteur's coefficient formula are formula (6)
Wherein, p, q 'iRespectively indicate original test sample Z and i class training sample xiHistogram data；
Step 302: calculating virtual test sampleWith the i class sample x ' in virtual training sample set1iHistogram it is similar
Degree, Pasteur's coefficient formula are formula (7):
Wherein, p1、q′1iRespectively indicate virtual test sampleWith i class virtual training sample x '1iHistogram data.
Step 303: calculating new virtual test sampleWith the i class sample x ' in new virtual training sample set2iIt is straight
Square figure similarity, Pasteur's coefficient formula are formula (8):
As further improvement of the invention, step 4 is specifically included:
Step 401: by every one kind original training sample X and original test samplePasteur's coefficient histogram it is similar
The minimum value and residual values for spending e carry out non-linear fusion in score level, and index non-linear fusion generates new residual error
resikFormula (9) and logarithm non-linear fusion generate new residual error distkFormula (10)
Step 402: by every a kind of virtual training sample X '1With virtual test samplePasteur's coefficient histogram phase
Like degree e1Minimum value and residual values non-linear fusion is carried out in score level, generate new residual error formula (11) and formula
(12),
Step 403: will be per a new class of virtual training sample X '2With new virtual test samplePasteur coefficient
Histogram similarity e2Minimum value and residual values non-linear fusion is carried out in score level, generate new residual error formula
(13) and formula (14),
Step 404: all residual errors newly obtained being subjected to non-linear fusion, generate new residual error formula (15) and formula
(16),
It is improved as of the invention further, in step 5, original test sample, virtual test sample and new virtual survey
The class tag definition of sample sheet is to be expressed as formula (17) and formula (18),
The invention has the benefit that new virtual sample strengthens the ignored detailed information of facial image；By
On the basis of enhancing the pixel of original image moderate strength and reducing the virtual sample of other image pixel intensities, it is former to introduce reinforcement
The more low and higher equicohesive pixel of beginning image, and other pixels are reduced, generate new virtual sample.Then with original
Training sample, virtual training sample and new virtual training sample difference linear expression test sample, virtual test sample and new
Virtual test sample, and Pasteur's coefficient similarity between test sample and training sample is calculated, by Pasteur's coefficient similarity
With Euclidean distance non-linear fusion, test sample is finally determined as to the classification of the smallest training sample of residual values.It joined new
Virtual sample, complementary effect is played to original virtual sample, and strengthen the ignored details letter of face picture
Breath.In non-linear fusion, residual sum similarity information is handled using logarithmic function and exponential function, increase compared with
Gap between small residual values；In sorting phase, Euclidean distance has merged the histogram information of Pasteur's coefficient similarity, non-
The information that linear fusion enables two kinds of distinct methods to generate preferably combines, and is easy to consolidate the weight of similar residual error
The property wanted.Before non-linear fusion, the present invention by residual sum similarity information using logarithmic function and exponential function at
Reason, increases the gap between lesser residual values, to differentiate that the classification of test sample improves accuracy rate.
Detailed description of the invention
Fig. 1 is a kind of face identification method flow chart that non-linear fusion Pasteur coefficient is indicated based on collaboration of the present invention；
Fig. 2 is a kind of recognition of face that non-linear fusion Pasteur coefficient is indicated based on collaboration described in the embodiment of the present invention
The schematic diagram of the sample in the library ORL of method；
Fig. 3 is a kind of recognition of face that non-linear fusion Pasteur coefficient is indicated based on collaboration described in the embodiment of the present invention
The schematic diagram of the sample in the library GT of method；
Fig. 4 is a kind of face identification method that non-linear fusion Pasteur coefficient is indicated based on collaboration described in the embodiment of the present invention
The library FERET sample schematic diagram.
Specific embodiment
The present invention is described in further detail below by specific embodiment and in conjunction with attached drawing.
As shown in Figure 1, a kind of face for indicating non-linear fusion Pasteur coefficient based on collaboration described in the embodiment of the present invention
Recognition methods, method includes the following steps:
Step 1: the intensity by enhancing the pixel of the moderate strength of original image reduces the intensity of other pixels, generates
Virtual sample then by reinforcing the more low and higher equicohesive pixel of original image, and reduces other pixels, generates
New virtual sample, selected part original image is as original training sample, remaining original image is as test sample；
Step 2: algorithm being indicated by collaboration, with virtual training sample, new virtual training sample and original training sample
Linear expression virtual test sample, new virtual test sample and test sample respectively；
Step 3: calculating virtual test sample, new virtual test sample, original test sample and virtual training sample,
Pasteur's coefficient similarity between new virtual training sample, original training sample；
Step 4: the Pasteur's coefficient similarity and Euclidean distance non-linear fusion that step 3 is calculated；
Step 5: being determined according to the calculated result of step 4 to virtual test sample, new virtual test sample and original survey
Sample is originally classified, and is merged, is differentiated according to fused residual error, residual values by class to all obtained residual errors
The smallest differentiation classification is original test sample, virtual test sample and new virtual test sample generic.
Further, in step 1, the image of virtual sample is expressed as Jij=Iij*(m-Iij), wherein IijIndicate original
The intensity of the pixel of image ith row and jth column, JijIndicate the intensity of the pixel of virtual sample image ith row and jth column.New
The image of virtual sample is expressed as:
Wherein, IijIndicate the pixel value of original image ith row and jth column, JijIndicate the i-th row of virtual sample image and the
The pixel value of j column.
Further, in step 2, existing c class face sample, every one kind has n training sample.Enable x1,...,xNIndicate institute
The N number of original training sample (N=nc) having.Assuming that xi∈RP×QIndicate i-th of training sample i ∈ (1,2 ..., N).
Collaboration indicates that algorithm specifically includes:
Step 201: image array is become into column vector,Indicate test sample Z-direction amount form,Indicate original training
Sample xiVector form, and meet formula (1)
Wherein, aiIndicate coefficient；
If withThen formula (1) can be write as formula (2)
Step 202: passing throughA is calculated, wherein λ indicates one
Positive integer (λ=0.01), I are a unit matrix, and the similarity of class training sample and test sample is calculated by formula (3),
Indicate residual error:
Wherein, dkFor the expression residual error of class training sample and test sample,For test sample Z-direction amount form,It indicates
Original training sample xiVector form；
Step 203: by virtual training sample X '1Carrying out collaboration indicates that algorithm, linear list show virtual test sample
And expression residual error is calculated by formula (4):
Wherein, dlFor the expression residual error of class virtual training sample and virtual test sample,For virtual test sample vector
Form,Indicate the vector form of virtual training sample；
Step 204: by new virtual training sample X '2Carrying out collaboration indicates that algorithm, linear list show new virtual test
SampleAnd expression residual error is calculated by formula (5),
Wherein, dqFor the expression residual error of every a new class of virtual training sample and new virtual test sample,It is virtual
Test sample vector form,Indicate the vector form of virtual training sample,It is new for new virtual training sample linear expression
Virtual test sample expression coefficient.
Further, step 3 specifically includes:
Step 301: calculating original test sampleThe i class sample x concentrated with original training sampleiHistogram it is similar
Degree, Pasteur's coefficient formula are formula (6)
Wherein, p, q 'iRespectively indicate original test sample Z and i class training sample xiHistogram data；
Step 302: calculating virtual test sampleWith the i class sample x ' in virtual training sample set1iHistogram it is similar
Degree, Pasteur's coefficient formula are formula (7):
Wherein, p1、q′1iRespectively indicate virtual test sampleWith i class virtual training sample x '1iHistogram data.
Step 303: calculating new virtual test sampleWith the i class sample x ' in new virtual training sample set2iIt is straight
Square figure similarity, Pasteur's coefficient formula are formula (8):
Wherein, p2、q′2iRespectively indicate virtual test sampleWith i class virtual training sample x '2iHistogram data.
Further, it is specifically included in step 4:
Step 401: by the original training sample X of every one kind and original test samplePasteur's coefficient histogram phase
Non-linear fusion is carried out in score level like the minimum value and residual values of degree e, index non-linear fusion generates new residual error
resikFormula (9) and logarithm non-linear fusion generate new residual error distkFormula (10)
Step 402: by every a kind of virtual training sample X '1With virtual test samplePasteur's coefficient histogram phase
Like degree e1Minimum value and residual values non-linear fusion is carried out in score level, generate new residual error formula (11) and formula
(12),
Step 403: will be per a new class of virtual training sample X '2With new virtual test samplePasteur coefficient
Histogram similarity e2Minimum value and residual values non-linear fusion is carried out in score level, generate new residual error formula
(13) and formula (14),
Step 404: all residual errors newly obtained being subjected to non-linear fusion, generate new residual error formula (15) and formula
(16),
Further, in step 5, original test sample, the class label of virtual test sample and new virtual test sample
It is defined to indicate that as formula (17) and formula (18),
The embodiment of the present invention carries out the experiment that small sample quickly identifies, ORL data set using ORL, GT, FERET data set
Comprising the sample from 40 people, 10 pictures of each offer.These pictures are to shoot to obtain under different time, include
Facial expression abundant, Fig. 2 are the example images in ORL data set.The face data library GT (Georgia Tech) includes
The face-image of 50 people.Everyone has 15 color images, and background is complicated.Image, which is shown, has different expressions, lighting condition
With the front of angle.The background for removing each image first, is then converted into gray image, and Fig. 3 is from GT data set
In example images.FERET subdata base includes 700 images from 100 people, everyone provides different posture changings
With 7 images of different illumination, Fig. 4 is the example images in FERET Sub Data Set.
These three data sets contain different time, facial expression, posture and light change abundant.For these feelings
Condition, each data set take everyone preceding 1,2,3,4 facial image as original training sample, remaining facial image respectively
As test sample, test proposes time and the discrimination of algorithm.Comparative experiments of the invention has used classical MSA, LRC,
RBTM, CIRLRC, NFRFR, KRBM, FSSP and DSSR algorithm.Experimental result is as shown in table 1, table 2 and table 3:
Table 1
Table 2
Table 3
By table 1, table 2 and table 3 it is found that mainly for the small scale changes in faces of face, (eye closing of opening eyes comes back in the library ORL
Bow, small scale deflection etc.), in the library GT primarily directed to the expression synthesis of face (smile, frown, large scale side head etc.) and
Lighting angle variation, and the library FERET is mainly illumination and the transformation of posture.This method compares the methods of other same types, right
In the identification of the facial expression transformation and light change of face, there is certain robustness.Having the same with MSA algorithm
On the basis of virtual sample, it joined new virtual sample and play complementary effect to the virtual sample in MSA algorithm, reinforce
It is easy ignored detailed information in face picture, so the discrimination of this paper has certain advantage.Since the present invention is limited
The histogram information for having used Euclidean distance fusion Pasteur's coefficient similarity, by many experiments we have found that the identification of the two
Mistake coincidence factor is low, and the discrimination after merging can improve.And present invention employs non-linear fusion, and it is right
Residual error data is pre-processed, so that the gap between smaller residual error increases, is more conducive to the classification of test sample.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, made any
Modification, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (6)
1. a kind of face identification method for indicating non-linear fusion Pasteur coefficient based on collaboration, which is characterized in that this method includes
Following steps:
Step 1: the intensity by enhancing the pixel of the moderate strength of original image reduces the intensity of other pixels, generates virtual
Sample then by reinforcing the more low and higher equicohesive pixel of original image, and reduces other pixels, generates new void
Quasi- sample, selected part original image is as original training sample, remaining original image is as test sample；
Step 2: algorithm being indicated by collaboration, is distinguished with virtual training sample, new virtual training sample and original training sample
Linear expression virtual test sample, new virtual test sample and original test sample；
Step 3: calculating virtual test sample, new virtual test sample, original test sample and virtual training sample, new void
Pasteur's coefficient similarity between quasi- training sample, original training sample；
Step 4: the Pasteur's coefficient similarity and Euclidean distance non-linear fusion that step 3 is calculated；
Step 5: being determined according to the calculated result of step 4 to virtual test sample, new virtual test sample and original test specimens
This is classified, and merges to all obtained residual errors by class, is differentiated that residual values are the smallest according to fused residual error
Differentiation classification is original test sample, virtual test sample and new virtual test sample generic.
2. a kind of face identification method that non-linear fusion Pasteur coefficient is indicated based on collaboration according to claim 1,
It is characterized in that, in step 1, the image of virtual sample is expressed as Jij=Iij*(m-Iij), wherein IijIndicate the i-th row of original image
With the pixel value of jth column, JijIndicate that the pixel value of virtual sample image ith row and jth column, the image of new virtual sample indicate
Are as follows:
Wherein, IijIndicate the pixel value of original image ith row and jth column, JijIndicate virtual sample image ith row and jth column
Pixel value.
3. a kind of face identification method that non-linear fusion Pasteur coefficient is indicated based on collaboration according to claim 1,
It is characterized in that, step 2 specifically includes:
Step 201: image array is become into column vector,Indicate test sample Z-direction amount form,Indicate original training sample xi
Vector form, and meet formula (1)
Wherein, aiIndicate coefficient；
If withA=[a1,…,aN], then formula (1) can be write as formula (2)
Step 202: passing throughA is calculated, wherein λ indicates a positive integer
(λ=0.01), I are a unit matrix, and the similarity of class training sample and test sample is calculated by formula (3), that is, is indicated
Residual error:
Wherein, dkFor the expression residual error of every a kind of training sample and test sample,For test sample Z-direction amount form,Indicate former
Beginning training sample xiVector form；
Step 203: by virtual training sample X '1Carrying out collaboration indicates that algorithm, linear list show virtual test sampleAnd pass through
Formula (4), which calculates, indicates residual error,
Wherein, dlFor the expression residual error of every a kind of virtual training sample and virtual test sample,For virtual test sample vector shape
Formula,Indicate the vector form of virtual training sample,For the expression system of virtual training sample linear expression virtual test sample
Number；
Step 204: by new virtual training sample X '2Carrying out collaboration indicates that algorithm, linear list show new virtual test sampleAnd expression residual error is calculated by formula (5),
Wherein, dqFor the expression residual error of every a new class of virtual training sample and new virtual test sample,For virtual test
Sample vector form,Indicate the vector form of virtual training sample,For the new void of new virtual training sample linear expression
The expression coefficient of quasi- test sample.
4. a kind of face identification method that non-linear fusion Pasteur coefficient is indicated based on collaboration according to claim 1,
It is characterized in that, step 3 specifically includes:
Step 301: calculating original test sampleThe i class sample x concentrated with original training sampleiHistogram similarity, this bar
Family name's coefficient formula is formula (6)
Wherein, p, q 'iRespectively indicate original test sample Z and i class training sample xiHistogram data；
Step 302: calculating virtual test sampleWith the i class sample x ' in virtual training sample set1iHistogram similarity, should
Pasteur's coefficient formula is formula (7):
Wherein, p1、q′1iRespectively indicate virtual test sampleWith i class virtual training sample x '1iHistogram data；
Step 303: calculating new virtual test sampleWith the i class sample x ' in new virtual training sample set2iHistogram
Similarity, Pasteur's coefficient formula are formula (8):
Wherein, p2、q′2iRespectively indicate virtual test sampleWith i class virtual training sample x '2iHistogram data.
5. a kind of face identification method that non-linear fusion Pasteur coefficient is indicated based on collaboration according to claim 1,
It is characterized in that, step 4 specifically includes:
Step 401: by the original training sample X of every one kind and original test samplePasteur's coefficient histogram similarity e
Minimum value and residual values carry out non-linear fusion in score level, and index non-linear fusion generates new residual error resikFormula
(9) and logarithm non-linear fusion generates new residual error distkFormula (10),
Step 402: by every a kind of virtual training sample X '1With virtual test samplePasteur's coefficient histogram similarity e1
Minimum value and residual values non-linear fusion is carried out in score level, generate new residual error formula (11) and formula (12),
Step 403: will be per a new class of virtual training sample X '2With new virtual test samplePasteur's coefficient histogram
Similarity e2Minimum value and residual values non-linear fusion is carried out in score level, generate new residual error formula (13) and formula
(14),
Step 404: all residual errors newly obtained being subjected to non-linear fusion, generate new residual error formula (15) and formula (16).
6. a kind of face identification method that non-linear fusion Pasteur coefficient is indicated based on collaboration according to claim 1,
It is characterized in that, in step 5, original test sample, the label of the generic of virtual test sample and new virtual test sample
Definition is expressed as formula (17) and formula (18).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910342740.6A CN110096992B (en) | 2019-04-26 | 2019-04-26 | Face recognition method based on collaborative representation nonlinear fusion Bhattacharyya coefficient |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910342740.6A CN110096992B (en) | 2019-04-26 | 2019-04-26 | Face recognition method based on collaborative representation nonlinear fusion Bhattacharyya coefficient |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110096992A true CN110096992A (en) | 2019-08-06 |
| CN110096992B CN110096992B (en) | 2022-12-16 |
Family
ID=67445891
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910342740.6A Active CN110096992B (en) | 2019-04-26 | 2019-04-26 | Face recognition method based on collaborative representation nonlinear fusion Bhattacharyya coefficient |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110096992B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115171190A (en) * | 2022-07-23 | 2022-10-11 | 贵州华数云谷科技有限公司 | Virtual image generation and fusion method for face recognition |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104700076A (en) * | 2015-02-13 | 2015-06-10 | 电子科技大学 | Face image virtual sample generating method |
| US20160034789A1 (en) * | 2014-08-01 | 2016-02-04 | TCL Research America Inc. | System and method for rapid face recognition |
| CN105426871A (en) * | 2015-12-16 | 2016-03-23 | 华南理工大学 | Similarity measure computation method suitable for moving pedestrian re-identification |
| CN105787430A (en) * | 2016-01-12 | 2016-07-20 | 南通航运职业技术学院 | Method for identifying second level human face with weighted collaborative representation and linear representation classification combined |
| CN107273845A (en) * | 2017-06-12 | 2017-10-20 | 大连海事大学 | A kind of facial expression recognizing method based on confidence region and multiple features Weighted Fusion |
| CN108376256A (en) * | 2018-05-08 | 2018-08-07 | 兰州大学 | One kind is based on ARM processing platform dynamic processing face identification systems and its equipment |
-
2019
- 2019-04-26 CN CN201910342740.6A patent/CN110096992B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160034789A1 (en) * | 2014-08-01 | 2016-02-04 | TCL Research America Inc. | System and method for rapid face recognition |
| CN104700076A (en) * | 2015-02-13 | 2015-06-10 | 电子科技大学 | Face image virtual sample generating method |
| CN105426871A (en) * | 2015-12-16 | 2016-03-23 | 华南理工大学 | Similarity measure computation method suitable for moving pedestrian re-identification |
| CN105787430A (en) * | 2016-01-12 | 2016-07-20 | 南通航运职业技术学院 | Method for identifying second level human face with weighted collaborative representation and linear representation classification combined |
| CN107273845A (en) * | 2017-06-12 | 2017-10-20 | 大连海事大学 | A kind of facial expression recognizing method based on confidence region and multiple features Weighted Fusion |
| CN108376256A (en) * | 2018-05-08 | 2018-08-07 | 兰州大学 | One kind is based on ARM processing platform dynamic processing face identification systems and its equipment |
Non-Patent Citations (5)
| Title |
|---|
| CHENGJUN XIE等: "Collaborative object tracking model with local sparse representation", 《JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION》 * |
| CHUNWEI TIAN等: "Integrating Sparse and Collaborative Representation Classifications for Image Classification", 《INTERNATIONAL JOURNAL OF IMAGE AND GRAPHICS》 * |
| YONG XU等: "Using the original and ‘symmetrical face’ training samples to perform representation based two-step face recognition", 《PATTERN RECOGNITION》 * |
| YONGXU等: "Multiple representations and sparse representation for image classification", 《PATTERN RECOGNITION LETTERS》 * |
| 何刚等: "基于镜像脸的FLDA单训练样本人脸识别方法", 《计算机与数字工程》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115171190A (en) * | 2022-07-23 | 2022-10-11 | 贵州华数云谷科技有限公司 | Virtual image generation and fusion method for face recognition |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110096992B (en) | 2022-12-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Research on face recognition based on deep learning | |
| Shao et al. | Joint discriminative learning of deep dynamic textures for 3D mask face anti-spoofing | |
| Li et al. | A review of face recognition technology | |
| Lin et al. | Face liveness detection by rppg features and contextual patch-based cnn | |
| Peng et al. | Face presentation attack detection using guided scale texture | |
| LeCun et al. | Learning methods for generic object recognition with invariance to pose and lighting | |
| CN106203391A (en) | Face identification method based on intelligent glasses | |
| KR20130037734A (en) | A system for real-time recognizing a face using radial basis function neural network algorithms | |
| Mao et al. | Face occlusion recognition with deep learning in security framework for the IoT | |
| CN107220598B (en) | Iris image classification method based on deep learning features and Fisher Vector coding model | |
| CN110263670A (en) | A kind of face Local Features Analysis system | |
| AU2013271337A1 (en) | Biometric verification | |
| Deng et al. | Similarity-preserving image-image domain adaptation for person re-identification | |
| Zeng et al. | A survey of micro-expression recognition methods based on lbp, optical flow and deep learning | |
| Hannan et al. | Analysis of detection and recognition of Human Face using Support Vector Machine | |
| Kumar et al. | One-shot face recognition | |
| CN110096992A (en) | A kind of face identification method indicating non-linear fusion Pasteur coefficient based on collaboration | |
| Li et al. | Foldover features for dynamic object behaviour description in microscopic videos | |
| Vareto et al. | Open-set face recognition with maximal entropy and Objectosphere loss | |
| Bhattacharya et al. | Qdf: A face database with varying quality | |
| CN110287973B (en) | Image feature extraction method based on low-rank robust linear discriminant analysis | |
| Dumitrescu et al. | Combining neural networks and global gabor features in a hybrid face recognition system | |
| Hassanpour et al. | ChatGPT and biometrics: an assessment of face recognition, gender detection, and age estimation capabilities | |
| Su et al. | An enhanced siamese angular softmax network with dual joint-attention for person re-identification | |
| Liang et al. | Exploring regularized feature selection for person specific face verification |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |