CN105654029B - The three-dimensional point cloud Ear recognition method of accuracy of identification and efficiency can be improved - Google Patents
The three-dimensional point cloud Ear recognition method of accuracy of identification and efficiency can be improved Download PDFInfo
- Publication number
- CN105654029B CN105654029B CN201510856611.0A CN201510856611A CN105654029B CN 105654029 B CN105654029 B CN 105654029B CN 201510856611 A CN201510856611 A CN 201510856611A CN 105654029 B CN105654029 B CN 105654029B
- Authority
- CN
- China
- Prior art keywords
- auricle
- point cloud
- model
- dimensional
- vector
- 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.)
- Expired - Fee Related
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
-
- 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
- G06V10/42—Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/64—Three-dimensional objects
- G06V20/653—Three-dimensional objects by matching three-dimensional models, e.g. conformal mapping of Riemann surfaces
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Human Computer Interaction (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention discloses a kind of three-dimensional point cloud Ear recognition method that accuracy of identification and efficiency can be improved, it is characterised in that carries out in accordance with the following steps: being decomposed based on PCA and SVD, pretreatment is normalized to the position and posture of three-dimensional auricle point cloud model；4 local characteristic regions of three-dimensional auricle point cloud model are extracted based on Iannarelli categorizing system；It is matched using local characteristic region of the Sparse ICP algorithm to three-dimensional auricle point cloud model, obtains whole initial matching points pair；It is based onSVDSolve rotation transformation between matching double pointsRAnd translation transformationT, minimize error functionE, until obtaining optimal registration when range error obtains minimum value between model.
Description
Technical field
The present invention relates to a kind of three-dimensional point cloud Ear recognition methods, especially a kind of to can be improved the three of accuracy of identification and efficiency
Dimension point cloud Ear recognition method.
Background technique
Auricle is made of the part such as helix, anthelix, fossa helicis, fossa triangularis, tragus, antitragus, ear-lobe, has significant rise
Shape and unique three-dimensional shape features are lied prostrate, generality, uniqueness, permanent and easy collectivity are had more.Ear recognition skill
Art is biological identification technology emerging in recent years, is mainly known using three-dimensional shape features such as the gully complications of human ear shape
Not, mankind's biological characteristic such as same palm, fingerprint, iris, DNA is the same, is permanent biological identification.With two-dimentional Ear recognition skill
The extraneous factors such as art is compared, and three-dimensional auricle posture is illuminated by the light influence is smaller, has significant healthy and strong sexual clorminance.The three-dimensional shaped of auricle
Shape feature is not influenced by factors such as hair style, expression, beard, makeup, glasses, the colour of skin, illumination, and between 7 ~ 70 years old, the mankind
The structure and shape of auricle will not be substantially change；There is also measurable differences for even twinborn auricle；Therefore with
Other mankind's biological characteristics are compared, and the three-dimensional shape features of auricle have the stability and uniqueness of height.
Three-dimensional Ear recognition generally comprises three steps: auricle detection, feature extraction and characteristic matching, wherein feature mentions
Take, characteristic matching be three-dimensional Ear recognition key problem.Existing three-dimensional Ear recognition is mostly based on iteration closest approach
(iterative closest point, ICP) algorithm matches the full auricle of three-dimensional auricle model.ICP algorithm passes through
Loop iteration, the repeatedly position of intense adjustment point cloud model and posture minimize overall registration error, to realize overall best
Matching.But ICP algorithm computation complexity is higher, efficiency is lower and accuracy of identification is to be improved.
The Iannarelli categorizing system in two dimensional image space is the anatomical features based on auricle, calculates auricle important set
It at the similarity of part, is indicated with the line segment on two dimensional image, line segment 1 ~ 4 is the width of outer helix as shown in Figure 1:, and line segment 5 is
The length of partial fossa triangularis, line segment 6 ~ 8 are helix the distance between to outer helix, line segment 9 ~ 11 for crus helicis to anthelix it
Between distance, line segment 12 be ear-lobe length.However not only the size of each part is required to manual survey to this method on 2d
Amount, can not be accurately positioned, while can not directly apply to three-dimensional point cloud Ear recognition.
Summary of the invention
The present invention be in order to solve above-mentioned technical problem present in the prior art, provide one kind can be improved accuracy of identification and
The three-dimensional point cloud Ear recognition method of efficiency.
The technical solution of the invention is as follows: a kind of three-dimensional point cloud Ear recognition side that accuracy of identification and efficiency can be improved
Method, it is characterised in that carry out in accordance with the following steps:
A. it is decomposed based on PCA and SVD, pretreatment is normalized to the position and posture of three-dimensional auricle point cloud model；
B. 4 local characteristic regions of three-dimensional auricle point cloud model are extracted based on Iannarelli categorizing system；
C. it is matched, is obtained complete using local characteristic region of the Sparse ICP algorithm to three-dimensional auricle point cloud model
Portion's initial matching point pair；It is based onSVDSolve rotation transformation between matching double pointsRAnd translation transformationT, minimize error functionE, directly
When obtaining minimum value to range error between model, optimal registration is obtained.
Steps are as follows by a:
If the vertex set of Arbitrary 3 D auricle point cloud model, thenV'sRank geometric moment are as follows:
ThenVThree 1 rank geometric moments and six 2 rank geometric moments be respectively as follows:
For vertex setVAny vertex (x i , y i , z i ) do such as down conversion:
Thus willVMass centerMove to coordinate origin；
It is constructed with six 2 rank geometric momentsVCovariance matrix it is as follows:
If, i.e., to covariance matrixMSVD decomposition is carried out, whereinUFor matrixMFeature vector
Matrix, corresponding auricle modelVThree main shafts, Δ is matrixMEigenvalue matrix, to auricle modelVAny vertex (x i ,y i , z i ) such as down conversion:
To make the first main shaft of auricle model withyAxis alignment, the second main shaft withxAxis alignment, third main shaft withzAxis pair
Together, realize that all auricle point cloud models normalize to almost the same posture and position in database.
The b step is as follows:
Using the first main shaft of Arbitrary 3 D auricle point cloud model and the second main shaft as a pair of of direction, while the two being led
The diagonal of axis is used as another pair of direction, does normal plane then along this 4 directions and auricle model asks friendship, finally extract auricle
In 4 local characteristic regions.
The step c is as follows:
IfXWithY4 local feature areas of respectively three-dimensional auricle point cloud reference model and three-dimensional auricle point cloud object module
Domain, noteXWithYBetween alignment error vector be, then, and noteXWithYBetween distance be, wherein, and, and remember between modelXWithYError function be
Alignment error vectorlpNorm:
Wherein,；
Enable reference modelXA upper pointx i With object moduleYOn the point nearest apart from the pointy j Form matching double points, iterate to calculate, obtainXWithYBetween whole initial matching points pair；
It is based onSVDSolve rotation transformation between matching double pointsRAnd translation transformationT, minimize error functionE, until model
Between range error obtain minimum value when, obtain optimal registration.
The step c can also be in accordance with the following steps:
IfXWithY4 local feature areas of respectively three-dimensional auricle point cloud reference model and three-dimensional auricle point cloud object module
Domain, noteXWithYBetween alignment error vector be, then, and noteXWithYBetween distance be, wherein, and, and remember between modelXWithYError function be
Alignment error vectorlpNorm, based on Lagrangian method by alignment error vectorlpNorm is defined as:
Wherein,,For Lagrange multiplier,For punishment
The factor is broken down into 3 subproblems using change of direction multiplier method:
(1)
(2)
(3)
Wherein,,, pass throughh i Vector contraction operator asks above-mentioned subproblem
Solution, algorithm flow are as follows:
4 local characteristic region X, Y initial alignments of step 1. reference model and object module, initialization Lagrange
Multiply, penalty factor μ and threshold tau；
Step 2. forXArbitrary point, calculateYThe nearest point of the upper Euclidean distance point, obtainsXWithYBetween it is all initial
Matching double points；
Step 3. solves, pass throughh i Vector contraction operator solves,
Renewal vector；
Step 4. updates, solve function, and calculate spin matrix R and
Translation matrix t；
Step 5. solves, update；
If step 6.Greater than threshold tau, step 2 is returned to, otherwise iteration terminates.
The present invention is decomposed using principal component analysis (Principal Component Analysis, PCA) and SVD to auricle
The position of all three-dimensional auricle point cloud models in database, posture are normalized, and are then based on Iannarelli classification system
System extracts four local characteristic regions in three-dimensional auricle model, finally special to the part extracted using Sparse ICP algorithm
Sign region is matched, and realizes Ear recognition according to the divergence measurement between the Distance Judgment auricle between characteristic point.Experimental result
Show that the present invention carries out auricle local characteristic region to match accuracy of identification and recognition efficiency with higher.Especially introduce
Degree of rarefication optimizes Model registration, and utilizesNorm replaces Euclidean distance, and the quantity that distance is zero between maximization corresponding points is kept away
The problems such as having exempted from local alignment generation.Meanwhile algorithm redefines error function using Lagrangian method, solves alignment side
The problems such as model surface nonconvex property caused by formula and unflatness, and the error letter that will be redefined using change of direction multiplier method
Number is divided into three simple subproblems, and is solved by contraction operator, and the precision and stability of algorithm is improved.
Detailed description of the invention
Fig. 1 is the Iannarelli categorizing system auricle stepwise schematic views in two dimensional image space.
Fig. 2 is three-dimensional auricle point cloud model of embodiment of the present invention normalization front and back contrast effect figure.
Fig. 3 is that the embodiment of the present invention extracts 4 sub-regions schematic diagram of auricle model.
Fig. 4 is the embodiment of the present invention and prior art matching precision contrast schematic diagram.
Fig. 5 is the CMC curve comparison schematic diagram of the embodiment of the present invention and ICP algorithm.
Fig. 6 is the ROC curve contrast schematic diagram of the embodiment of the present invention and ICP algorithm.
Specific embodiment
The specific embodiment of the invention carries out on UND three-dimensional auricle database, which includes from 415 people's
1800 width three-dimensional auricles.
Specifically carry out in accordance with the following steps:
A. it is decomposed based on PCA and SVD, pretreatment is normalized to the position and posture of three-dimensional auricle point cloud model:
Since three-dimensional auricle data set building time span is larger, the same auricle data obtained on different acquisition times
It will be influenced by the factors such as the distance between collected auricle and acquisition equipment, angle, therefore firstly the need of to three-dimensional auricle
Pretreatment is normalized in auricle model in database.
Specific step is as follows:
If the vertex set of Arbitrary 3 D auricle point cloud model,
ThenV'sRank geometric moment are as follows:
ThenVThree 1 rank geometric moments and six 2 rank geometric moments be respectively as follows:
For vertex setVAny vertex (x i , y i , z i ) do such as down conversion:
Thus willVMass centerMove to coordinate origin；
It is constructed with six 2 rank geometric momentsVCovariance matrix it is as follows:
If, i.e., to covariance matrixMSVD decomposition is carried out, whereinUFor matrixMFeature vector
Matrix, corresponding auricle modelVThree main shafts, Δ is matrixMEigenvalue matrix, to auricle modelVAny vertex (x i ,y i , z i ) such as down conversion:
To make the first main shaft of auricle model withyAxis alignment, the second main shaft withxAxis alignment, third main shaft withzAxis pair
Together, realize that all auricle point cloud models normalize to almost the same posture and position in database.
Three-dimensional auricle point cloud model normalization front and back contrast effect is as shown in Fig. 2, wherein a is that three-dimensional auricle point cloud model is returned
Schematic diagram before one change；B is the effect picture before three-dimensional auricle point cloud model normalization.
B. 4 local characteristic regions of three-dimensional auricle point cloud model are extracted based on Iannarelli categorizing system:
Using the first main shaft of Arbitrary 3 D auricle point cloud model and the second main shaft as a pair of of direction, while the two being led
The diagonal of axis is used as another pair of direction, does normal plane then along this 4 directions and auricle model asks friendship, finally extract auricle
In 4 local characteristic regions.
As shown in Figure 3: it is special that this 4 local characteristic regions contain 12 important geometry in Iannarelli categorizing system
Sign.
C. it is matched, is obtained complete using local characteristic region of the Sparse ICP algorithm to three-dimensional auricle point cloud model
Portion's initial matching point pair；It is based onSVDSolve rotation transformation between matching double pointsRAnd translation transformationT, minimize error functionE, directly
When obtaining minimum value to range error between model, optimal registration is obtained.
It specifically can be as follows:
IfXWithY4 local feature areas of respectively three-dimensional auricle point cloud reference model and three-dimensional auricle point cloud object module
Domain, noteXWithYBetween alignment error vector be, then, and noteXWithYBetween distance be, wherein, and, and remember between modelXWithYError function be
Alignment error vectorlpNorm:
Wherein,；
Enable reference modelXA upper pointx i With object moduleYOn the point nearest apart from the pointy j Form matching double points, iterate to calculate, obtainXWithYBetween whole initial matching points pair；
It is based onSVDSolve rotation transformation between matching double pointsRAnd translation transformationT, minimize error functionE, until model
Between range error obtain minimum value when, obtain optimal registration.
It in order to advanced optimize registration problems, minimizes in error function solving, will be aligned based on Lagrangian method
Error vectorl p Norm redefines are as follows:
Wherein,For Lagrange multiplier,For penalty factor.
Using change of direction multiplier method (alternating direction method of multipliers, ADMM),
It is broken down into 3 subproblems:
(1)
(2)
(3)
Wherein,,.Pass throughh i Vector contraction operator (shrinkage
Operator) above-mentioned subproblem is solved, algorithm flow is as follows:
4 local characteristic region X, Y initial alignments of step 1. reference model and object module, initialization Lagrange
Multiply, penalty factorμAnd threshold tau；
Step 2. forXArbitrary point, calculateYThe nearest point of the upper Euclidean distance point, obtainsXWithYBetween it is all initial
Matching double points；
Step 3. solves, pass throughh i Vector contraction operator solves
, renewal vector；
Step 4. updates, solve function, and calculate spin matrix R and
Translation matrix t；
Step 5. solves, update；
If step 6.Greater than threshold tau, step 2 is returned to, otherwise iteration terminates.
Experimental result and analysis:
One, matching precision
Fig. 4 is to count after auricle model 05129d002ear to be measured is matched with auricle model Sparse ICP algorithm in database
The Euclidean distance quadratic sum similarity of calculation is distributed, and wherein lines a indicates that the embodiment of the present invention is carried out using the auricle region extracted
It is after matching as a result, lines b indicate matched using complete auricle model after result.Figure 4, it is seen that of the invention
Embodiment is higher using the matched auricle model accuracy of regional area.
Equally, it is matched for auricle model to be measured respectively at other auricle models in database with 05129d002ear,
Average distance between matching double points between computation model, it may be assumed that
Wherein,nIndicate the number of matching double points; MinDisIndicate the distance between matching double points; iIndicate matching double points
Serial number;kIndicate the serial number in region.Obviously,DisAvgSmaller, two distortions are higher；DisAvgIt is bigger, two moulds
Type similarity is lower.The average distance of average each local, can be obtained the similarity between auricle model.As shown in table 1.It is aobvious
So, for the embodiment of the present invention compared with ICP, ICNP algorithm, precision is higher, and can effective district split-phase with auricle different data mould
Type.
Average distance compares between 1 auricle of table
Assessment algorithm is come using cumulative matches characteristic (cumulative match characteristics, CMC) curve
Recognition efficiency, wherein CMC curve abscissa indicate auricle matching experimental result match it is best beforekA model is indulged
Coordinate representation is the accuracy of three-dimensional Ear recognition.Fig. 5 is to compare the CMC curve of the embodiment of the present invention and ICP algorithm.
From this figure, it can be seen that discrimination of the embodiment of the present invention is higher than the discrimination of ICP algorithm, wherein rank-1 has reached 93.8%.
It is correct using the reflection of receiver operating characteristic (receiver operating characteristic, ROC) curve
Receptance (genuine acceptance rate, GAR) and false acceptance rate (false acceptance rate, FAR)
Correlation, GAR indicates that correct auricle is considered as correct percentage, and FAR indicates that the auricle of mistake is considered as correct
Percentage.Fig. 6 is to compare the ROC curve of the embodiment of the present invention and ICP algorithm.From this figure, it can be seen that the present invention is real
The discrimination for applying example is higher than the discrimination of ICP algorithm.
Two, match times
This experiment Intel (R) Xeon (R) CPU based on 2.40 GHz, 16.0 GB RAM, 64 bit manipulation systems
It calculates in environment, pinna characteristics region is extracted respectively to 1 800 three-dimensional auricle models in UND three-dimensional auricle database, and
It compared carrying out Sparse ICP using complete auricle model, ICP, the match condition of ICNP algorithm is corresponding to match the used time pair
Than as shown in table 2, it can be seen that the matching used time based on auricle local characteristic region of the embodiment of the present invention is obviously smaller than and is based on
The matching used time of complete auricle model, and compared with ICP and ICNP algorithm, Sparse ICP algorithm time loss is less.
2 match time of table compares (s)
Conclusion: it is demonstrated experimentally that the embodiment of the present invention has very high accuracy of identification and efficiency compared with other algorithms.
Claims (4)
1. a kind of three-dimensional point cloud Ear recognition method that accuracy of identification and efficiency can be improved, it is characterised in that in accordance with the following steps into
Row:
A. it is decomposed based on PCA and SVD, pretreatment is normalized to the position and posture of three-dimensional auricle point cloud model；
B. 4 local characteristic regions of three-dimensional auricle point cloud model are extracted based on Iannarelli categorizing system:
Using the first main shaft of Arbitrary 3 D auricle point cloud model and the second main shaft as a pair of of direction, while by the two main shafts
It is diagonal to do normal plane as another pair of direction then along this 4 directions and auricle model asks friendship, it finally extracts 4 in auricle
A local characteristic region；
C. it is matched, is obtained all just using local characteristic region of the Sparse ICP algorithm to three-dimensional auricle point cloud model
Beginning matching double points；It is based onSVDSolve rotation transformation between matching double pointsRAnd translation transformationT, minimize error functionE, until mould
When range error obtains minimum value between type, optimal registration is obtained.
2. the three-dimensional point cloud Ear recognition method according to claim 1 that accuracy of identification and efficiency can be improved, feature exist
In a, steps are as follows:
If the vertex set of Arbitrary 3 D auricle point cloud model, thenV'sRank geometric moment are as follows:
ThenVThree 1 rank geometric moments and six 2 rank geometric moments be respectively as follows:
For vertex setVAny vertex (x i , y i , z i ) do such as down conversion:
,
Thus willVMass centerMove to coordinate origin；
It is constructed with six 2 rank geometric momentsVCovariance matrix it is as follows:
If, i.e., to covariance matrixMSVD decomposition is carried out, whereinUFor matrixMFeature vector square
Battle array, corresponding auricle modelVThree main shafts, Δ is matrixMEigenvalue matrix, to auricle modelVAny vertex (x i , y i ,z i ) such as down conversion:
To make the first main shaft of auricle model withyAxis alignment, the second main shaft withxAxis alignment, third main shaft withzAxis alignment, it is real
All auricle point cloud models normalize to almost the same posture and position in existing database.
3. the three-dimensional point cloud Ear recognition method according to claim 1 that accuracy of identification and efficiency can be improved, feature exist
It is as follows in the step c:
IfXWithY4 local characteristic regions of respectively three-dimensional auricle point cloud reference model and three-dimensional auricle point cloud object module,
NoteXWithYBetween alignment error vector be, then, and noteXWithYBetween distance be, wherein, and, and remember between modelXWithYError function be
Alignment error vectorlpNorm:
Wherein,；
Enable reference modelXA upper pointWith object moduleYOn the point nearest apart from the pointForm matching double points, iterate to calculate, obtainXWithYBetween whole initial matching points pair；
It is based onSVDSolve rotation transformation between matching double pointsRAnd translation transformationT, minimize error functionE, until model spacing
When obtaining minimum value from error, optimal registration is obtained.
4. the three-dimensional point cloud Ear recognition method according to claim 1 that accuracy of identification and efficiency can be improved, feature exist
It is as follows in the step c:
IfXWithY4 local characteristic regions of respectively three-dimensional auricle point cloud reference model and three-dimensional auricle point cloud object module,
NoteXWithYBetween alignment error vector be, then, and noteXWithYBetween distance be, wherein, and, and remember between modelXWithYError function be
Alignment error vectorlpNorm, based on Lagrangian method by alignment error vectorNorm is defined as:
Wherein,,For Lagrange multiplier,For punishment because
Son is broken down into 3 subproblems using change of direction multiplier method:
Wherein,,, pass throughVector contraction operator is to above-mentioned
Subproblem solves, and algorithm flow is as follows:
4 local characteristic region X, Y initial alignments of step 1. reference model and object module initialize Lagrange multiplier, penalty factorAnd threshold value；
Step 2. forXArbitrary point, calculateYThe nearest point of the upper Euclidean distance point, obtainsXWithYBetween whole initial matchings
Point pair；
Step 3. solves, pass throughVector contraction operator solves, renewal vector；
Step 4. updates, solve function, and calculate rotation
Matrix R and translation matrix t；
Step 5. solves, update；
If step 6.Greater than threshold value, step 2 is returned to, otherwise iteration terminates.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510856611.0A CN105654029B (en) | 2015-11-30 | 2015-11-30 | The three-dimensional point cloud Ear recognition method of accuracy of identification and efficiency can be improved |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510856611.0A CN105654029B (en) | 2015-11-30 | 2015-11-30 | The three-dimensional point cloud Ear recognition method of accuracy of identification and efficiency can be improved |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105654029A CN105654029A (en) | 2016-06-08 |
| CN105654029B true CN105654029B (en) | 2019-04-05 |
Family
ID=56481905
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510856611.0A Expired - Fee Related CN105654029B (en) | 2015-11-30 | 2015-11-30 | The three-dimensional point cloud Ear recognition method of accuracy of identification and efficiency can be improved |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105654029B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101812001B1 (en) | 2016-08-10 | 2017-12-27 | 주식회사 고영테크놀러지 | Apparatus and method for 3d data registration |
| CN106970956A (en) * | 2017-03-16 | 2017-07-21 | 天津大学 | A kind of method for searching three-dimension model based on tensor |
| CN108871284B (en) * | 2018-05-08 | 2020-06-23 | 中国矿业大学 | Non-initial value solving method of three-dimensional space similarity transformation model parameters based on linear feature constraint |
| CN110222580B (en) * | 2019-05-09 | 2021-10-22 | 中国科学院软件研究所 | Human hand three-dimensional attitude estimation method and device based on three-dimensional point cloud |
| CN111127505B (en) * | 2019-11-27 | 2024-03-26 | 天津津航技术物理研究所 | Online learning tracking and engineering realization method based on space planning |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101770566A (en) * | 2008-12-30 | 2010-07-07 | 复旦大学 | Quick three-dimensional human ear identification method |
| CN103810751A (en) * | 2014-01-29 | 2014-05-21 | 辽宁师范大学 | Three-dimensional auricle point cloud shape feature matching method based on IsoRank algorithm |
| CN105069403A (en) * | 2015-07-20 | 2015-11-18 | 同济大学 | Three-dimensional ear recognition based on block statistic features and dictionary learning sparse representation classification |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7605812B2 (en) * | 2006-12-19 | 2009-10-20 | Siemens Aktiengesellschaft | Intelligent modeling method and system for earmold shell and hearing aid design |
-
2015
- 2015-11-30 CN CN201510856611.0A patent/CN105654029B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101770566A (en) * | 2008-12-30 | 2010-07-07 | 复旦大学 | Quick three-dimensional human ear identification method |
| CN103810751A (en) * | 2014-01-29 | 2014-05-21 | 辽宁师范大学 | Three-dimensional auricle point cloud shape feature matching method based on IsoRank algorithm |
| CN105069403A (en) * | 2015-07-20 | 2015-11-18 | 同济大学 | Three-dimensional ear recognition based on block statistic features and dictionary learning sparse representation classification |
Non-Patent Citations (1)
| Title |
|---|
| 基于Sparse ICP的三维点云耳廓识别;王森 等;《图学学报》;20151231;第36卷(第6期);862-861页 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105654029A (en) | 2016-06-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105654029B (en) | The three-dimensional point cloud Ear recognition method of accuracy of identification and efficiency can be improved | |
| CN104268593B (en) | The face identification method of many rarefaction representations under a kind of Small Sample Size | |
| CN101398886B (en) | Rapid three-dimensional face identification method based on bi-eye passiveness stereo vision | |
| CN101833654B (en) | Sparse representation face identification method based on constrained sampling | |
| CN101609507A (en) | Gait recognition method | |
| Liu et al. | Ear-parotic face angle: A unique feature for 3D ear recognition | |
| CN103984922A (en) | Face identification method based on sparse representation and shape restriction | |
| CN102214299A (en) | Method for positioning facial features based on improved ASM (Active Shape Model) algorithm | |
| CN101770566A (en) | Quick three-dimensional human ear identification method | |
| Soltanpour et al. | Multimodal 2D–3D face recognition using local descriptors: pyramidal shape map and structural context | |
| Bai et al. | Person recognition using 3-D palmprint data based on full-field sinusoidal fringe projection | |
| Chen et al. | Unconstrained face verification using fisher vectors computed from frontalized faces | |
| KR20120089504A (en) | Apparatus for recognizing a subject and method using thereof | |
| CN110210293A (en) | A kind of gender identification method based on three-dimensional data and face-image | |
| CN106778491A (en) | The acquisition methods and equipment of face 3D characteristic informations | |
| Bastias et al. | A method for 3D iris reconstruction from multiple 2D near-infrared images | |
| CN105404883B (en) | A kind of heterogeneous three-dimensional face identification method | |
| CN110298275B (en) | Three-dimensional human ear identification method based on key points and local features | |
| CN111062343B (en) | Three-dimensional human ear recognition method based on average curvature and normal distribution transformation | |
| Lv et al. | 3D Face Recognition based on Local Conformal Parameterization and Iso‐Geodesic Stripes Analysis | |
| Xiao et al. | 3d face reconstruction via feature point depth estimation and shape deformation | |
| Vyas et al. | Efficient features for smartphone-based iris recognition | |
| Chauhan et al. | Study of ridge based and image based approach for fingerprint gender classification | |
| Zou et al. | 3-D face recognition based on warped example faces | |
| Chowdhury et al. | Fast and robust biometric authentication scheme using human ear |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190405 Termination date: 20191130 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |