CN105550642A - Gender identification method and system based on multi-scale linear difference characteristic low-rank expression - Google Patents
Gender identification method and system based on multi-scale linear difference characteristic low-rank expression Download PDFInfo
- Publication number
- CN105550642A CN105550642A CN201510895459.7A CN201510895459A CN105550642A CN 105550642 A CN105550642 A CN 105550642A CN 201510895459 A CN201510895459 A CN 201510895459A CN 105550642 A CN105550642 A CN 105550642A
- Authority
- CN
- China
- Prior art keywords
- human face
- rank
- face region
- low
- linear differential
- 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
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000012549 training Methods 0.000 claims abstract description 19
- 239000011159 matrix material Substances 0.000 claims description 26
- 238000012360 testing method Methods 0.000 claims description 21
- 238000007477 logistic regression Methods 0.000 claims description 17
- 230000001815 facial effect Effects 0.000 claims description 7
- 238000001914 filtration Methods 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 5
- 238000011478 gradient descent method Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 2
- 230000003993 interaction Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000005286 illumination Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 241000228740 Procrustes Species 0.000 description 1
- 241000405217 Viola <butterfly> Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
- G06V40/171—Local features and components; Facial parts ; Occluding parts, e.g. glasses; Geometrical relationships
-
- 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
- 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
Abstract
The invention discloses a gender identification method and system based on multi-scale linear difference characteristic low-rank expression. The method comprises the steps of: carrying out human face detection based on Haar-like characteristics and an Adaboost classifier algorithm, and cutting out a human face region in an image to be detected; extracting texture characteristics based on multi-scale linear difference characteristics in the cut-out human face region; and then under a low-rank constraint condition, projecting the texture characteristics to a sub-space, training a logic regression classifier model in the sub-space, and utilizing the model to carry out gender identification of the human face image.
Description
Technical field
The present invention relates to Techniques of Gender Recognition field, particularly relate to the gender identification method and system that represent based on multiple dimensioned linear Differential Characteristics low-rank.
Background technology
Face picture contains abundant personal information, and comprise identity, age, sex and race etc., these information are widely used in field of human-computer interaction.Along with ecommerce development and and various mobile device universal, the gender information of user in man-machine interaction in occupation of more and more important effect.Sex based on face picture is identified in man-machine interaction and has a wide range of applications, the user management in such as security monitoring, ecommerce, the access control of website, image and video frequency searching and more humane human-computer interaction function etc.
Rely on this biological characteristic of face to carry out sex identification, do not need user to cooperate with on one's own initiative, thus operation is disguised strong, can provide better Consumer's Experience.Meanwhile, because face picture is contactless collection, because the not property invaded, more meet the identification custom of the mankind, easily accepted by users.
But its recognition efficiency of existing recognition methods and accuracy rate need to improve.
Therefore, prior art has yet to be improved and developed.
Summary of the invention
In view of above-mentioned the deficiencies in the prior art, the object of the present invention is to provide the gender identification method and system that represent based on multiple dimensioned linear Differential Characteristics low-rank, be intended to solve its recognition efficiency of existing recognition methods and accuracy rate has the problem needing to improve.
Technical scheme of the present invention is as follows:
Based on the gender identification method that multiple dimensioned linear Differential Characteristics low-rank represents, wherein, comprise step:
A, use Face datection algorithm based on Haar-like characteristic sum Adaboost sorter, shear out the human face region in testing image;
B, on the human face region sheared out, extract textural characteristics based on multiple dimensioned linear Differential Characteristics;
C, under low-rank constraint condition, by textural characteristics form eigenmatrix project to proper vector subspace;
D, in proper vector subspace, training logistic regression sorter model, utilizes described logistic regression sorter model to carry out sex identification to human face region.
The described gender identification method represented based on multiple dimensioned linear Differential Characteristics low-rank, wherein, described steps A specifically comprises:
A1, use integrogram calculate the rectangular characteristic of testing image as Weak Classifier;
A2, use Adaboost algorithm pick out the Weak Classifier representing face, according to the mode of Nearest Neighbor with Weighted Voting, Weak Classifier are configured to strong classifier;
A3, the cascade filtering of finally will the some strong classifiers obtained be trained to be composed in series a cascade structure.
The described gender identification method represented based on multiple dimensioned linear Differential Characteristics low-rank, wherein, described step B specifically comprises:
B1, active shape model is utilized to carry out facial modeling;
B2, the human face characteristic point position obtained according to active shape model, rotate the human face region of testing image and ajust, carry out convergent-divergent, row interpolation process of going forward side by side to this human face region;
B3, centered by groups of people's face characteristic point position, in human face region, mark off multiple image block;
B4, on each image block, extract textural characteristics.
The described gender identification method represented based on multiple dimensioned linear Differential Characteristics low-rank, wherein, in described step C, projects by following formula:
Wherein, W is the subspace projection matrix obtained that projects,
for between class scatter matrix,
for Scatter Matrix in class,
for eigenmatrix projects to the nuclear norm behind proper vector subspace, for limiting the size of the rear subspace projection rank of matrix of projection.
The described gender identification method represented based on multiple dimensioned linear Differential Characteristics low-rank, wherein, in described step D, by subspace projection matrix W the value on the different dimensions of textural characteristics is weighted and obtains weighted value, and weighted value is compressed to the scope of 0 ~ 1 with logical function.
The described gender identification method represented based on multiple dimensioned linear Differential Characteristics low-rank, wherein, in described step D, the sex of training sample for woman's probability is
,
for the low-rank of textural characteristics represents.
The described gender identification method represented based on multiple dimensioned linear Differential Characteristics low-rank, wherein, in described step D, utilizes gradient descent method, and the loss function minimizing subspace projection matrix W obtains the optimum solution of subspace projection matrix W.
Based on the sex recognition system that multiple dimensioned linear Differential Characteristics low-rank represents, wherein, comprising:
Shear module, for using the Face datection algorithm based on Haar-like characteristic sum Adaboost sorter, shears out the human face region in testing image;
Characteristic extracting module, for extracting the textural characteristics based on multiple dimensioned linear Differential Characteristics on the human face region sheared out;
Low-rank representation module, under low-rank constraint condition, projects to proper vector subspace by the eigenmatrix that textural characteristics is formed;
Identification module, in proper vector subspace, training logistic regression sorter model, utilizes described logistic regression sorter model to carry out sex identification to human face region.
The described sex recognition system represented based on multiple dimensioned linear Differential Characteristics low-rank, wherein, described shear module specifically comprises:
Computing unit, calculates the rectangular characteristic of testing image as Weak Classifier for using integrogram;
Ballot unit, picking out the Weak Classifier representing face, according to the mode of Nearest Neighbor with Weighted Voting, Weak Classifier being configured to strong classifier for using Adaboost algorithm;
Series unit, for last cascade filtering of will the some strong classifiers obtained be trained to be composed in series a cascade structure.
The described sex recognition system represented based on multiple dimensioned linear Differential Characteristics low-rank, wherein, described characteristic extracting module specifically comprises:
Positioning unit, carries out facial modeling for utilizing active shape model;
Unit for scaling, for the human face characteristic point position obtained according to active shape model, rotates the human face region of testing image and ajusts, carry out convergent-divergent, row interpolation process of going forward side by side to this human face region;
Image block retrieval unit, for centered by groups of people's face characteristic point position, marks off multiple image block in human face region;
Feature extraction unit, for extracting textural characteristics on each image block.
Beneficial effect: the Face datection that the present invention is based on Haar-like characteristic sum Adaboost classifier algorithm, shears out the human face region in testing image; And the textural characteristics extracted on the human face region sheared out based on multiple dimensioned linear Differential Characteristics; Then under low-rank constraint condition, textural characteristics is projected to subspace, in this subspace, training logistic regression sorter model, utilizes model to carry out the sex identification of face picture.Its recognition accuracy of recognition methods of the present invention is high, and effectively can improve recognition efficiency.
Accompanying drawing explanation
Fig. 1 is the process flow diagram that the present invention is based on the gender identification method preferred embodiment that multiple dimensioned linear Differential Characteristics low-rank represents.
Fig. 2 is the process flow diagram of the Face datection process in the present invention.
Embodiment
The invention provides the gender identification method and system that represent based on multiple dimensioned linear Differential Characteristics low-rank, for making object of the present invention, technical scheme and effect clearly, clearly, the present invention is described in more detail below.Should be appreciated that specific embodiment described herein only in order to explain the present invention, be not intended to limit the present invention.
Refer to Fig. 1, Fig. 1 is the process flow diagram of present pre-ferred embodiments, and as shown in the figure, it comprises step:
S101, use Face datection algorithm based on Haar-like characteristic sum Adaboost sorter, shear out the human face region in testing image;
S102, on the human face region sheared out, extract textural characteristics based on multiple dimensioned linear Differential Characteristics;
S103, under low-rank constraint condition, by textural characteristics form eigenmatrix project to proper vector subspace;
S104, in proper vector subspace, training logistic regression sorter model, utilizes described logistic regression sorter model to carry out sex identification to human face region.
Wherein, step S101 mainly completes Face datection.As shown in Figure 2, the present invention adopts Viola method for detecting human face, and it is a kind of method based on integrogram, cascade detectors and AdaBoost algorithm.
Specifically, it comprises step:
S201, use integrogram calculate the rectangular characteristic of testing image as Weak Classifier;
S202, use Adaboost algorithm pick out the Weak Classifier representing face, according to the mode of Nearest Neighbor with Weighted Voting, Weak Classifier are configured to strong classifier;
S203, the cascade filtering of finally will the some strong classifiers obtained be trained to be composed in series a cascade structure.
First, use Harr-like character representation face, specifically use integrogram to calculate the rectangular characteristic of testing image; Then, use Adaboost algorithm to pick out the rectangular characteristic (Weak Classifier) that some can represent face, become multi-scale image to form subimage to be measured by testing image by the continuous scaling of ratio; According to the mode of Nearest Neighbor with Weighted Voting, Weak Classifier is configured to strong classifier again; Finally, by the cascade filtering of training the some strong classifiers obtained to be composed in series a cascade structure, cascade structure can improve the detection speed of sorter effectively.
Step S102 mainly completes the extraction of multiple dimensioned linear Differential Characteristics (LDF).Linear differential feature (LDF) utilizes the second order difference information of image to carry out reasonable coding.Specifically, it comprises step:
S301, active shape model is utilized to carry out facial modeling;
S302, the human face characteristic point position obtained according to active shape model, rotate the human face region of testing image and ajust, carry out convergent-divergent, row interpolation process of going forward side by side to this human face region;
S303, centered by groups of people's face characteristic point position, in human face region, mark off multiple image block;
S304, on each image block, extract textural characteristics.
Specifically, the process of multiple dimensioned LDF extraction is as follows: first utilize active shape model (ActiveShapeModel, ASM) to carry out facial modeling, as adopted 68 point models.The structure of active shape model is divided into training and search two steps.
During training: the position constraint setting up human face characteristic point, build the local feature obtaining human face characteristic point;
During search: the coupling comprising iteration.The first step is training: first collect N number of training sample and build active shape model; Hand labeled human face characteristic point; The coordinate of human face characteristic point in training sample is conspired to create proper vector; Proper vector is normalized and align (alignment adopts Procrustes method); Proper vector after alignment does PCA process.Then, for everyone face characteristic point builds local feature, object is that everyone can find new position by face characteristic point in each iterative search procedures.Local feature generally uses Gradient Features, in case illumination variation.Second step is search: the first position of eyes in calculation training sample (or eyes and face), does yardstick and rotates change, alignment face; Then, near each human face characteristic point after alignment, search, mates each local feature (often adopting mahalanobis distance), obtains preliminary configuration; Average face (active shape model) is used to revise matching result again; Iteration is until convergence.
Obtain the position of human face characteristic point according to active shape model, based on 2 eye coordinates points, human face region rotation is ajusted, and according to two eye distances from carrying out scaling to three yardsticks, adopt bilinearity cubic interpolation.Then, centered by groups of people's face characteristic point position, in human face region, divide multiple subregion form image block and take out, for different size human face region, the size of taking out image block is set to necessarily, the present invention selects 16 human face characteristic points as the center of the image block that will obtain, these human face characteristic points mainly face inside some points and do not choose some points at face edge.On each image block, extract the textural characteristics of linear differential feature.Linear differential feature has more sign ability, and can strengthen picture quality to a certain extent.First, computing center's pixel and the radius in " level ", " vertically ", " positive diagonal line ", " counter-diagonal " these four position line are the magnitude relationship of the neighbor of 2, calculate and not simply mutually subtract each other, but subtract each other again after subtracting each other successively, what namely extract is second order difference information in this position line; After obtaining difference, then carry out two-value according to the result that compares with 0 and turn to 0 or 1; Finally, using encoded translated for decimal value is as eigenwert for obtain 01, four position line due to this algorithm have included the direction of 8 neighborhoods, therefore coding value is out between 0 ~ 15 instead of between 0 ~ 255 of picture LBP, when doing statistics with histogram, histogrammic dimension also can be lower, even if compared with tieing up histogram 0 with 59 of the LBP adopting " uniform pattern ", dimension is also lower.
The low-rank that step S103 mainly completes textural characteristics represents.
Specifically, consider multiple face picture of same person, if the textural characteristics each width face picture being extracted to linear differential feature obtains a row vector, using the textural characteristics of multiple face picture of multiple people as row vector constitutive characteristic matrix, then it this matrix theory should be low-rank matrix.But due in reality, every width face picture can be subject to the impact of Different factor, such as illumination, block, the noise such as translation.The impact of the eigenmatrix that these noise on human face pictures are formed can be regarded as the effect of a noise matrix.So the present invention, when carrying out subspace projection, adds low-rank constraint condition, the impact of the noise such as can effectively reduce illumination, block.Subspace projection algorithm of the present invention, based on linear discriminant analysis (LinearDiscriminantAnalysis, LDA), is with the difference of conventional linear discriminatory analysis model: on traditional model basis, add low-rank constraint condition.Optimizing expression is as follows:
Wherein, W is the subspace projection matrix obtained that projects,
for between class scatter matrix,
for Scatter Matrix in class,
for eigenmatrix projects to the nuclear norm behind proper vector subspace (this formula bottom right is " * " number), for limiting the size of the rear subspace projection rank of matrix of projection.By the optimization to above-mentioned goal expression, subspace projection matrix can be obtained.
The main completion logic of step S104 returns sorter model.
Specifically, a kind of method based on discriminant of logistic regression sorter model (LogicRegression), the example of its supposition class is linear separability, by the parameter of direct estimation discriminant, obtains final forecast model.Mainly consider based on binary classification logistic regression forecast model in the present invention, i.e. the class label of sorter identification is 0(man) and 1(female).In logistic regression sorter model, by subspace projection matrix W the value on the different dimensions of textural characteristics is weighted and obtains weighted value, and with logic (Sigmoid) function, weighted value is compressed to the scope of 0 ~ 1, be the probability of positive sample as this sample using functional value.Logic (Sigmoid) function is
In Logic Regression Models, the sex of training sample for woman's probability is
,
for the low-rank of textural characteristics represents.
Solving for Logic Regression Models, finds suitable subspace projection matrix W exactly, makes the maximized classification of men and women's picture in training sample correct.The loss function of this problem is
Utilize gradient descent method, the loss function minimizing subspace projection matrix W obtains the optimum solution of feature weight vector W, can obtain the logistic regression sorter model for sex identification.
Based on said method, the present invention also provides the sex recognition system represented based on multiple dimensioned linear Differential Characteristics low-rank, and it comprises:
Shear module, for using the Face datection algorithm based on Haar-like characteristic sum Adaboost sorter, shears out the human face region in testing image;
Characteristic extracting module, for extracting the textural characteristics based on multiple dimensioned linear Differential Characteristics on the human face region sheared out;
Low-rank representation module, under low-rank constraint condition, projects to proper vector subspace by the eigenmatrix that textural characteristics is formed;
Identification module, in proper vector subspace, training logistic regression sorter model, utilizes described logistic regression sorter model to carry out sex identification to human face region.
Further, described shear module specifically comprises:
Computing unit, calculates the rectangular characteristic of testing image as Weak Classifier for using integrogram;
Ballot unit, picking out the Weak Classifier representing face, according to the mode of Nearest Neighbor with Weighted Voting, Weak Classifier being configured to strong classifier for using Adaboost algorithm;
Series unit, for last cascade filtering of will the some strong classifiers obtained be trained to be composed in series a cascade structure.
Further, described characteristic extracting module specifically comprises:
Positioning unit, carries out facial modeling for utilizing active shape model;
Unit for scaling, for the human face characteristic point position obtained according to active shape model, rotates the human face region of testing image and ajusts, carry out convergent-divergent, row interpolation process of going forward side by side to this human face region;
Image block retrieval unit, for centered by groups of people's face characteristic point position, marks off multiple image block in human face region;
Feature extraction unit, for extracting textural characteristics on each image block.
Ins and outs about above-mentioned modular unit are existing in method above to be described in detail, therefore repeats no more.
Should be understood that, application of the present invention is not limited to above-mentioned citing, for those of ordinary skills, can be improved according to the above description or convert, and all these improve and convert the protection domain that all should belong to claims of the present invention.
Claims (10)
1., based on the gender identification method that multiple dimensioned linear Differential Characteristics low-rank represents, it is characterized in that, comprise step:
A, use Face datection algorithm based on Haar-like characteristic sum Adaboost sorter, shear out the human face region in testing image;
B, on the human face region sheared out, extract textural characteristics based on multiple dimensioned linear Differential Characteristics;
C, under low-rank constraint condition, by textural characteristics form eigenmatrix project to proper vector subspace;
D, in proper vector subspace, training logistic regression sorter model, utilizes described logistic regression sorter model to carry out sex identification to human face region.
2. the gender identification method represented based on multiple dimensioned linear Differential Characteristics low-rank according to claim 1, it is characterized in that, described steps A specifically comprises:
A1, use integrogram calculate the rectangular characteristic of testing image as Weak Classifier;
A2, use Adaboost algorithm pick out the Weak Classifier representing face, according to the mode of Nearest Neighbor with Weighted Voting, Weak Classifier are configured to strong classifier;
A3, the cascade filtering of finally will the some strong classifiers obtained be trained to be composed in series a cascade structure.
3. the gender identification method represented based on multiple dimensioned linear Differential Characteristics low-rank according to claim 1, it is characterized in that, described step B specifically comprises:
B1, active shape model is utilized to carry out facial modeling;
B2, the human face characteristic point position obtained according to active shape model, rotate the human face region of testing image and ajust, carry out convergent-divergent, row interpolation process of going forward side by side to this human face region;
B3, centered by groups of people's face characteristic point position, in human face region, mark off multiple image block;
B4, on each image block, extract textural characteristics.
4. the gender identification method represented based on multiple dimensioned linear Differential Characteristics low-rank according to claim 1, is characterized in that, in described step C, project by following formula:
Wherein, W is the subspace projection matrix obtained that projects,
for between class scatter matrix,
for Scatter Matrix in class,
for eigenmatrix projects to the nuclear norm behind proper vector subspace, for limiting the size of the rear subspace projection rank of matrix of projection.
5. the gender identification method represented based on multiple dimensioned linear Differential Characteristics low-rank according to claim 4, it is characterized in that, in described step D, by subspace projection matrix W the value on the different dimensions of textural characteristics is weighted and obtains weighted value, and weighted value is compressed to the scope of 0 ~ 1 with logical function.
6. the gender identification method represented based on multiple dimensioned linear Differential Characteristics low-rank according to claim 5, is characterized in that, in described step D, the sex of training sample for woman's probability is
,
for the low-rank of textural characteristics represents.
7. the gender identification method represented based on multiple dimensioned linear Differential Characteristics low-rank according to claim 6, it is characterized in that, in described step D, utilize gradient descent method, the loss function minimizing subspace projection matrix W obtains the optimum solution of subspace projection matrix W.
8., based on the sex recognition system that multiple dimensioned linear Differential Characteristics low-rank represents, it is characterized in that, comprising:
Shear module, for using the Face datection algorithm based on Haar-like characteristic sum Adaboost sorter, shears out the human face region in testing image;
Characteristic extracting module, for extracting the textural characteristics based on multiple dimensioned linear Differential Characteristics on the human face region sheared out;
Low-rank representation module, under low-rank constraint condition, projects to proper vector subspace by the eigenmatrix that textural characteristics is formed;
Identification module, in proper vector subspace, training logistic regression sorter model, utilizes described logistic regression sorter model to carry out sex identification to human face region.
9. the sex recognition system represented based on multiple dimensioned linear Differential Characteristics low-rank according to claim 8, it is characterized in that, described shear module specifically comprises:
Computing unit, calculates the rectangular characteristic of testing image as Weak Classifier for using integrogram;
Ballot unit, picking out the Weak Classifier representing face, according to the mode of Nearest Neighbor with Weighted Voting, Weak Classifier being configured to strong classifier for using Adaboost algorithm;
Series unit, for last cascade filtering of will the some strong classifiers obtained be trained to be composed in series a cascade structure.
10. the sex recognition system represented based on multiple dimensioned linear Differential Characteristics low-rank according to claim 8, it is characterized in that, described characteristic extracting module specifically comprises:
Positioning unit, carries out facial modeling for utilizing active shape model;
Unit for scaling, for the human face characteristic point position obtained according to active shape model, rotates the human face region of testing image and ajusts, carry out convergent-divergent, row interpolation process of going forward side by side to this human face region;
Image block retrieval unit, for centered by groups of people's face characteristic point position, marks off multiple image block in human face region;
Feature extraction unit, for extracting textural characteristics on each image block.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510895459.7A CN105550642B (en) | 2015-12-08 | 2015-12-08 | Gender identification method and system based on multiple dimensioned linear Differential Characteristics low-rank representation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510895459.7A CN105550642B (en) | 2015-12-08 | 2015-12-08 | Gender identification method and system based on multiple dimensioned linear Differential Characteristics low-rank representation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105550642A true CN105550642A (en) | 2016-05-04 |
CN105550642B CN105550642B (en) | 2019-03-22 |
Family
ID=55829825
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510895459.7A Active CN105550642B (en) | 2015-12-08 | 2015-12-08 | Gender identification method and system based on multiple dimensioned linear Differential Characteristics low-rank representation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105550642B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107085734A (en) * | 2017-05-24 | 2017-08-22 | 南京华设科技股份有限公司 | IN service accepts robot |
CN107341457A (en) * | 2017-06-21 | 2017-11-10 | 北京小米移动软件有限公司 | Method for detecting human face and device |
CN107609459A (en) * | 2016-12-15 | 2018-01-19 | 平安科技(深圳)有限公司 | A kind of face identification method and device based on deep learning |
CN110263621A (en) * | 2019-05-06 | 2019-09-20 | 北京迈格威科技有限公司 | Image-recognizing method, device and readable storage medium storing program for executing |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102902986A (en) * | 2012-06-13 | 2013-01-30 | 上海汇纳网络信息科技有限公司 | Automatic gender identification system and method |
CN103198303A (en) * | 2013-04-12 | 2013-07-10 | 南京邮电大学 | Gender identification method based on facial image |
-
2015
- 2015-12-08 CN CN201510895459.7A patent/CN105550642B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102902986A (en) * | 2012-06-13 | 2013-01-30 | 上海汇纳网络信息科技有限公司 | Automatic gender identification system and method |
CN103198303A (en) * | 2013-04-12 | 2013-07-10 | 南京邮电大学 | Gender identification method based on facial image |
Non-Patent Citations (2)
Title |
---|
李同宇 等: "基于特征融合的人脸图像性别识别", 《智能系统学报》 * |
杨卫国 等: "自适应搜索目标的快速彩色复合人脸检测方法", 《计算机工程与设计》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107609459A (en) * | 2016-12-15 | 2018-01-19 | 平安科技(深圳)有限公司 | A kind of face identification method and device based on deep learning |
CN107609459B (en) * | 2016-12-15 | 2018-09-11 | 平安科技(深圳)有限公司 | A kind of face identification method and device based on deep learning |
CN107085734A (en) * | 2017-05-24 | 2017-08-22 | 南京华设科技股份有限公司 | IN service accepts robot |
CN107341457A (en) * | 2017-06-21 | 2017-11-10 | 北京小米移动软件有限公司 | Method for detecting human face and device |
CN110263621A (en) * | 2019-05-06 | 2019-09-20 | 北京迈格威科技有限公司 | Image-recognizing method, device and readable storage medium storing program for executing |
CN110263621B (en) * | 2019-05-06 | 2021-11-26 | 北京迈格威科技有限公司 | Image recognition method and device and readable storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN105550642B (en) | 2019-03-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Yao et al. | Revisiting co-saliency detection: A novel approach based on two-stage multi-view spectral rotation co-clustering | |
CN102902986A (en) | Automatic gender identification system and method | |
Guo et al. | Facial expression recognition using ELBP based on covariance matrix transform in KLT | |
CN110163239A (en) | A kind of Weakly supervised image, semantic dividing method based on super-pixel and condition random field | |
Ming | Robust regional bounding spherical descriptor for 3D face recognition and emotion analysis | |
Ming | Rigid-area orthogonal spectral regression for efficient 3D face recognition | |
CN105550641B (en) | Age estimation method and system based on multi-scale linear differential texture features | |
Zhou et al. | Pose-robust face recognition with Huffman-LBP enhanced by divide-and-rule strategy | |
Sharma et al. | Emotion recognition using facial expression by fusing key points descriptor and texture features | |
Zhao et al. | Bisecting k-means clustering based face recognition using block-based bag of words model | |
Huang et al. | Local circular patterns for multi-modal facial gender and ethnicity classification | |
Fang et al. | Real-time hand posture recognition using hand geometric features and fisher vector | |
Duan et al. | Expression of Concern: Ethnic Features extraction and recognition of human faces | |
Li et al. | Head-shoulder based gender recognition | |
CN108446672A (en) | A kind of face alignment method based on the estimation of facial contours from thick to thin | |
Paul et al. | Extraction of facial feature points using cumulative histogram | |
CN105550642A (en) | Gender identification method and system based on multi-scale linear difference characteristic low-rank expression | |
Wan et al. | Palmprint recognition system for mobile device based on circle loss | |
CN104376312A (en) | Face recognition method based on word bag compressed sensing feature extraction | |
CN103942572A (en) | Method and device for extracting facial expression features based on bidirectional compressed data space dimension reduction | |
Kumar et al. | Automatic face detection using genetic algorithm for various challenges | |
Jiang et al. | Discriminating features learning in hand gesture classification | |
Kaur et al. | Comparative analysis of image classification techniques using statistical features in CBIR systems | |
Li et al. | Spatial and temporal information fusion for human action recognition via Center Boundary Balancing Multimodal Classifier | |
Muthukumar et al. | Vision based hand gesture recognition for Indian sign languages using local binary patterns with support vector machine classifier |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right | ||
TR01 | Transfer of patent right |
Effective date of registration: 20200702 Address after: 23 / F, Guangdong hi tech District Patentee after: Shenzhen Konka Holding Group Co., Ltd Address before: 518057 Konka 28 R & D building, twelve hi tech Industrial Zone, Shenzhen hi tech Industrial Park, Guangdong, Nanshan District 23 Patentee before: KONKA GROUP Co.,Ltd. |