CN107704810A - A kind of expression recognition method suitable for medical treatment and nursing - Google Patents
A kind of expression recognition method suitable for medical treatment and nursing Download PDFInfo
- Publication number
- CN107704810A CN107704810A CN201710828671.0A CN201710828671A CN107704810A CN 107704810 A CN107704810 A CN 107704810A CN 201710828671 A CN201710828671 A CN 201710828671A CN 107704810 A CN107704810 A CN 107704810A
- Authority
- CN
- China
- Prior art keywords
- shape
- mrow
- point
- expression
- msub
- 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
- G06V40/174—Facial expression recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a kind of expression recognition method suitable for medical treatment and nursing, comprise the following steps：Active shape model is established using training image；Test image is input in active shape model and searches for human face characteristic point；Extraction includes the mood vector characteristics of angle information and range information, and human face characteristic point is substituted into mood vector characteristics, and the information of mood vector characteristics is inputted and voted into SVM classifier, and who gets the most votes's class is affiliated class.
Description
Technical field
The present invention relates to a kind of image processing techniques, particularly a kind of expression recognition method suitable for medical treatment and nursing.
Background technology
Expression recognition technology is in recent years with the rapid development such as machine learning of some association areas, at image
Reason, become the technology of a focus development.The influence of expression recognition system and potentiality are generalized to vast application simultaneously
In occasion, such as man-machine interaction, intelligent robot, driver status supervision etc..Expression recognition system is computer understanding
The premise of people's emotion, and people explore intelligence, understand the effective way of intelligence.Personalizing for computer how is realized, is made
It can adaptively provide most friendly operation ring according to contents such as the states of the environment of surrounding and object for communicatee
Border, oneself is intelligent robot asking of coming into that people's daily life must solve through turning into the target of man-machine interface of future generation development
Topic, to establishing multi information intelligent man-machine interaction system important in inhibiting.Meanwhile as China human mortality old-age group trend is aggravated,
Medical treatment and nursing industry welcomes rapid development, on the road of medical care industry intelligent development, to the Expression Recognition of old physical disabilities,
The identification of particularly negative feeling has great importance in actual applications.
Expression recognition technology generally comprises three parts content：(1) Face datection；(2) extraction of human face expression feature；
(3) classification of expressive features.Wherein human face expression feature extraction be in whole system the most core the step of, feature extraction is direct
Have influence on the precision, robustness and real-time of identification.The method of common face characteristic extraction includes：Based on geometric properties, table
See feature, the method for behavioral characteristics.Generally, although expression recognition accurately extracts expression by development for many years
Feature is still exactly the technical barrier of a urgent need to resolve so as to carry out the identification of strong robustness human face expression, while carries out expression certainly
The accuracy and real-time of dynamic identification still have to be hoisted.
The content of the invention
It is an object of the invention to provide a kind of expression recognition method suitable for medical treatment and nursing, comprise the following steps：
Step 1, active shape model is established using training image；
Step 2, test image is input in active shape model and searches for human face characteristic point；
Step 3, extraction includes the mood vector characteristics of angle information and range information, by human face characteristic point substitution mood to
In measure feature, and the information of mood vector characteristics is inputted and voted into SVM classifier, who gets the most votes's class is affiliated
Class.
Using the above method, step 1 specifically includes：
Step 1.1, N training images are obtained and n mark point is marked on training image and represents flat shape with X
X=[x1,x2,...xj...xn,y1,y2,...yj...yn]T (1)
Wherein, (xj,yj) it is j-th of coordinate for marking point；
Step 1.2, the average shape of training image is obtained
Step 1.3, the characteristic vector of the covariance matrix of average shape is obtained
Step 1.4, by characteristic vectorIn answer characteristic value to arrange from big to small and q characteristic vector forms before therefrom choosing
Matrix Φs
Φs=[Φ1,...Φj,...Φq] (4)
Step 1.5, active shape model Y is obtaineds
Wherein, bsIt is the parameter set of shape.
Using the above method, middle test image is input in active shape model searches for face characteristic according to the following steps
Point：
Step 2.1, i=0 is made, uses average shapeAs initialization shape Xi；
Step 2.2, to current shape, mahalanobis distance is calculated for each mark point, selects the point of minimum range to be used as and is somebody's turn to do
Mark the new position (x of pointj',yj')；
Step 2.3, adjusting parameter tx、ty、θ、s、bs, make (xj',yj') withMatching, (tx,ty) it is figure
As translation distance, θ is image rotation angle, and s is image scaling size, bsFor the parameter set of shape, M represents the mistake of adjustment
Journey, M-1For M inverse transformation；
Step 2.4, makeIf Xi=Xi+1, then exit；Otherwise, i=i+1,2.2 are gone to step.
Using the above method, the angle information angle triangle for three human face characteristic points in step 3, if this three
Angular angle is in the angular range corresponding to an expression, then expression is then fallen into corresponding to the angular range corresponding to the angle
Expression in；Range information is the distance between two human face characteristic points in step 3, if the distance corresponding to an expression away from
From in the range of, then expression is then fallen into the expression corresponding to the distance range corresponding to the distance.
The present invention proposes a kind of new Extraction of Geometrical Features method based on ASM, calculates vector angle and distance feature,
Make Expression Recognition more accurate.
The present invention is described further with reference to Figure of description.
Brief description of the drawings
Fig. 1 is expression recognition schematic flow sheet.
Fig. 2 is the manual calibration maps of human face characteristic point in embodiment.
Fig. 3 is face mood vector characteristics angle and distance feature schematic diagram in embodiment.
Embodiment
With reference to Fig. 1, a kind of expression recognition method suitable for medical treatment and nursing, comprise the following steps：
Step 1, active shape model is established using training image；
Step 2, test image is input in active shape model and searches for human face characteristic point；
Step 3, extraction includes the mood vector characteristics of angle information and range information, by human face characteristic point substitution mood to
In measure feature, and the information of mood vector characteristics is inputted and voted into SVM classifier, who gets the most votes's class is affiliated
Class.
" class " can simply be interpreted as the recognition result of svm classifier, and identification process is to utilize software LIBSVM adjusting parameters
Classification results are obtained, are prior arts, therefore do not illustrate design parameter in the patent.Last recognition result is glad respectively, in
Property and sad three major types.Obtain this result and think that identification is completed.
Specifically
Step 1：Active shape model is established using training image
An ASM model is established first with training image, marks a number of point on training picture manually, these
Point all concentrates on as countenance changes and changes most places, such as eyebrow, lip, eyes etc..If these point marks
Inaccuracy, recognition result can decline.The facial image of characteristic point mark is as shown in Figure 2.
Active shape model (ASM) is a kind of Object shape description technology, is a kind of feature matching method based on model,
It both can neatly change the shape of model to adapt to the uncertain characteristic of target shape, and the change of shape is controlled in mould
Type allow in the range of, so as to ensure model change when will not be affected by various factors and there is irrational shape.Its base
This thought is to choose one group of training sample, and the shape of sample is described with one group of characteristic point, and then the shape of each sample is carried out
Registering (so that shape is similar as much as possible), carries out statistics using principal component method to the shape vector after these registrations and builds
Mould obtains the statistical description of body form, finally contour of object is searched in new image using the model of foundation, depending on
Position goes out target object.
Assuming that n point forms an object, N is the number of all training images, and vectorial X represents flat shape：
X=[x1,x2,...xj...xn,y1,y2,...yj...yn]T
Wherein, (xj,yj) coordinate of j-th of mark point is represented, n is the number for forming all points of body form.
The average shape of one object, it can be obtained from following formula：
The characteristic vector of the covariance matrix of average shapeExpression formula is：
Therefrom select q maximum characteristic vector of character pair value, composition matrix Φs, expression formula is：
Then obtaining ASM models from the training atlas of manual feature point for calibration is
Wherein, YsIt is the shape of target object,It is average shape, bsIt is the parameter set of shape, ΦsIt is association
The matrix of the characteristic vector composition of variance matrix, it is orthogonal.
Step 2：Test image is input in ASM models, scans for matching.
Step 201：I=0 is made, uses average shapeAs initialization shape Xi；
Step 202：To current shape in each mark point, mahalanobis distance is calculated, selects that point conduct of ultimate range
New position (the x of the calibration pointj',yj')；
Step 203：The transformation parameter t such as adjustment translation, rotation and yardstickx,ty, θ, s, and form parameter bs, make (xj',
yj') withMatching, orderM represents such a operation, is by image translation (tx、
ty), image rotation θ, graphical rule scaling s, bsThe parameter set of shape, can using visual interpretation as：Arbitrary shape can be with
Approximate representation is " deformation " to average shape, and this " deformation " is to variously-shaped change weighting by form parameter
With model.These parameters are adjusted, are commonly called as " aliging ", in order to make the optimal mark glyph calculated under mahalanobis distance
Close ASM models.M-1That is M inverse transformation, I did not attempted this algorithm in person, was seen in bibliography；
Step 204：If Xi=Xi+1Then exit, otherwise go to step 202.
Mahalanobis distance calculating process in step 202 includes：
First to every width training image feature point for calibration edge by its angular bisector direction, using the characteristic point of demarcation in
The heart takes npIndividual pixel, the following vector of its gray value composition
Gray scale can be obtained prolong the derivative in angular bisector direction and be：
After the derivative is standardized, have
To carrying out identical calculating on each image, obtain j-th of mark point on all images half-tone information it is flat
Average
Wherein, i represents i width images, represents j-th of mark point of jth width image
Calculate covariance
For standardized grayscale vector h' corresponding to a certain mark point in test imagej, with Average normalized gray scale derivative
The mahalanobis distance of vector
In summary, the matching of ASM models is exactly to utilize the profile of the target object in algorithm search test image.Searching
Suo Zhong, the optimized parameter of target shape is obtained by comparison reference model and test image, object wheel is searched in new image
Exterior feature, so as to orient target object.
Further, the present invention is improved for the method for existing extraction geometric properties, in order that it is more applicable
In medical treatment and nursing industry, i.e., there is more unitized judgement to patient's passiveness expression (including sadness, pain etc.), to geometric properties
Improved, it is proposed that mood vector characteristics (Motion Vector Feature)
Easily found from the picture in JAFFE expressions storehouse, people's horizontal length of mouth when glad and sad be it is different,
But the direct length of mouth is extracted to carry out judgement be unscientific, and because the mouth of people may be big, Ke Neng little, but the nose of people
It is rigid, it is not easy to change, just form a triangle with three points on nose and corners of the mouth both sides, the triangle can be utilized
Some features carry out expression identification, it is specific as follows：
Triangle OAB, it is extracted first mood vector characteristics angle information.Assuming that it is respectively with 3 points of coordinate：
Wherein i=1,2 ... N
Wherein, N is the number of facial image,J=43,44,49 is the seat on three summits in the i-th width image
Mark represents that the label of the point of nose is 43, and the label at corners of the mouth both ends is 44,49 respectively.
It is possible to further calculateWithAngle, can be used for that difference is sad, neutral glad expression,
Formula is as follows：
By observing JAFFE expression datas storehouse, it can be seen that when happiness, the mouth of people is usually what is opened, and sad
When, the mouth of people closes, and extracts distance feature, that is, represents four points of upper lip and lower mouth
The distance between four corresponding points of lip, can be obtained by following formula
d5961=| y59-y61|
Wherein, y59It is the 59th point of ordinate, y61It is the 61st point of ordinate, d5961It is the vertical seat of the 59th point and the 61st point
The absolute value of the difference of mark, remaining three similar.
Because every face there are 66 points, the coordinate of these points, also drop it off in mood vector characteristics, it is possible to use
Following vector expression represents the shape of face：
X=[x1,x2,...xj...xn,y1,y2,...yj...yn]T
The feature thus needed, it will send these characteristic informations in SVM classifier voted below,
Who gets the most votes's class is affiliated class.
Claims (4)
1. a kind of expression recognition method suitable for medical treatment and nursing, it is characterised in that comprise the following steps：
Step 1, active shape model is established using training image；
Step 2, test image is input in active shape model and searches for human face characteristic point；
Step 3, extraction includes the mood vector characteristics of angle information and range information, and it is special that human face characteristic point is substituted into mood vector
In sign, and the information of mood vector characteristics is inputted and voted into SVM classifier, who gets the most votes's class is affiliated class.
2. according to the method for claim 1, it is characterised in that step 1 specifically includes：
Step 1.1, N training images are obtained and n mark point is marked on training image and represents flat shape with X
X=[x1,x2,...xj...xn,y1,y2,...yj...yn]T (1)
Wherein, (xj,yj) it is j-th of coordinate for marking point；
Step 1.2, the average shape of training image is obtained
<mrow>
<mover>
<mi>X</mi>
<mo>&OverBar;</mo>
</mover>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mi>N</mi>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<msub>
<mi>X</mi>
<mi>j</mi>
</msub>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
Step 1.3, the characteristic vector of the covariance matrix of average shape is obtained
Step 1.4, by characteristic vectorIn answer characteristic value to arrange from big to small and q characteristic vector forms matrix Φ before therefrom choosings
Φs=[Φ1,...Φj,...Φq] (4)
Step 1.5, active shape model Y is obtaineds
<mrow>
<msub>
<mi>Y</mi>
<mi>s</mi>
</msub>
<mo>=</mo>
<mover>
<mi>X</mi>
<mo>&OverBar;</mo>
</mover>
<mo>+</mo>
<msub>
<mi>&Phi;</mi>
<mi>s</mi>
</msub>
<msub>
<mi>b</mi>
<mi>s</mi>
</msub>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>5</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, bsIt is the parameter set of shape.
3. according to claim, the method described in 2, it is characterised in that in by test image be input in active shape model by
Following steps search for human face characteristic point：
Step 2.1, i=0 is made, uses average shapeAs initialization shape Xi；
Step 2.2, to current shape, mahalanobis distance is calculated for each mark point, selects the point of minimum range as the mark
New position (the x of pointj',yj')；
Step 2.3, adjusting parameter tx、ty、θ、s、bs, make (xj',yj') withMatching, (tx,ty) it is image translation
Distance, θ are image rotation angle, and s is image scaling size, bsFor the parameter set of shape, M represents the process of adjustment, M-1
For M inverse transformation；
Step 2.4, makeIf Xi=Xi+1, then exit；Otherwise, i=i+1,2.2 are gone to step.
4. according to the method for claim 1, it is characterised in that angle information is three human face characteristic point institute structures in step 3
Into the angle of triangle, if the angle of the triangle in the angular range corresponding to an expression, expression corresponding to the angle
Then fall into the expression corresponding to the angular range；
Range information is the distance between two human face characteristic points in step 3, if the distance corresponding to an expression apart from model
In enclosing, then expression is then fallen into the expression corresponding to the distance range corresponding to the distance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710828671.0A CN107704810A (en) | 2017-09-14 | 2017-09-14 | A kind of expression recognition method suitable for medical treatment and nursing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710828671.0A CN107704810A (en) | 2017-09-14 | 2017-09-14 | A kind of expression recognition method suitable for medical treatment and nursing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107704810A true CN107704810A (en) | 2018-02-16 |
Family
ID=61171744
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710828671.0A Pending CN107704810A (en) | 2017-09-14 | 2017-09-14 | A kind of expression recognition method suitable for medical treatment and nursing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107704810A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109409296A (en) * | 2018-10-30 | 2019-03-01 | 河北工业大学 | The video feeling recognition methods that facial expression recognition and speech emotion recognition are merged |
| CN109753950A (en) * | 2019-02-11 | 2019-05-14 | 河北工业大学 | Dynamic human face expression recognition method |
| CN111127521A (en) * | 2019-10-25 | 2020-05-08 | 上海联影智能医疗科技有限公司 | System and method for generating and tracking the shape of an object |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101777131A (en) * | 2010-02-05 | 2010-07-14 | 西安电子科技大学 | Method and device for identifying human face through double models |
| CN102831388A (en) * | 2012-05-23 | 2012-12-19 | 上海交通大学 | Method and system for detecting real-time characteristic point based on expanded active shape model |
| CN104036255A (en) * | 2014-06-21 | 2014-09-10 | 电子科技大学 | Facial expression recognition method |
| CN104376333A (en) * | 2014-09-25 | 2015-02-25 | 电子科技大学 | Facial expression recognition method based on random forests |
| CN104951743A (en) * | 2015-03-04 | 2015-09-30 | 苏州大学 | Active-shape-model-algorithm-based method for analyzing face expression |
-
2017
- 2017-09-14 CN CN201710828671.0A patent/CN107704810A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101777131A (en) * | 2010-02-05 | 2010-07-14 | 西安电子科技大学 | Method and device for identifying human face through double models |
| CN102831388A (en) * | 2012-05-23 | 2012-12-19 | 上海交通大学 | Method and system for detecting real-time characteristic point based on expanded active shape model |
| CN104036255A (en) * | 2014-06-21 | 2014-09-10 | 电子科技大学 | Facial expression recognition method |
| CN104376333A (en) * | 2014-09-25 | 2015-02-25 | 电子科技大学 | Facial expression recognition method based on random forests |
| CN104951743A (en) * | 2015-03-04 | 2015-09-30 | 苏州大学 | Active-shape-model-algorithm-based method for analyzing face expression |
Non-Patent Citations (4)
| Title |
|---|
| CHEN CHIUNG HSIEH ET AL: "A Facial Expression Classification System Based on Active Shape Model and Support Vector Machine", 《2011 INTERNATIONAL SYMPOSIUM ON COMPUTER SCIENCE AND SOCIETY》 * |
| KE WANG ET AL: "Real-time facial expressions recognition system for service robot based-on ASM and SVMs", 《2010 8TH WORLD CONGRESS ON INTELLIGENT CONTROL AND AUTOMATION》 * |
| YU ZHENGHONG ET AL: "Research of Facial Expression Recognition Based on ASM Model and RS-SVM", 《2014 FIFTH INTERNATIONAL CONFERENCE ON INTELLIGENT SYSTEMS DESIGN AND ENGINEERING APPLICATIONS》 * |
| 王林路等: "基于情感几何特征和支持向量机的人脸表情识别研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109409296A (en) * | 2018-10-30 | 2019-03-01 | 河北工业大学 | The video feeling recognition methods that facial expression recognition and speech emotion recognition are merged |
| CN109409296B (en) * | 2018-10-30 | 2020-12-01 | 河北工业大学 | Video emotion recognition method integrating facial expression recognition and voice emotion recognition |
| CN109753950A (en) * | 2019-02-11 | 2019-05-14 | 河北工业大学 | Dynamic human face expression recognition method |
| CN109753950B (en) * | 2019-02-11 | 2020-08-04 | 河北工业大学 | Dynamic facial expression recognition method |
| CN111127521A (en) * | 2019-10-25 | 2020-05-08 | 上海联影智能医疗科技有限公司 | System and method for generating and tracking the shape of an object |
| CN111127521B (en) * | 2019-10-25 | 2024-03-01 | 上海联影智能医疗科技有限公司 | System and method for generating and tracking shape of target |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109325398B (en) | Human face attribute analysis method based on transfer learning | |
| CN104850825B (en) | A kind of facial image face value calculating method based on convolutional neural networks | |
| CN105447473B (en) | A kind of any attitude facial expression recognizing method based on PCANet-CNN | |
| Lemaire et al. | Fully automatic 3D facial expression recognition using differential mean curvature maps and histograms of oriented gradients | |
| CN100382751C (en) | Canthus and pupil location method based on VPP and improved SUSAN | |
| WO2020103417A1 (en) | Bmi evaluation method and device, and computer readable storage medium | |
| CN108614999B (en) | Eye opening and closing state detection method based on deep learning | |
| CN107704810A (en) | A kind of expression recognition method suitable for medical treatment and nursing | |
| CN112232184B (en) | Multi-angle face recognition method based on deep learning and space conversion network | |
| CN101763503A (en) | Face recognition method of attitude robust | |
| JP2018055470A (en) | Facial expression recognition method, facial expression recognition apparatus, computer program, and advertisement management system | |
| CN103902992B (en) | Human face recognition method | |
| CN111985532B (en) | Scene-level context-aware emotion recognition deep network method | |
| CN1687957A (en) | Man face characteristic point positioning method of combining local searching and movable appearance model | |
| CN106980819A (en) | Similarity judgement system based on human face five-sense-organ | |
| CN1776711A (en) | Method for searching new position of feature point using support vector processor multiclass classifier | |
| CN107392151A (en) | Face image various dimensions emotion judgement system and method based on neutral net | |
| CN104794441B (en) | Human face characteristic positioning method based on active shape model and POEM texture models under complex background | |
| CN106897706B (en) | A kind of Emotion identification device | |
| CN109409298A (en) | A kind of Eye-controlling focus method based on video processing | |
| CN111611912B (en) | Detection method for pedestrian head-falling abnormal behavior based on human body joint point | |
| CN1936925A (en) | Method for judging characteristic point place using Bayes network classification device image | |
| CN108898623A (en) | Method for tracking target and equipment | |
| CN114758382A (en) | Face AU detection model establishing method and application based on adaptive patch learning | |
| TW201040846A (en) | Face detection apparatus and face detection method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180216 |