CN106951819A - The single sample face recognition method screened based on sparse probability distribution and multistage classification - Google Patents
The single sample face recognition method screened based on sparse probability distribution and multistage classification Download PDFInfo
- Publication number
- CN106951819A CN106951819A CN201610698162.6A CN201610698162A CN106951819A CN 106951819 A CN106951819 A CN 106951819A CN 201610698162 A CN201610698162 A CN 201610698162A CN 106951819 A CN106951819 A CN 106951819A
- Authority
- CN
- China
- Prior art keywords
- pixel
- image
- block
- classification
- sub
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—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 single sample face recognition method screened based on sparse probability distribution and multistage classification, comprise the following steps：First, block is carried out to facial image and sub-block two-stage is divided, assuming that the different sub-blocks in same are to belong to the other different samples of same class, so that solving many sub-space learning methods and sparse representation method can not work in the case of single sample or the problem of hydraulic performance decline；Then each block of facial image is identified, the class probability for obtaining the face using the method for ballot is distributed；Next using a multistage classification screening structure, uncorrelated classification is iteratively weeded out by the thought of entropy minimization, so as to obtain preferable recognition of face classifying quality.The present invention is to expression, illumination variation and blocks etc. and to have good robustness, and accuracy of identification is high, so as to provide a kind of simple and effective solution for list sample recognition of face problem.
Description
Technical field
The invention belongs to field of face identification, it is related to one kind using brand-new effective single sample recognition of face solution,
Specifically related to each object to be identified only has the automatic human face recognition system of a width training image.
Background technology
As a kind of contactless biometric authentication technique, recognition of face is always pattern-recognition and computer vision neck
One of domain most popular research topic, it is widely used in authentication, security monitoring, multimedia recreation, the man-machine friendship of intelligence
The various fields such as mutual.In general, recognition of face is primarily referred to as in digital picture or video image, passes through human body face vision
Information, carries out the computer technology of human identity discriminating.Other biological characteristics such as recognition of face and fingerprint recognition, personal recognition are known
Other technology is compared, and with sampling, more convenient, effect is easy to check, using more extensive etc. outstanding feature.
In numerous existing face identification methods, the extraction of diagnostic characteristics is the key of recognition of face.In the past few decades
In, it is of great interest in numerous feature extraction algorithm sub-spaces learning methods.Its basic thought is by certain
Criterion function find one group of base vector, original high dimensional feature is projected to the lower-dimensional subspace that this group of base vector is opened
In, so that data are compacter respectively to have more distinctive, amount of calculation can also be reduced in addition.In sub-space learning method
It is most representational to surely belong to principal component analysis (M.A.Turk, A.P.Pentland, " Eigenfaces for
recognition,”Journal of cognitive neuroscience,1991,3(1):71-86) and linear discriminant analysis
(P.N.Belhumeur,J.P.Hespanha.Kriegman,D.J.“Eigenfaces vs.Fisherfaces:
Recognition using class specific linear projection,”IEEE Transactions on
Pattern Analysis and Machine Intelligence,1997,19(7):711-720.), it is by widely
Applied in recognition of face and achieving good recognition result.In recent years J.Wright et al. propose based on rarefaction representation
Face identification method (A.Yang, A.Ganesh, S.Sastry, and Y.Ma, " Robust Face Recognition via
Sparse Representation,”IEEE Trans.Pattern Analysis and Machine Intelligence,
Vol.31, no.2, pp.210-227,2009) with the antinoise of its robust, block ability and received much concern, it is substantially former
Reason is to construct dictionary using all training images, then tries to achieve the most sparse line of test image by solving a underdetermined system of equations
Property combination coefficient, then image is identified classification according to these coefficients.Because this method will solve l1- norm minimums
Problem, computation complexity is higher, and Zhang Lei et al. is had found by studying：Be collaboration of the training image to test image represent without
It is the openness accuracy for being more conducive to lift recognition of face of l1- norms induction, they thus propose is represented based on collaboration
Face identification method (L.Zhang, M.Yang, and X.Feng, " Sparse representation or
collaborative representation:which helps face recognition" in ICCV 2011), it is proposed that
Replace l1- norms as regular terms with l2- norms, calculating is substantially reduced while competitive recognition effect can be obtained strong
Degree.
The above-mentioned face identification method represented based on sparse or collaboration or traditional face based on sub-space learning are known
Other method all depends critically upon the number of training sample, and wherein most method is recognized when facing single sample recognition of face problem
Hydraulic performance decline it is very serious, or even some methods can not work at all in the case of single sample, only the training sample in each class
It can just be shown when this quantity is more preferably for noise, the robustness blocked.Therefore for such as identity card identification, customs
In the case of one training sample being there is normally only in many practical applications such as passport verification, security monitoring, the identification of these methods
Performance will drastically decline even completely infeasible.
The content of the invention
It is an object of the invention to provide a kind of single sample face screened based on sparse probability distribution and multistage classification
Recognition methods.It is not enough that the present invention tackles training samples number by image block and using the high similitude of sub-block in image block
The challenge brought, is that traditional sub-space learning method and sparse or collaboration represent that being applied to single sample recognition of face problem carries
A practicable solution is supplied.
The technical solution for realizing the object of the invention is：It is a kind of to be screened based on sparse probability distribution and multistage classification
Single sample face recognition method, passes through the high similitude of image block and utilization image block neutron block structure, it is proposed that should first
The a little piece of hypothesis as same category of different samples.Based on this hypothesis, it is possible to the localized mass application tradition of image
Face identification method classified：Test image carries out the division of block and sub-block, and each piece is individually known using sub-block
Not, the other probability distribution of row of the image is tried to achieve.Then, according to the principle of entropy minimization, sieved by multistage classification and classification
Choosing, constantly rejects the classification of small probability, and then improves the probability of correct classification, then terminates iteration until meeting stop condition, most
Classification results of output image afterwards.The present invention specifically includes following steps：
1. pair all one training samples and the image of test sample are first carried out in piecemeal, block with the square block of fixed size
Each pixel in heart correspondence image.Then continue to mark off different sub-blocks in an overlapping manner each piece of inside,
Propose that the different sub-blocks in each block of facial image can regard the reasonable vacation for belonging to the other different samples of same class as
If；
2. based on above-mentioned it is assumed that using the different sub-blocks in same as sample, using traditional learning method, to figure
Each piece of picture is individually classified, and is distributed according to the class probability that the image is obtained in ballot；
3. the probability distribution for belonging to every class obtained using 2, classification screening is carried out according to entropy minimization principle, weed out general
Rate is 0 or relatively low classification, completes primary screening process；
4. the category classification above-mentioned to the progress of test image iteration and screening process, until meeting stop condition, so that defeated
Go out the final classification results of the test image；
Compared with prior art, its remarkable advantage is the present invention：(1) traditional face identification method is usually constructed with training sample
This number has certain requirement.The sub-block that the present invention has been formed by image block is used as same category of different samples and made up
This defect so that many conventional methods can be applied to single specimen discerning occasion (2) propose one it is multistage
Classification screens the problem of framework is to solve entropy minimization, and amalgamation of global provides more accurately classification foundation with local advantage, because
And with preferable robustness and accuracy of identification.(4) parallel computation is supported, computational efficiency is high：Image block class is differentiated in the present invention
Other work is committed step, and each image block can greatly save the calculating time with parallel processing completely, can more be met in real time
Property require higher application scenario.
Brief description of the drawings
The algorithm for single sample face recognition method that Fig. 1 is screened for the present invention based on sparse probability distribution and multistage classification
Flow chart.
The neighborhood for single sample face recognition method that Fig. 2 is screened for the present invention based on sparse probability distribution and multistage classification
Pixel schematic diagram.
The partition for single sample recognition of face that Fig. 3 is screened for the present invention based on sparse probability distribution and multistage classification
Schematic diagram.
The block sub-block two for single sample recognition of face that Fig. 4 is screened for the present invention based on sparse probability distribution and multistage classification
Level divides schematic diagram.
The acquisition for single sample recognition of face that Fig. 5 is screened for the present invention based on sparse probability distribution and multistage classification is minimum
The schematic diagram of entropy.
Embodiment
The present invention is described in further detail below in conjunction with the accompanying drawings.
With reference to Fig. 1, single sample face recognition method that the present invention is screened based on sparse probability distribution and multistage classification, bag
Include following steps：
1.1 in view of the Limited information that single image is provided, is from making full use of limited information to be classified, image is entered
Row piecemeal, analyzes each block respectively, finally merges each piece of classification results.The present invention proposes a kind of method of partition, specifically
It is as follows：
Assuming that a facial image hasIndividual pixel, we centered on each pixel, take R to be built as radius respectively
One square area is as the neighborhood territory pixel of the pixel, and each pixel that neighborhood is concentrated is corresponding centered on it, greatly
It is small be S × S fritter (we make S for odd number not less than 3), we are by the S in fritter2Table is linked in sequence in individual pixel
It is shown as vector formSimilarly, the fritter of a S × S corresponding to center pixel i, being expressed as vector form isDivided by such, we have just obtained a localized mass of an image, bag has expanded its center pixel and its neighborhood picture
All sub-blocks corresponding to element, size is (S+2R) × (S+2R).As shown in Figure 3.In above-mentioned partition process, marginal portion
The problem of partial pixel value missing is so as to lead to not divide occurs.We take the method for edge pixel mirror image to solve this
It is, using edge as symmetry axis, the pixel on edge to be copied over by symmetrical form, is used as in individual problem, specific implementation
Missing pixel values are filled up, so as to complete the correct division of marginal portion image block.
First carry out piecemeal with the square block of fixed size to all training samples and the image of test sample, block center according to
Some neighboring sub-patch groups on secondary each pixel taken in image, each block center sub-block as shown in Figure 3 and periphery
Into.Fig. 4 divides schematic diagram for the two-stage of image, due to the same piece of smaller area for being confined to facial image, its internal institute
There is sub-block again overlapped, therefore these sub-blocks have stronger similitude, therefore may be assumed that these sub-blocks are to belong to same category
Different samples.Thus it can be assumed that whole sub-blocks at the pixel i of the training facial image of all classes in corresponding piece are constituted
Block training set Bi, test the center sub-block at pixel i in corresponding piece in facial imageThen as test sample.
1.2 based on above-mentioned it is assumed that being illustrated exemplified by cooperateing with the face identification method that represents, for block training setCollaboration representForm can be written as：
Concrete operations are as follows：
1.2.1 first, normalizing is carried out with 2- norms to the corresponding column vector of each facial image of single sample facial image
Change is handled.
1.2.2 then, when carrying out rarefaction representation, each square block marked off is handled as follows：
Solve l1Minimization problem：Wherein subscript i is represented currently
I-th piece of processing (i=1,2 ... N, N are image total pixel number), matrix BiRow by all sons in all i-th piece of training samples
Block number removes the center sub-block y in rarefaction representation test image in i-th piece according to composition in this, as local dictionary0 i, pass through above-mentioned l1
Minimize the coefficient for solving and obtaining rarefaction representation.
1.2.3 secondly, test image is reconstructed using each class training image blocks and both residual errors are sought, formula is：Wherein,The rarefaction representation coefficient tried to achieve for 1.2.2, functionFor taking out expression
Correspond to those components of kth class in coefficient.
1.2.4 it is last, by test image block sort to that minimum class of reconstructed residual, complete once to be based on sparse collaboration
The face recognition process of expression.
1.3 illustrate by taking LDA methods as an example, for block training setIn its class Scatter Matrix and
Class scatter matrix can be calculated as follows：
1.3.1 next we are kept by maximizing following target while classification ga s safety degree, are reduced in class
Change, caused by illumination etc.：
1.3.2 we are by a matrix U, and its row areCharacteristic vector (also as Fisherfaces),
To carry out linear transformation to face, so as to complete the identification of face.
Every piece of classification information of 1.4 images obtained using 1.2 or 1.3, obtains the class probability distribution of the image, so
The screening process of classification is carried out afterwards, is comprised the following steps that：
(1) after classifying according to the method described above to each image block, the classification for calculating image according to equation below is general
Rate is distributed β=[p1,p2,...,pK]：
Wherein, VkIt is the votes that kth class is obtained, ∑ VkFor total votes.
(2) after the class probability distribution for obtaining image, the category set Ω of image is generated1={ C1,C2,...CN1},
N1Then=K. removes the classification of wherein a series of 0 probability or small probability, and classification screening operation generates a new classification
Set omega2={ C1,C2,...,CN2, the set of this new generation is used for the classification in next stage.
The above-mentioned category classification of 1.5 pairs of test image iteration progress and screening process, until meeting stop condition, so that defeated
Go out the final classification results of the test image, as shown in Figure 5.
Claims (7)
1. a kind of single sample face recognition method screened based on sparse probability distribution and multistage classification, it is characterised in that：
Step 1. carries out piecemeal, block with the square block of fixed size to all one training samples and the image of test sample first
Each pixel in the correspondence image of center；Then continue to mark off different sons in an overlapping manner each piece of inside
Block, proposes the different sub-blocks in each block of facial image, regards the reasonable vacation for belonging to the other different samples of same class as
If；
Step 2. is based on above-mentioned it is assumed that using the different sub-blocks in same as sample, using traditional learning method, to figure
Each piece of picture is individually classified, and is distributed according to the class probability that voting results obtain the test image；
The probability distribution for belonging to every class that step 3. is obtained using step 2, classification screening is carried out according to entropy minimization principle, is rejected
Fall probability for 0 or relatively low classification, complete primary screening process；
Step 4. carries out the category classification and screening process described in step 2 and step 3 to test image iteration, stops until meeting
Condition, so as to export the final classification results of the test image.
2. the single sample recognition of face side according to claim 1 screened based on sparse probability distribution and multistage classification
Method, it is characterised in that block and sub-block two-stage are divided in the step 1 operates as follows：
Assuming that image has N number of pixel, centered on each pixel, radius be neighborhood that the pixel on R square is the pixel
Pixel, then each pixel one S × S fritter centered on it of correspondence that pixel i neighborhood territory pixel is concentrated, S is taken more than or equal to 3
Odd number, the S in fritter2Individual pixel is expressed as vector formJ=1 ..., P；Similarly, center pixel i also corresponds to one
Individual S × S fritter, being expressed as vector form isCenter pixel i and its corresponding all fritters formation of neighborhood territory pixel one with
Localized mass centered on pixel i, size is (S+2R) × (S+2R)；It is corresponding at the pixel i of the training facial image of all classes
Whole sub-blocks in block constitute block training set Bi, test the center sub-block at pixel i in corresponding piece in facial imageThen make
For test sample.
3. the single sample recognition of face side according to claim 2 screened based on sparse probability distribution and multistage classification
Method, it is characterised in that：In the partition process, marginal portion is solved using the method for edge pixel mirror image and partial pixel value occurs
The problem of missing is so as to lead to not divide；Specific implementation method is using edge as symmetry axis, by the pixel on edge by symmetrical
Form be copied over, be used as filling up for missing pixel values, so as to complete the correct division of marginal portion image block.
4. the single sample recognition of face side according to claim 1 screened based on sparse probability distribution and multistage classification
Method, it is characterised in that：In the step 1, same piece be confined to facial image smaller area, its internal all sub-block phase
It is mutually overlapping, therefore these sub-blocks have stronger similitude, therefore assume that these sub-blocks are to belong to same category of different samples.
5. the single sample recognition of face side according to claim 1 screened based on sparse probability distribution and multistage classification
Method, it is characterised in that the class probability distribution that image is asked in the step 2 is operated according to the following steps：
2.1 for block training setScatter Matrix and class scatter matrix are calculated as follows in its class：
2.2 are classified each image block using sparse representation methods, based on block training set BiTo testCollaboration represent to be write as
Lower form：
After 2.3 classify to each image block according to the method described above, the class probability point of image is calculated according to equation below
Cloth β=[p1,p2,...,pK]：
Wherein, VkIt is the votes that kth class is obtained, ∑ VkFor total votes.
6. the single sample recognition of face side according to claim 1 screened based on sparse probability distribution and multistage classification
Method, it is characterised in that the classification screening process in the step 3 is as follows：
3.1 class probability for having tried to achieve is distributed, and its entropy is calculated according to equation below：
E (β)=- ∑ pkln pk
After 3.2 class probabilities for obtaining image are distributed, the category set Ω of image is generated1={ C1,C2,...CN1},N1=
K, then removes the classification of wherein a series of 0 probability or small probability, and classification screening operation produces a new category set Ω2
={ C1,C2,...,CN2, the set of this new generation is used for the classification in next stage.
7. the single sample recognition of face side according to claim 1 screened based on sparse probability distribution and multistage classification
Method, it is characterised in that：In the step 4：
The new category set tried to achieve in step 3.2 is carried out to the classification of next stage, classification is then iteratively performed and screens
The operation of this two step exports the classification results of final image until meeting stop condition.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610698162.6A CN106951819A (en) | 2016-08-19 | 2016-08-19 | The single sample face recognition method screened based on sparse probability distribution and multistage classification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610698162.6A CN106951819A (en) | 2016-08-19 | 2016-08-19 | The single sample face recognition method screened based on sparse probability distribution and multistage classification |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106951819A true CN106951819A (en) | 2017-07-14 |
Family
ID=59465519
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610698162.6A Pending CN106951819A (en) | 2016-08-19 | 2016-08-19 | The single sample face recognition method screened based on sparse probability distribution and multistage classification |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106951819A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107886090A (en) * | 2017-12-15 | 2018-04-06 | 苏州大学 | A kind of single sample face recognition method, system, equipment and readable storage medium storing program for executing |
| CN109214307A (en) * | 2018-08-15 | 2019-01-15 | 长安大学 | Low resolution illumination robust human face recognition methods based on probability rarefaction representation |
| CN109840567A (en) * | 2018-11-16 | 2019-06-04 | 中电科新型智慧城市研究院有限公司 | A kind of steady differentiation feature extracting method indicated based on optimal collaboration |
| CN109902657A (en) * | 2019-03-12 | 2019-06-18 | 哈尔滨理工大学 | A kind of face identification method indicated based on piecemeal collaboration |
| CN111738244A (en) * | 2020-08-26 | 2020-10-02 | 腾讯科技（深圳）有限公司 | Image detection method, image detection device, computer equipment and storage medium |
| CN112565614A (en) * | 2021-02-22 | 2021-03-26 | 四川赛狄信息技术股份公司 | Signal processing module and method |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104732186A (en) * | 2013-12-18 | 2015-06-24 | 南京理工大学 | Single sample face recognition method based on local subspace sparse representation |
-
2016
- 2016-08-19 CN CN201610698162.6A patent/CN106951819A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104732186A (en) * | 2013-12-18 | 2015-06-24 | 南京理工大学 | Single sample face recognition method based on local subspace sparse representation |
Non-Patent Citations (3)
| Title |
|---|
| FAN LIU: "《Local Structure based Sparse Representation for Face Recognition with Single Sample per Person》", 《2014 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP)》 * |
| 刘凡: "《人脸识别中基于图像局部结构的特征提取与分类研究》", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 朱杰等: "《基于字典学习的核稀疏表示人脸识别方法》", 《模式识别与人工智能》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107886090A (en) * | 2017-12-15 | 2018-04-06 | 苏州大学 | A kind of single sample face recognition method, system, equipment and readable storage medium storing program for executing |
| CN107886090B (en) * | 2017-12-15 | 2021-07-30 | 苏州大学 | Single-sample face recognition method, system, equipment and readable storage medium |
| CN109214307A (en) * | 2018-08-15 | 2019-01-15 | 长安大学 | Low resolution illumination robust human face recognition methods based on probability rarefaction representation |
| CN109840567A (en) * | 2018-11-16 | 2019-06-04 | 中电科新型智慧城市研究院有限公司 | A kind of steady differentiation feature extracting method indicated based on optimal collaboration |
| CN109840567B (en) * | 2018-11-16 | 2021-12-17 | 中电科新型智慧城市研究院有限公司 | Robust discriminant feature extraction method based on optimal collaborative representation |
| CN109902657A (en) * | 2019-03-12 | 2019-06-18 | 哈尔滨理工大学 | A kind of face identification method indicated based on piecemeal collaboration |
| CN109902657B (en) * | 2019-03-12 | 2022-07-08 | 哈尔滨理工大学 | Face recognition method based on block collaborative representation |
| CN111738244A (en) * | 2020-08-26 | 2020-10-02 | 腾讯科技（深圳）有限公司 | Image detection method, image detection device, computer equipment and storage medium |
| CN112565614A (en) * | 2021-02-22 | 2021-03-26 | 四川赛狄信息技术股份公司 | Signal processing module and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107609497B (en) | Real-time video face recognition method and system based on visual tracking technology | |
| CN106951819A (en) | The single sample face recognition method screened based on sparse probability distribution and multistage classification | |
| CN103605972B (en) | Non-restricted environment face verification method based on block depth neural network | |
| CN106599797B (en) | A kind of infrared face recognition method based on local parallel neural network | |
| Wang et al. | SymNet: A simple symmetric positive definite manifold deep learning method for image set classification | |
| CN109063666A (en) | The lightweight face identification method and system of convolution are separated based on depth | |
| Ming et al. | Simple triplet loss based on intra/inter-class metric learning for face verification | |
| CN102938065B (en) | Face feature extraction method and face identification method based on large-scale image data | |
| CN106372581A (en) | Method for constructing and training human face identification feature extraction network | |
| CN103679158A (en) | Face authentication method and device | |
| CN111414862A (en) | Expression recognition method based on neural network fusion key point angle change | |
| CN113076927B (en) | Finger vein identification method and system based on multi-source domain migration | |
| CN102521575A (en) | Iris identification method based on multidirectional Gabor and Adaboost | |
| CN104732186B (en) | Single sample face recognition method based on Local Subspace rarefaction representation | |
| CN103077378B (en) | Contactless face recognition algorithms based on extension eight neighborhood Local textural feature and system of registering | |
| CN108875907A (en) | A kind of fingerprint identification method and device based on deep learning | |
| CN110472495B (en) | Deep learning face recognition method based on graphic reasoning global features | |
| CN106295609A (en) | The single sample face recognition method represented based on block sparsity structure low-rank | |
| CN109344856B (en) | Offline signature identification method based on multilayer discriminant feature learning | |
| CN107545243A (en) | Yellow race's face identification method based on depth convolution model | |
| CN104298974A (en) | Human body behavior recognition method based on depth video sequence | |
| CN107092884A (en) | Rapid coarse-fine cascade pedestrian detection method | |
| CN110414587A (en) | Depth convolutional neural networks training method and system based on progressive learning | |
| CN105160290A (en) | Mobile boundary sampling behavior identification method based on improved dense locus | |
| CN107122725A (en) | A kind of face identification method and its system based on joint sparse discriminant analysis |
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 |
Application publication date: 20170714 |
|
| RJ01 | Rejection of invention patent application after publication |