CN105608737B - A kind of human foot three-dimensional rebuilding method based on machine learning - Google Patents
A kind of human foot three-dimensional rebuilding method based on machine learning Download PDFInfo
- Publication number
- CN105608737B CN105608737B CN201610069095.1A CN201610069095A CN105608737B CN 105608737 B CN105608737 B CN 105608737B CN 201610069095 A CN201610069095 A CN 201610069095A CN 105608737 B CN105608737 B CN 105608737B
- Authority
- CN
- China
- Prior art keywords
- foot
- point
- machine learning
- dimensional
- model
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Computer Graphics (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to dimensional Modeling Technologies, and in particular to a kind of human foot three-dimensional rebuilding method based on machine learning belongs to computer vision technique application field.The invention discloses a kind of human foot three-dimensional rebuilding method based on machine learning, contact type measurement scheme measuring speed is slow, precision is low, troublesome in poeration in solution traditional technology, although and non-contact measurement scheme measurement accuracy is high, problem still cumbersome, at high cost.This method obtains the foot picture of several positions using image acquiring device at random, then the method for machine learning is used, utilize the good foot model of precondition, obtain the key point of foot, then it is approached using the foot threedimensional model of key point driving standard, the threedimensional model of foot is obtained finally by iteration, and the parameter of foot can be calculated according to the threedimensional model of reconstruction.
Description
Technical field
The present invention relates to dimensional Modeling Technologies, and in particular to a kind of human foot three-dimensional reconstruction side based on machine learning
Method belongs to computer vision technique application field.
Background technique
Three-dimensional reconstruction is the mathematical procedure and computer technology for restoring object dimensional information (shape etc.) using two-dimensional projection.
It provides the threedimensional model of accurate geometry information and photorealistic according to the data reconstruction of real scene.
Three-dimensional reconstruction is carried out to human foot, can more improve and effectively digitize foot physiological characteristic, it is simple with
Its relevant form characteristic parameter is advantageously measured, is had in fields such as Foot-biomechanics research, medicine, sport, shoemaking, 3D printings
Be widely applied.
There is contact and two kinds contactless to foot three-dimensional reconstruction scheme at present:
1) contact measurement method uses probe contacts foot surfaces, is squeezed foot and is caused to a certain extent
Foot surfaces deform, and the foot parameter being achieved in that is inaccurate；In addition the measuring speed of contact type measurement mode is slow, if
Measure entire foot profile, measuring speed is also the restraining factors considered in foot measurement, while contact type measurement side
Formula operation is more troublesome, time-consuming and laborious, will necessarily be replaced quick Contactless foot measurement method.
2) contactless measurement method combination photoelectric technology, computer technology and electronic scanning device, in foot measurement
During do not need directly to contact the foot tri-dimensional facial type that foot surfaces can restore the three-dimensional appearance of foot, and obtain
Data have that data volume is more, measuring speed is fast, advantage with high accuracy compared with the foot relevant parameter that contact method obtains,
Heavy manual labor is got rid of in measurement process, is the prefered method in foot measurement field.
The method that foot non-contact measurement rebuilds threedimensional model at present has: laser scanning, structure light or binocular triangle
Measurement etc., although its measurement accuracy is high, measuring system and measuring principle are complicated, at high cost, cumbersome, only professional people
Scholar could complete.
Summary of the invention
The technical problems to be solved by the present invention are: proposing a kind of human foot three-dimensional reconstruction side based on machine learning
Method, solves in traditional technology that contact type measurement scheme measuring speed is slow, precision is low, troublesome in poeration, and non-contact measurement scheme
Although measurement accuracy is high, problem still cumbersome, at high cost.
The present invention solves scheme used by above-mentioned technical problem:
A kind of human foot three-dimensional rebuilding method based on machine learning, comprising the following steps:
A, choosing, there is the article of standard two-dimensional scale to be placed near foot as object of reference, and guarantee abundant with foot
Contact；
B, obtained at random using image acquiring device foot two open more than image；
C, the foot feature on the image obtained in trained foot feature point model identification and annotation step B is utilized
Point；
D, the attitude parameter of foot is determined according to the foot characteristic point obtained in step C, and according to the attitude parameter tune
The posture of whole general three-dimensional foot model；
E, corresponding points of each foot characteristic point obtained in step C on general three-dimensional foot model, acquisition pair are determined
Answer the foot point cloud of part；
F, the step E foot point cloud obtained is normalized to obtain the point cloud data comprising true coordinate；
G, matching alignment is carried out to all the points cloud, obtains the point cloud data of entire foot；
H, interpolation is carried out to the point cloud data of entire foot, the intensive point cloud data of entire foot is obtained, to obtain foot
The 3D model in portion；
I, texture mapping, the 3D foot finally rebuild are carried out to the foot 3D model of acquisition.
As advanced optimizing, in step A, selection standard A4 paper steps on the foot of 3D model to be reconstructed as object of reference
On the standard A4 paper；Here it is not limited in A4 paper, but any object that can provide standard two-dimensional scale, such as identity
Card etc. can also be with.
As advanced optimizing, in step B, when obtaining foot image file at random using image acquiring device, guarantee to obtain
At least each image at left and right sides of to foot.
As advanced optimizing, in step C, the characteristic point is used to characterize the key point and profile of foot, comprising: after foot
With point, longest toe cusp, l articulationes metatarsophalangeae bump, the 5th articulationes metatarsophalangeae bump, internal malleolus cusp, external malleolus cusp, navicular bone highest
It selects, medial arch highest point, big toe side salient point, toe front end highest point, l articulationes metatarsophalangeae upper limb are selected, support millet cake.
The beneficial effects of the present invention are: the present invention passes through the foot picture of single image acquisition device random shooting, utilize
The method of machine learning carries out 3D reconstruction to foot, multiple sensors is not necessarily to, without calibration；Because the method is simple, it is easy to real
Existing, precision is high, strong operability.
Detailed description of the invention
Fig. 1 is foot characteristic point schematic diagram.
Specific embodiment
The present invention is directed to propose a kind of method for carrying out foot three-dimensional reconstruction using machine learning principle, may be implemented pair
The measurement of foot relevant parameter describes human foot physiological phenomenon with more perfect, accurate digitized forms.
This method obtains the foot picture of several positions using image acquiring device at random, then uses the side of machine learning
Method obtains the key point of foot using the good foot model of precondition, and then the foot using key point driving standard is three-dimensional
Model is approached, and obtains the threedimensional model of foot finally by iteration, and foot can be calculated according to the threedimensional model of reconstruction
The parameter in portion.
Embodiment:
Human foot three-dimensional rebuilding method in this example includes following implemented step:
(1) it is placed around an article with standard two-dimensional (long and wide) scale in the foot of 3D model to be reconstructed and is used as and join
According to object, which has with foot adequately contacts (such as: foot is stepped on a standard A4 paper)；
(2) two images above (left and right of foot is obtained using image acquiring device random (without accurately knowing angle)
Two sides are one at least each)；
(3) using the foot feature of the image obtained in trained foot feature point model identification and mark (2)
Point, characteristic point are used to characterize the key point and profile of foot, as shown in Figure 1, comprising: heel point 1, longest toe cusp 2, the
L articulationes metatarsophalangeae bump 3, the 5th articulationes metatarsophalangeae bump 4, internal malleolus cusp 5, external malleolus cusp 6, navicular bone highest point 7, medial arch are most
High point 8, big toe side salient point 9, toe front end highest point 10, l articulationes metatarsophalangeae upper limb select 11, support millet cake 12.
(4) 3 are utilized) in characteristic point determine the attitude parameter of foot, and adjust general three-dimensional foot according to attitude parameter
The posture of model；Here general three-dimensional foot model is a standard 3D model, and the purpose of this step is by universal model
Pose adjustment is the same visual angle of foot image file to be reconstructed.
(5) corresponding points of the characteristic point in (4) on general three-dimensional foot model in (3) are determined, corresponding part is obtained
Foot point cloud；
(6) object of reference in (1) is utilized, the foot point cloud of acquisition is normalized to obtain the point cloud number comprising true coordinate
According to；
(7) matching alignment is carried out to all the points cloud, obtains the point cloud data of entire foot；
(8) interpolation is carried out to above-mentioned point cloud data, the intensive point cloud data of foot is obtained, to obtain the 3D mould of foot
Type；
(9) texture mapping, the 3D foot finally rebuild are carried out to the foot model in (8).
Claims (4)
1. a kind of human foot three-dimensional rebuilding method based on machine learning, which comprises the following steps:
A, choosing, there is the article of standard two-dimensional scale to be placed near foot as object of reference, and guarantee sufficiently to connect with foot
Touching；
B, using single image acquisition device random shooting foot two open more than image；
C, referring to the foot characteristic point on the image obtained in trained foot feature point model identification and annotation step B；
D, the attitude parameter of foot is determined according to the foot characteristic point obtained in step C, and is adjusted and led to according to the attitude parameter
With the posture of three-dimensional foot model；
E, corresponding points of each foot characteristic point obtained in step C on general three-dimensional foot model are determined, correspondence portion is obtained
The foot point cloud divided；
F, the step E foot point cloud obtained is normalized to obtain the point cloud data comprising true coordinate；
G, matching alignment is carried out to all the points cloud, obtains the point cloud data of entire foot；
H, interpolation is carried out to the point cloud data of entire foot, the intensive point cloud data of entire foot is obtained, to obtain foot
3D model；
I, texture mapping, the 3D foot finally rebuild are carried out to the foot 3D model of acquisition.
2. a kind of human foot three-dimensional rebuilding method based on machine learning as described in claim 1, which is characterized in that step
In A, selection standard A4 paper steps down in the foot of 3D model to be reconstructed on the standard A4 paper as object of reference.
3. a kind of human foot three-dimensional rebuilding method based on machine learning as described in claim 1, which is characterized in that step
In B, when obtaining foot image file at random using image acquiring device, guarantee gets at least each figure at left and right sides of foot
Picture.
4. a kind of human foot three-dimensional rebuilding method based on machine learning as described in claim 1, which is characterized in that step
In C, the characteristic point is used to characterize the key point and profile of foot, comprising: heel point, longest toe cusp, l articulationes metatarsophalangeae
Bump, the 5th articulationes metatarsophalangeae bump, internal malleolus cusp, external malleolus cusp, navicular bone highest point, medial arch highest point, big toe side are convex
It selects, toe front end highest point, l articulationes metatarsophalangeae upper limb are selected, support millet cake.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610069095.1A CN105608737B (en) | 2016-02-01 | 2016-02-01 | A kind of human foot three-dimensional rebuilding method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610069095.1A CN105608737B (en) | 2016-02-01 | 2016-02-01 | A kind of human foot three-dimensional rebuilding method based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105608737A CN105608737A (en) | 2016-05-25 |
| CN105608737B true CN105608737B (en) | 2019-01-08 |
Family
ID=55988648
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610069095.1A Active CN105608737B (en) | 2016-02-01 | 2016-02-01 | A kind of human foot three-dimensional rebuilding method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105608737B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| IT202100023615A1 (en) * | 2021-09-14 | 2023-03-14 | Trya S R L | METHOD AND SYSTEM FOR MEASURING A FOOT AND GENERATING A THREE-DIMENSIONAL MODEL OF A FOOT |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI607412B (en) * | 2016-09-10 | 2017-12-01 | 財團法人工業技術研究院 | Measurement systems and methods for measuring multi-dimensions |
| CN106815428B (en) * | 2017-01-13 | 2020-05-19 | 中国空气动力研究与发展中心高速空气动力研究所 | Wind tunnel balance calibration data processing method based on intelligent optimization algorithm |
| CN109330106B (en) * | 2018-11-01 | 2021-08-24 | 成都牛晶科技有限公司 | Foot code size measuring method based on mobile phone photographing |
| CN109636907A (en) * | 2018-12-13 | 2019-04-16 | 谷东科技有限公司 | A kind of terrain reconstruction method and system based on AR glasses |
| CN109840592B (en) * | 2018-12-24 | 2019-10-18 | 梦多科技有限公司 | A kind of method of Fast Labeling training data in machine learning |
| CN109815830A (en) * | 2018-12-28 | 2019-05-28 | 梦多科技有限公司 | A method of obtaining foot information in the slave photo based on machine learning |
| CN109887077B (en) * | 2019-03-07 | 2022-06-03 | 百度在线网络技术（北京）有限公司 | Method and apparatus for generating three-dimensional model |
| CN110379001B (en) * | 2019-07-04 | 2023-04-07 | 新拓三维技术(深圳)有限公司 | Foot product customization method and device, terminal equipment and computer readable storage medium |
| CN112257582A (en) * | 2020-10-21 | 2021-01-22 | 北京字跳网络技术有限公司 | Foot posture determination method, device, equipment and computer readable medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090306801A1 (en) * | 2006-11-27 | 2009-12-10 | Northeastern University | Patient specific ankle-foot orthotic device |
| CN101658347A (en) * | 2009-09-24 | 2010-03-03 | 浙江大学 | Method for obtaining dynamic shape of foot model |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104021589A (en) * | 2014-06-27 | 2014-09-03 | 江苏中佑石油机械科技有限责任公司 | Three-dimensional fitting simulating method |
-
2016
- 2016-02-01 CN CN201610069095.1A patent/CN105608737B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090306801A1 (en) * | 2006-11-27 | 2009-12-10 | Northeastern University | Patient specific ankle-foot orthotic device |
| CN101658347A (en) * | 2009-09-24 | 2010-03-03 | 浙江大学 | Method for obtaining dynamic shape of foot model |
Non-Patent Citations (1)
| Title |
|---|
| 基于线结构光扫描的脚型重构测量研究;袁振宇;《中国优秀硕士学位论文全文数据库》;20140815(第08期);第9-56页 |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| IT202100023615A1 (en) * | 2021-09-14 | 2023-03-14 | Trya S R L | METHOD AND SYSTEM FOR MEASURING A FOOT AND GENERATING A THREE-DIMENSIONAL MODEL OF A FOOT |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105608737A (en) | 2016-05-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105608737B (en) | A kind of human foot three-dimensional rebuilding method based on machine learning | |
| CN103971409A (en) | Measuring method for foot three-dimensional foot-type information and three-dimensional reconstruction model by means of RGB-D camera | |
| CN104126989B (en) | A kind of based on the foot surfaces 3 D information obtaining method under multiple stage RGB-D pick up camera | |
| US10460517B2 (en) | Mobile device human body scanning and 3D model creation and analysis | |
| CN107111833B (en) | Fast 3D model adaptation and anthropometry | |
| CN107767442A (en) | A kind of foot type three-dimensional reconstruction and measuring method based on Kinect and binocular vision | |
| Peyer et al. | Subject-specific body segment parameter estimation using 3D photogrammetry with multiple cameras | |
| CN102157013A (en) | System for fully automatically reconstructing foot-type three-dimensional surface from a plurality of images captured by a plurality of cameras simultaneously | |
| CN104408762A (en) | Method for obtaining object image information and three-dimensional model by using monocular unit and two-dimensional platform | |
| US11779242B2 (en) | Systems and methods to estimate human length | |
| CN107041585A (en) | The measuring method of human dimension | |
| CN110074788A (en) | A kind of body data acquisition methods and device based on machine learning | |
| Bragança et al. | Current state of the art and enduring issues in anthropometric data collection | |
| Kaashki et al. | Deep learning-based automated extraction of anthropometric measurements from a single 3-D scan | |
| CN105212436B (en) | The measuring system and measuring method of non-contact 3-D laser foot type | |
| CN104665107B (en) | A kind of vola three dimensional data collection processing system and data acquisition treatment method | |
| CN106767433B (en) | A kind of method and system measuring foot sizing | |
| Bragança et al. | An overview of the current three-dimensional body scanners for anthropometric data collection | |
| CN103075977A (en) | Automatic combining algorithm for point cloud data in binocular stereoscopic vision system | |
| KR101557492B1 (en) | Apparatus and Method for generating user's three dimensional body model based on depth information | |
| CN116966086A (en) | Human back acupoints calibrating method and system based on real-time image optimization | |
| TWI542320B (en) | Human weight estimating method by using depth images and skeleton characteristic | |
| Krefer et al. | A method for generating 3D thermal models with decoupled acquisition | |
| CN109815830A (en) | A method of obtaining foot information in the slave photo based on machine learning | |
| CN109801326A (en) | It is a kind of for obtaining the image measuring method of human somatotype data |
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 |