CN106971190A - Sexual discriminating method based on human somatotype - Google Patents
Sexual discriminating method based on human somatotype Download PDFInfo
- Publication number
- CN106971190A CN106971190A CN201710132428.5A CN201710132428A CN106971190A CN 106971190 A CN106971190 A CN 106971190A CN 201710132428 A CN201710132428 A CN 201710132428A CN 106971190 A CN106971190 A CN 106971190A
- Authority
- CN
- China
- Prior art keywords
- sex
- trunk
- steps
- physical characteristic
- discriminating method
- 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.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/48—Extraction of image or video features by mapping characteristic values of the pattern into a parameter space, e.g. Hough transformation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/752—Contour matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/755—Deformable models or variational models, e.g. snakes or active contours
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The invention discloses a kind of Sexual discriminating method based on human somatotype, the Sexual discriminating method comprises the following steps：S01, set up human somatotype gender data storehouse；S02, the mathematical modeling for setting up trunk physical characteristic model and sex；S03, the human body front photograph for obtaining the measured and side are shone；S04, the mathematical modeling for setting up the trunk physical characteristic vector input of the measured, obtain sex.User of the present invention provides front and shone, sideways according to the other judgement of i.e. realizability, simple to operate, efficiency high and accuracy of measurement height.
Description
Technical field
The invention belongs to human engineering of clothing field, and in particular to a kind of to carry out human body using human body contour outline graph data
Method for distinguishing is not known.
Background technology
Sex identification is the process for allowing computer to judge sex according to information such as sound, the images of the people of input, artificial
There is important prospect in terms of intelligence, pattern-recognition.Traditional sex identification is main special using physiology such as voice, face-images
Levy progress computer identification (CN201310128913, CN201110170752, CN201210515116 etc.).
The present invention carries out sex identification using the trunk outline shape of human body, and can apply to monitoring etc. can not obtain face
The neighborhood such as image or voice.
In addition in 3D anthropometric scanning system, obtain after human 3d model data, it is necessary to know from three-dimensional data
The measurement characteristic point of others' body.Due to the otherness of men and women's build, before human body feature point identification is carried out, first to measurement object
Sex carry out automatic discrimination and be then more beneficial for the accurate extraction of characteristics of human body's information.
The content of the invention
The technical problem to be solved in the present invention is to overcome the mode of Sexual discriminating of the prior art to exist not
Foot carries out the side of Sexual discriminating there is provided a kind of morphological differences according to men and women's build using the contour pattern of trunk part
Method.
The present invention is to solve above-mentioned technical problem by following technical proposals；
A kind of Sexual discriminating method based on human somatotype, its feature is that the Sexual discriminating method includes following step
Suddenly：
S01, set up human somatotype gender data storehouse.Gather a large amount of human somatotype gender datas, including trunk build
Characteristic vector and sex.
First, in S011 steps, collection human body front is shone, shone sideways and sex.
In S012 steps, from human body front according to in people's body side surface photograph, rear neck point, side neck point, thigh root point, armpit are chosen
Lower point, shoulder point are characterized a little, obtain front face human body trunk contour curve and side trunk contour curve, to front and side
Contour curve carries out curve fitting, and to selected curvilinear function dimensionality reduction and chooses component of the contribution rate more than preset value as human body
The characteristic vector of trunk build.
Preferably, in S012, curve matching selection elliptic curve function.
Preferably, in S012, selecting Fourier transformation mode.The oval Fourier descriptor carries out principal component point
Analysis, calculating obtains dimensionality reduction matrix, and takes accumulation contribution rate to be more than the dimensionality reduction principal component of default contribution rate, sets up trunk feature
Vector.
Preferably, in S012, it is 90% to preset contribution rate.
Finally, in S013, the trunk characteristic vector obtained in S011, S012 and sex input database are built
Vertical human somatotype gender data storehouse.
S02 steps, set up Sexual discriminating model.S02 is using linearly or nonlinearly model, and linear model is preferred to use polynary
Linear regression method；Nonlinear model optimization BP neural network method.
Preferably, S02 steps use BP neural network method, pass through the human somatotype gender data storehouse set up to S01 steps
Study obtain the neutral net mould of input parameter (trunk physical characteristic vector) and output parameter (sex) mapping relations
Type.
Trunk physical characteristic vector and the measurement model of sex are set up by S01, S02, collection can be passed through accordingly
Measured people's trunk physical characteristic vector, obtains sex.Judgement to the sex of single the measured is walked by S03, S04
It is rapid to complete.
S03 steps, obtain the measured relevant information, are shone and side photograph by shooting the front of acquisition the measured.
S04 steps, complete the judgement of sex.First, shone in the S041 human body fronts for obtaining S03 and side is shone and used
S012 identical modes obtain trunk physical characteristic vector；Then, by the trunk physical characteristic obtained from S041 to
The Sexual discriminating mathematical modeling that amount input S02 steps are obtained, can obtain sex.
The present invention has the advantages that easy realization, efficiency high, recognition accuracy are high.
(1) sex identification can be carried out to object by only needing to the contour pattern geometric data of trunk.
(2) for front profile, the frontal outline of human body can be used, back side profile can also be used.Also without
The mug shot of high-resolution.
(3) growing up, there were significant differences on build by men and women, and the accuracy rate recognized using this method is high.
Brief description of the drawings
Fig. 1 is the flow chart of the Sexual discriminating method based on human somatotype of a preferred embodiment of the present invention.
Fig. 2 is the specific implementation flow chart of the Sexual discriminating method based on human somatotype of a preferred embodiment of the present invention.
Fig. 3 extracts the schematic diagram of trunk profile for the present invention from direct picture and side image.
Embodiment
The present invention is described in further details below with reference to Figure of description and specific embodiment.But therefore will not
Limit the invention among described scope of embodiments.
As shown in figure 1, the Sexual discriminating method based on human somatotype of the present invention comprises the following steps：
S01, set up human somatotype gender data storehouse；
S02, the mathematical modeling for setting up trunk physical characteristic model and sex；
S03, the human body front photograph for obtaining the measured and side are shone；
S04, the mathematical modeling for setting up the trunk physical characteristic vector input of the measured, obtain sex.
Wherein S01 specifically includes following steps：
S011：Obtain full face, side photo and the sex of human body；
S012：Obtain trunk physical characteristic vector；
S013：By trunk physical characteristic vector sum sex typing human somatotype gender data storehouse.
S04 specifically includes following steps：
S041, the trunk physical characteristic vector for obtaining the measured；
S042, input trunk physical characteristic vector and the mathematical modeling of sex obtain sex.
As Figure 2-3, the present invention is in specific implementation process, and idiographic flow comprises the following steps:
Step 101 shoots front and lateral plan photo and the outline data for extracting human body of human body.
In the present embodiment, carrying out the front of target body, lateral plan picture-taking step is：
(1.1) target body, which, can highlight the fitted garment of human body contour outline；Both arms are stretched, and with horizontal line into about 45 °
Angle opens one's arms；Two legs are opened about with shoulder with wide；Ensure that left and right wheels clean up clear separation (Fig. 2) under oxter and crotch.
(1.2) camera heights are adjusted to concordant with human body waist, keep camera to look squarely, human body is occupied in shooting picture
Middle position, shoots complete human body photo.Preferably：Use 35~50mm of equivalent focal length lens shooting；Using with human body
The colour of skin and the high solid background of dressing contrast.
(1.3) application image contour detecting and extracting method, human body wheel is carried out to the front of shooting and lateral plan photo
The extraction of wide geometric data.Preferably：Use the contours extract algorithm based on Active Shape Model Method and contour mould.
The characteristic point information of template is remained in the outline data of extraction.
(1.4) the human body feature point position in Fig. 2, takes out trunk partial contour graph data.
Step 102, the outline data extracted in step 101 is carried out into oval Fourier techniques to describe.Specially：
(2.1) profile is described as a series of pixel point set U=(xi, yi) i=1,2 ..., n.Wherein n is composition profile
Pixel number evidence.
(2.2) contour curve is converted into oval Fourier descriptor by the following method：
The oval Fourier expansion that each pixel is projected on x, y-axis is
Wherein：A0 is profile central point x coordinate；C0 is profile central point y-coordinate；N is overtone order；N is maximum harmonic wave
The oval number of number of times, i.e. close approximation；Accumulation displacements of the t for point along profile, i.e. profile plays point-to-point p arc length.T tires out for total
Product displacement, i.e. profile girth.Profile is discrete for K sampled point approximate description, then the oval coefficient An of X-direction, Bn is respectively：
The oval coefficient Cn of Y-direction, Dn are respectively：
Wherein：K is total sampling number of profile；N is overtone order；Δ xp is from profile point p to profile point p+1 point-to-point transmissions
Along the displacement of X-direction；Δ yp is the displacement from profile point p to profile point p+1 point-to-point transmissions along Y direction；T is profile week
It is long；Δ tp is the distance between profile point p to profile point p+1, i.e.,:
The central point O of profile is in x, the DC component of the projection in y-axis direction, i.e. Fourier space, is respectively：
Wherein：
And：ε 1=σ 1=0
Step 103, the result of step 102 is carried out to the normalization of oval Fourier descriptor using following methods.
(a) location specification：The DC component A0=C0=0 of oval Fourier, i.e., first ellipse described profile
Center translation is to the origin of coordinates.
(b) measurement regulation：First oval size E of Fourier descriptor is calculated, by each description subsystem number divided by E.
After normalization is handled, obtain one group of oval Fourier descriptor unrelated with outline position, orientation, yardstick to
Amount.Preferably：The overtone order of oval Fourier is not less than 30.
Repeat step 101~103, obtains the trunk contour pattern data of multiple known sex samples, and is converted to specification
Oval Fourier descriptor after change.Using PCA, principal component analysis is carried out to all outline datas measured, will
Outline data dimensionality reduction is that one group of accumulation contribution rate is more than a certain proportion of principal component.Preferably accumulation contribution rate is more than 90%.
Step 201 and 202 sets up BP pattern recognition neural networks, using the principal component P after dimensionality reduction as input vector, by people
As output vector after body sex coding, by the use of the sample data measured as training data, neural network model is carried out
Train (step 201).
For the measured, outline data, the god that applying step 401 is set up using step 202 are obtained using 301~303
Through sex automatic identification of the network sex identification model to object to be measured.Wherein 301 steps are specifically scanned using camera or 3D
Instrument obtains human somatotype profile, and 302 steps are using the same processing mode of step 102；Wherein 303 steps use step 103
Same processing mode.
Claims (12)
1. a kind of Sexual discriminating method based on human somatotype, it is characterised in that the Sexual discriminating method comprises the following steps：
S01, set up human somatotype gender data storehouse；
S02, the mathematical modeling for setting up trunk physical characteristic model and sex；
S03, the human body front photograph for obtaining the measured and side are shone；
S04, the mathematical modeling for setting up the trunk physical characteristic vector input of the measured, obtain sex.
2. Sexual discriminating method according to claim 1, it is characterised in that comprise the following steps in the step S01,
S011：Obtain full face, side photo and the sex of human body；
S012：Obtain trunk physical characteristic vector；
S013：By trunk physical characteristic vector sum sex typing human somatotype gender data storehouse.
3. Sexual discriminating method according to claim 2, it is characterised in that obtained by step S01 and set up human somatotype
The trunk physical characteristic vector sum sex that gender data place needs, wherein：Sex is in S011 steps by traditional
Manual input-mode obtains；The human body front that trunk physical characteristic vector is obtained in S011 steps is shone and side is shone and obtained
By handling generation, and the typing human somatotype gender data storehouse in S013 steps in S012 steps.
4. Sexual discriminating method according to claim 3, it is characterised in that obtained in S012 steps using step S011
, according to extracting, choose rear neck point, side neck point, thigh root point, underarm point, shoulder point according to side and be characterized a little, acquisition is just in human body front
Face trunk contour curve and side trunk contour curve, carry out curve fitting to front and side profile curve, and
To selected curvilinear function dimensionality reduction and choose contribution rate more than the component of preset value and be used as the characteristic vector of trunk build to make
For trunk physical characteristic vector.
5. Sexual discriminating method according to claim 4, it is characterised in that matched curve uses elliptic function curve.
6. Sexual discriminating method according to claim 4, it is characterised in that curve dimensionality reduction uses elliptic function, by ellipse
Circle Fourier techniques describe the trunk contour curve to obtain oval Fourier descriptor, and to oval Fourier descriptor
Carry out principal component analysis and obtain dimensionality reduction matrix and dimensionality reduction principal component；Default contribution rate uses 90%, chooses contribution rate more than default
The component of contribution rate obtains trunk characteristic vector.
7. Sexual discriminating method according to claim 1, it is characterised in that S02 steps set up linearly or nonlinearly model,
The mathematical formulae of the output inputted by the fitting in the human somatotype gender data storehouse set up to S01 steps, the input
It is vectorial for trunk physical characteristic, it is described to be output as sex；Linear model uses multiple linear regression method；Nonlinear model BP
Neural network.
8. Sexual discriminating method according to claim 7, it is characterised in that S02 sets up trunk physical characteristic vector
BP neural network method is used with the mathematical modeling step of sex, passes through the human somatotype gender data storehouse set up to S01 steps
Study obtains the neural network model of input parameter and output parameter mapping relations；The input parameter be trunk feature to
Amount, the output parameter is sex.
9. Sexual discriminating method according to claim 1, it is characterised in that obtained by S03 steps and judge the measured
Information required for sex, the front that the measured is obtained by shooting is shone according to side.
10. Sexual discriminating method according to claim 1, it is characterised in that S04 obtains the sex of the measured, including
Following steps,
S041, the trunk physical characteristic vector for obtaining the measured；
S042, input trunk physical characteristic vector and the mathematical modeling of sex obtain sex.
11. Sexual discriminating method according to claim 10, it is characterised in that S041 steps to S03 steps by obtaining
The measured front according to and side shine, using S012 identicals method extract measurement needed for trunk physical characteristic to
Amount.
12. Sexual discriminating method according to claim 11, it is characterised in that input parameter is inputted into S02 steps and obtained
Mathematical modeling, exported, the input parameter be from step S041 steps obtain trunk physical characteristic vector, institute
State and be output as sex.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710132428.5A CN106971190A (en) | 2017-03-07 | 2017-03-07 | Sexual discriminating method based on human somatotype |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710132428.5A CN106971190A (en) | 2017-03-07 | 2017-03-07 | Sexual discriminating method based on human somatotype |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106971190A true CN106971190A (en) | 2017-07-21 |
Family
ID=59329278
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710132428.5A Withdrawn CN106971190A (en) | 2017-03-07 | 2017-03-07 | Sexual discriminating method based on human somatotype |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106971190A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110210293A (en) * | 2018-10-30 | 2019-09-06 | 上海市服装研究所有限公司 | A kind of gender identification method based on three-dimensional data and face-image |
| CN110222564A (en) * | 2018-10-30 | 2019-09-10 | 上海市服装研究所有限公司 | A method of sex character is identified based on three-dimensional data |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101799865A (en) * | 2010-02-25 | 2010-08-11 | 上海复控华龙微系统技术有限公司 | Pedestrian space-time outline presenting method based on ellipse Fourier decomposition |
| CN105125172A (en) * | 2015-06-24 | 2015-12-09 | 广州三瑞医疗器械有限公司 | Analgesia control method and device based on facial image recognition |
| CN106022343A (en) * | 2016-05-19 | 2016-10-12 | 东华大学 | Fourier descriptor and BP neural network-based garment style identification method |
| CN106326834A (en) * | 2016-07-29 | 2017-01-11 | 华讯方舟科技有限公司 | Human body gender automatic identification method and apparatus |
-
2017
- 2017-03-07 CN CN201710132428.5A patent/CN106971190A/en not_active Withdrawn
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101799865A (en) * | 2010-02-25 | 2010-08-11 | 上海复控华龙微系统技术有限公司 | Pedestrian space-time outline presenting method based on ellipse Fourier decomposition |
| CN105125172A (en) * | 2015-06-24 | 2015-12-09 | 广州三瑞医疗器械有限公司 | Analgesia control method and device based on facial image recognition |
| CN106022343A (en) * | 2016-05-19 | 2016-10-12 | 东华大学 | Fourier descriptor and BP neural network-based garment style identification method |
| CN106326834A (en) * | 2016-07-29 | 2017-01-11 | 华讯方舟科技有限公司 | Human body gender automatic identification method and apparatus |
Non-Patent Citations (1)
| Title |
|---|
| 程伟: ""非接触式人体三维测量—人体特征点识别及轮廓线提取"", 《中国优秀硕士学位论文全文数据库 工程科技I辑》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110210293A (en) * | 2018-10-30 | 2019-09-06 | 上海市服装研究所有限公司 | A kind of gender identification method based on three-dimensional data and face-image |
| CN110222564A (en) * | 2018-10-30 | 2019-09-10 | 上海市服装研究所有限公司 | A method of sex character is identified based on three-dimensional data |
| CN110232309A (en) * | 2018-10-30 | 2019-09-13 | 上海市服装研究所有限公司 | A method of based on three-dimensional data and neural network recognization sex character |
| CN110222564B (en) * | 2018-10-30 | 2022-12-13 | 上海市服装研究所有限公司 | Method for identifying gender characteristics based on three-dimensional data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108921100B (en) | Face recognition method and system based on visible light image and infrared image fusion | |
| CN102982557B (en) | Method for processing space hand signal gesture command based on depth camera | |
| CN107016721A (en) | The modeling method of human 3d model | |
| CN108520216B (en) | Gait image-based identity recognition method | |
| CN104063702B (en) | Three-dimensional gait recognition based on shielding recovery and partial similarity matching | |
| CN109919141A (en) | A kind of recognition methods again of the pedestrian based on skeleton pose | |
| CN103049758B (en) | Merge the remote auth method of gait light stream figure and head shoulder mean shape | |
| CN107145842A (en) | With reference to LBP characteristic patterns and the face identification method of convolutional neural networks | |
| CN102982323B (en) | Gait recognition method fast | |
| CN107908288A (en) | A kind of quick human motion recognition method towards human-computer interaction | |
| CN107423712B (en) | 3D face recognition method | |
| CN109299659A (en) | A kind of human posture recognition method and system based on RGB camera and deep learning | |
| CN103984922B (en) | Face identification method based on sparse representation and shape restriction | |
| CN104200200B (en) | Fusion depth information and half-tone information realize the system and method for Gait Recognition | |
| CN104573634A (en) | Three-dimensional face recognition method | |
| CN106991380A (en) | A kind of preprocess method based on vena metacarpea image | |
| CN111985332B (en) | Gait recognition method of improved loss function based on deep learning | |
| CN109101864A (en) | The upper half of human body action identification method returned based on key frame and random forest | |
| CN109766796A (en) | A kind of depth pedestrian detection method towards dense population | |
| CN107240128A (en) | A kind of X-ray film and photochrome method for registering based on contour feature | |
| CN111858997B (en) | Cross-domain matching-based clothing template generation method | |
| CN110210293A (en) | A kind of gender identification method based on three-dimensional data and face-image | |
| CN110348344B (en) | Special facial expression recognition method based on two-dimensional and three-dimensional fusion | |
| CN108154176A (en) | A kind of 3D human body attitude algorithm for estimating for single depth image | |
| CN106971190A (en) | Sexual discriminating method based on human somatotype |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20170721 |
|
| WW01 | Invention patent application withdrawn after publication |