CN104933448B - The curve matching algorithm unrelated with Location Scale in a kind of image recognition - Google Patents
The curve matching algorithm unrelated with Location Scale in a kind of image recognition Download PDFInfo
- Publication number
- CN104933448B CN104933448B CN201510410862.6A CN201510410862A CN104933448B CN 104933448 B CN104933448 B CN 104933448B CN 201510410862 A CN201510410862 A CN 201510410862A CN 104933448 B CN104933448 B CN 104933448B
- Authority
- CN
- China
- Prior art keywords
- curve
- point
- sub
- curved section
- shape indexes
- 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
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses curve matching algorithms unrelated with Location Scale in a kind of image recognition, including：（1）The Shape Indexes I of discretization is built to template curve；（2）On curve to be matched, all sub- curved sections are traversed, and use step（1）In identical Shape Indexes construction method, build Shape Indexes in identical point, Shape Indexes i built for every sub- curved sectionn；（3）Shape Indexes I and i are calculated successivelynSimilarity parameter；（4）If similarity parameter is less than or equal to the threshold value of setting, then it is assumed that corresponding sub- curved section is identical as template curve shape.The method of the present invention builds curve shape index by defining curve local coordinate system in this coordinate system so that matching process of the invention is not influenced by curvilinear translation, rotation parameter.
Description
Technical field
The present invention relates to a kind of image-recognizing method, more particularly to the curve unrelated with Location Scale in a kind of image recognition
Matching process.
Background technology
It is frequently necessary to carry out in the image recognition in the fields such as machine vision, reverse engineering, automation control, pattern-recognition
Curve to be matched and known template curve are carried out match cognization by the matching and identification of curve, so realize data splicing,
The purpose of target identification, classification；The challenges such as vision guided navigation, posture and three-dimension object image recognition, majority can convert
For the match cognization problem of several space curves.
Using computer disposal problems, space curve is all provided in a manner of equation or sorted points.Sorted points,
An as point set, putting in order for each point decides the spatial shape of curve in set, i.e., same point is with different order
Arrangement, constitutes different curve.The curve provided with equation form can easily be converted into the form of sorted points.
Currently, it is following several to realize that the main method of Curve Matching has：The closest approach iteration (ICP) of D space curve matching is calculated
Method, curve similar matrix matching algorithm.
Closest approach iteration (ICP) algorithm of D space curve matching has the advantages that monotone convergence, but to the selection of initial value ten
Divide sensitivity, and is easily absorbed in local extremum.
Curve similar matrix matching algorithm obtains similar matrix, examines that is, by calculating the local shape label of space curve
Optimum Matching section is measured, matching is completed.It such as uses calculus of finite differences to calculate curvature and torsion as local shape label, constitutes similar square
Battle array detects that longest matches sequence, and as Optimum Matching, but detection algorithm complexity is high, and it is ideal that algorithm is generally suitable only for matching
Curve.
In general, the space position parameter of curve to be matched（Translation, rotation）, scale parameter meeting and standard curve
There are larger differences, and above method is caused to be difficult to carry out, it is therefore desirable to a kind of unrelated to curvilinear translation rotation and scale low
Complexity Curve Matching method.
Invention content
It is an object of the invention to overcome the above-mentioned deficiency in the presence of the prior art, provide in a kind of image recognition with position
The unrelated Curve Matching method of scale is set, this method is by defining curve local coordinate system, and the curve defined in this coordinate system
Shape Indexes so that do not influenced by curvilinear translation, rotation parameter.
In order to achieve the above-mentioned object of the invention, the present invention provides following technical schemes：
The Curve Matching method unrelated with Location Scale, includes the following steps in a kind of image recognition：
（1）The Shape Indexes I of discretization is built to template curve；
（2）On curve to be matched, all sub- curved sections are traversed, and use step（1）In identical Shape Indexes structure
Method, Shape Indexes are built in identical point, and Shape Indexes i is built for every sub- curved sectionn；
（3）Shape Indexes I and i are calculated successivelynSimilarity parameter；
（4）If similarity parameter is less than or equal to the threshold value of setting, then it is assumed that corresponding sub- curved section and template curve shape
It is identical.
Preferably, the step（1）The Shape Indexes I of discretization is built to template curve, specially：
A. the sorted points of template curve are expressed as T={ tp1,tp2…tpn, wherein tpnN-th point is indicated in point range,
Its two dimensional form coordinate is [xtpn, ytpn], by point tp1、tp2Establish local coordinate system, as structure Shape Indexes coordinate system,
The x-axis unit direction vector Vxt of coordinate system is：Vxt=(tp2-tp1)/|tp2-tp1|, the y-axis unit direction vector of coordinate system
Vyt is that Vxt is rotated by 90 ° counterclockwise；
B. by tp1To tpnPoint coordinates is converted into the coordinate in the coordinate system that Vxt-Vyt is defined, and i-th point of coordinate is set as
[xti,yti], wherein tp1The coordinate of point is [0,0]；
C. the distance of adjacent point-to-point transmission in point range is calculated successively, and sets tpiAnd tpi+1Distance between point is li, calculated curve
Overall length L=l1+l2…+ln-1, wherein 1≤i≤n-1；
D. the corresponding normalization arc length parameters in every place, i.e. a are calculatedn=(l1+…+ln-1)/L, tp1Corresponding a at point1=
0；And tpnLocate corresponding an=1；
E. structure Shape Indexes I is [a1,a2,…an;xt1,xt2…xtn;yt1,yt2…ytn]。
Preferably, the step（2）In be every sub- curved section structure Shape Indexes in, specially：
A. the total length L ＇ of sub- curved section is calculated；
B. the arc length of specified location on sub- curved section, as L ＇ × [a are calculated1,a2,…an], wherein a1,a2,…anFor
The normalization arc length parameters；
C. on sub- curved section, obtain the space coordinate of each arc length corresponding points according to interpolation, and according to template curve
Identical method builds sub- curved section Shape Indexes in。
Preferably, the interpolation is specially and, n parts of the equal arc length of specified location, is then calculated on sub- curved section
Corresponding point coordinates at each arc length.
Preferably, the step（2）In sub- curved section traversal using following methods carry out：If curve to be matched is by { p1,
p2…pnSorted points definition, and stipulated that include at least N number of point in per strip curved section, 2≤N≤n, then all starting points be
p1Sub- curve be：{p1,p2…pN, { p1,p2…pN,pN+1, { p1,p2…pN,…pN+2... ..., { p1,p2…pN,…pn,
Wherein, all subscript values comprising N are smaller than equal to n；It is identical to be write out successively with p2、p3…、pn-N+1For the sub- curve of starting point
Section, this little curved section are to constitute all sub- curved sections that minimum length is N.
Preferably, the step（3）In Shape Indexes I and inSimilarity calculating method be：
A. poor M=I-i of 2 Shape Indexes is calculatedn；
B. the norm of M is calculated as similarity parameter, and wherein norm is a number more than or equal to 0, as Shape Indexes are complete
Complete the same, then norm is equal to 0, indicates equal, that is, has maximum similarity；Norm is bigger, then it represents that similarity is smaller.
Preferably, the threshold value value range is [0,1].
Compared with prior art, beneficial effects of the present invention：
1. by building curve local coordinate system, and building curve shape index in this coordinate system so that of the invention
Matching process is not influenced by curvilinear translation, rotation parameter；
2. defining curve using sorted points so that limited sub- curved section can be generated, so that the method for the present invention
Whole matching can not only be carried out, can also realize the local matching of curve；
3. using normalized arc length parameters structure Shape Indexes so that method of the invention can be to different zoom ruler
Curve under degree carries out shape matching；
4. the foundation using the norm of Shape Indexes difference as shape difference between appraisal curve so that method of the invention
Shape matching can be carried out to the curve comprising noise；By adjusting similarity parameter threshold value, you can determine the matching essence of curve
Degree；
5. the method for the present invention is applicable not only to two dimension, three-dimensional curve, can also expand to higher dimensional space curve；
6. the present invention method without using etc. arc length sorted points, therefore can be optimized according to curvature of curve situation from
The selection of scatterplot reduces the quantity at sorted points midpoint, and then reduce sub- curve in the case where ensureing discrete rear curve quality
The quantity of section, improves matching speed.
Specific implementation mode
With reference to test example and specific implementation mode, the present invention is described in further detail.But this should not be understood
It is only limitted to embodiment below for the range of the above-mentioned theme of the present invention, it is all that this is belonged to based on the technology that the content of present invention is realized
The range of invention.
In the method for the present invention, it is related to 2 class curves：Template curve, curve to be matched.Matching, is from curve to be matched
It is upper to find and orient and the immediate sub- curved section of template curve shape（Sub- curved section can be equal to curve to be matched）.It is all
The curve being related to, is all defined with sorted points.
The sorted points of template curve are expressed as T={ tp1,tp2…tpn, wherein tpnN-th point is indicated in point range, secondly
Dimension form coordinate is [xtpn,ytpn]；The sorted points of curve to be matched are expressed as C={ p1,p2…pn, wherein pnIt indicates in point range
N-th point, two dimensional form coordinate is [xn, yn]。
The method of the present invention mainly comprises the following steps：
（1）Build the template curve Shape Indexes I of discretization；
（2）On curve to be matched, all sub- curved sections are traversed, and use step（1）In identical Shape Indexes structure
Method, Shape Indexes are built in identical point, and Shape Indexes i is defined for every sub- curved sectionn；
（3）The similarity parameter of I and in is calculated successively；
（4）If similarity parameter is less than or equal to the threshold θ of setting, then it is assumed that corresponding sub- curved section and template curve shape
It is identical.
" identical " point refers to the point with same index.I.e. in two point ranges, the point with same sequence number,
This belongs in Curve Matching known.
Under two-dimensional case, to illustrate step（1）、（2）In template curve and the shape of the sub- curved section of curve to be matched refer to
Mark construction method.
1, template curve：
A. the sorted points of template curve are expressed as T={ tp1,tp2…tpn, wherein tpnN-th point is indicated in point range,
Its two dimensional form coordinate is [xtpn, ytpn], by point tp1、tp2Establish local coordinate system, as structure Shape Indexes coordinate system,
The x-axis unit direction vector Vxt of coordinate system is：Vxt=(tp2-tp1)/|tp2-tp1|, the y-axis unit direction vector of coordinate system
Vyt is that Vxt is rotated by 90 ° counterclockwise；
B. by tp1To tpnPoint coordinates is converted into the coordinate in the coordinate system that Vxt-Vyt is defined, and i-th point of coordinate is set as
[xti,yti], wherein tp1The coordinate of point is [0,0]；
C. the distance of adjacent point-to-point transmission in point range is calculated successively, and sets tpiAnd tpi+1Distance between point is li, calculated curve
Overall length L=l1+l2…+ln-1, wherein 1≤i≤n-1；
D. the corresponding normalization arc length parameters in every place, i.e. a are calculatedn=(l1+…+ln-1)/L, tp1Corresponding a at point1=
0；And tpnLocate corresponding an=1；
E. structure Shape Indexes I is [a1,a2,…an;xt1,xt2…xtn;yt1,yt2…ytn]。
2, the sub- curved section of curve to be matched：
A. by point p1、p2Local coordinate system is established, as the coordinate system for defining Shape Indexes.The x-axis unit side of coordinate system
It is calculated to vectorial Vx by following methods：
Vx=(p2-p1)/|p2-p1|
The y-axis unit direction vector Vy of coordinate system is that Vx is rotated by 90 ° counterclockwise；
B. by p1To pnPoint coordinates is converted into the coordinate in the coordinate system that Vx-Vy is defined, and i-th point of coordinate is set as [xi,
yi], particularly, p1The coordinate of point is [0,0]；
Step（2）Involved in identical point carry out Shape Indexes computational methods, compared with template curve method more than one
A interpolation procedure, as：
A. the total length L ＇ of sub- curved section is calculated；
B. the arc length of specified location on sub- curved section, as L ＇ × [a are calculated1,a2,…an], wherein a1,a2,…anFor
The normalized arc length parameters that above-mentioned template curve is calculated；
C. on sub- curved section, obtain the space coordinate of each arc length corresponding points according to interpolation, and according to template curve
Identical method builds Shape Indexes.Interpolation method ripe in the prior art, such as simplest one can be used in the interpolation
It plants and is：Adjacent 2 points given in point range are connected with straight line, are done if then being taken as needed on this straight line.
On curve, it is determined that after starting point, you can to uniquely determine a point with the arc length apart from starting point, normalization
Afterwards, between the arc length value of arbitrary point is 0 to 1 on curve.The method carries out curve when comparing, and needs on 2 curves that there are numbers
Measure identical point.Designated position refers to just determining a series of corresponding points on 2 curves.Interpolation can be used specifically sub bent
On line segment, n parts of the equal arc length of specified location, corresponding point coordinates at each arc length is then calculated.
Step（2）In sub- curved section ergodic algorithm using following methods carry out, if curve is by { p1,p2…pnOrderly point
Row definition, and stipulated that include at least N number of point in per strip curved section, 2≤N≤n, then all starting points are p1Sub- curved section
For：
{p1,p2…pN}
{p1,p2…pN,pN+1}
{p1,p2…pN,…pN+2}
……
{p1,p2…pN,…pn}
Wherein, all subscript values comprising N are smaller than equal to n.
Similarly, it can successively write out with p2、p3…pn-N+1Deng the sub-line section for starting point.These sub-line sections constitute minimum
" length " is all sub- curved sections of N.
Step（3）In Shape Indexes similarity I and i computational methods be（It must assure that 2 Shape Indexes are having the same
Normalize arc length parameters a1,a2,…an）：
A. poor M=I-i of 2 indexs is calculated；
B. the norm of M is calculated（Norm is a number more than or equal to 0）；If Shape Indexes are just the same, then norm is equal to
0, indicate that there is maximum similarity（It is i.e. equal）；Norm is bigger, then it represents that similarity is smaller.
The type of norm can arbitrarily be chosen, such as the norm of ∞, 1,2.Under normal circumstances, under the premise of ensureing effect, it is
Reduce calculation amount, square of 2 norms may be used, meaning is constant.
The selection of threshold θ will be according to particular problem.The dot it is necessary to setting is accurately identified if necessary, can generally be chosen
[0,1] is more than or equal to 0, the number being less than or equal between 1.Usually select minimum norm（It is most like）Result as final
Recognition result.
By the above content it is found that in the case where the operational speed of a computer is certain, the quantity of sub- curved section affects the calculation
The speed of service of method；And the main factor of determinant curve segment number, then it is the sorted points midpoint by defining the curve
Quantity determine, i.e., in the case where most short sub- curved section is certain, the point quantity in sorted points is fewer, it includes sub- curved section
Quantity is fewer.
The point that the method for the present invention is not required for defining curve such as is at the arc length distributions, therefore can be fixed by bent curvature of a curve
The justice arc length point range collection such as non-.In the smaller part of curvature, with big arc length discretization curve；In curvature major part, with small arc length
Discretization curve is to retain more detailed information.Reach in the case of ensureing Shape Indexes precision, reduces sorted points midpoint
Quantity, improve matching speed.
Claims (6)
1. a kind of Curve Matching method unrelated with Location Scale in image recognition, which is characterized in that include the following steps：
(1) the Shape Indexes I of discretization is built to template curve；
(2) on curve to be matched, all sub- curved sections are traversed, and use identical Shape Indexes construction method in step (1),
Shape Indexes are built in identical point, Shape Indexes i is built for every sub- curved sectionn；
(3) Shape Indexes I and i are calculated successivelynSimilarity parameter；
(4) if similarity parameter is less than or equal to the threshold value of setting, then it is assumed that corresponding sub- curved section is identical as template curve shape；
The step (1) builds template curve the Shape Indexes I of discretization, specially：
A. the sorted points of template curve are expressed as T={ tp1,tp2…tpn, wherein tpnN-th point is indicated in point range, two dimension
Form coordinate is [xtpn, ytpn], by point tp1、tp2Local coordinate system is established, as the coordinate system of structure Shape Indexes, coordinate system
X-axis unit direction vector Vxt be：Vxt=(tp2-tp1)/|tp2-tp1|, the y-axis unit direction vector Vyt of coordinate system is
Vxt is rotated by 90 ° counterclockwise；
B. by tp1To tpnPoint coordinates is converted into the coordinate in the coordinate system that Vxt-Vyt is defined, and i-th point of coordinate is set as [xti,
yti], wherein tp1The coordinate of point is [0,0]；
C. the distance of adjacent point-to-point transmission in point range is calculated successively, and sets tpiAnd tpi+1Distance between point is li, calculated curve overall length L
=l1+l2…+ln-1, wherein 1≤i≤n-1；
D. the corresponding normalization arc length parameters in every place, i.e. a are calculatedn=(l1+…+ln-1)/L, tp1Corresponding a at point1=0；And
tpnLocate corresponding an=1；
E. structure Shape Indexes I is [a1,a2,…an；xt1,xt2…xtn；yt1,yt2…ytn]。
2. the Curve Matching method unrelated with Location Scale in image recognition according to claim 1, which is characterized in that institute
It is every sub- curved section structure Shape Indexes i to state in step (2)n, specially：
A. the total length L ＇ of sub- curved section is calculated；
B. the arc length of specified location on sub- curved section, as L ＇ × [a are calculated1,a2,…an], wherein a1,a2,…anIt is described
Normalize arc length parameters；
C. on sub- curved section, the space coordinate of each arc length corresponding points is obtained according to interpolation, and according to identical as template curve
Method build sub- curved section Shape Indexes in。
3. the Curve Matching method unrelated with Location Scale in image recognition according to claim 2, which is characterized in that institute
It is specially, n parts of the equal arc length of specified location, then to calculate corresponding point at each arc length on sub- curved section to state interpolation
Coordinate.
4. the Curve Matching method unrelated with Location Scale in image recognition according to claim 1, which is characterized in that institute
The sub- curved section traversal stated in step (2) is carried out using following methods：If curve to be matched is by { p1,p2…pnSorted points are fixed
Justice, and stipulated that including at least N number of point in per strip curved section, 2≤N≤n, then all starting points are p1Sub- curve be：{p1,
p2…pN, { p1,p2…pN,pN+1, { p1,p2…pN,…pN+2... ..., { p1,p2…pN,…pn, wherein all includes N's
Subscript value is smaller than equal to n；It is identical to be write out successively with p2、p3…、pn-N+1For the sub- curved section of starting point, this little curved section is
Constitute all sub- curved sections that minimum length is N.
5. the Curve Matching method unrelated with Location Scale in image recognition according to claim 2, which is characterized in that institute
State the Shape Indexes I and i in step (3)nSimilarity calculating method be：
A. the poor M=I-i of 2 Shape Indexes is calculatedn；
B. the norm of M is calculated as similarity parameter, and wherein norm is a number more than or equal to 0, such as Shape Indexes complete one
Sample, then for norm equal to 0, expression is equal, that is, has maximum similarity；Norm is bigger, then it represents that similarity is smaller.
6. the Curve Matching method unrelated with Location Scale in image recognition according to claim 5, which is characterized in that institute
It is [0,1] to state threshold value value range.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510410862.6A CN104933448B (en) | 2015-07-14 | 2015-07-14 | The curve matching algorithm unrelated with Location Scale in a kind of image recognition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510410862.6A CN104933448B (en) | 2015-07-14 | 2015-07-14 | The curve matching algorithm unrelated with Location Scale in a kind of image recognition |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104933448A CN104933448A (en) | 2015-09-23 |
| CN104933448B true CN104933448B (en) | 2018-09-18 |
Family
ID=54120608
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510410862.6A Expired - Fee Related CN104933448B (en) | 2015-07-14 | 2015-07-14 | The curve matching algorithm unrelated with Location Scale in a kind of image recognition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104933448B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106054593B (en) * | 2016-05-27 | 2019-05-31 | 浙江工业大学 | Variable precision optimization matching method of NURBS curve based on Kabsch algorithm |
| CN107862748A (en) * | 2017-11-10 | 2018-03-30 | 佛山科学技术学院 | A kind of unrelated thermonuclear characteristic processing method of improved yardstick |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AU2007254666A1 (en) * | 2007-12-24 | 2009-07-09 | Canon Kabushiki Kaisha | Recognition of overlapping shapes from document images |
| CN101794379A (en) * | 2010-02-04 | 2010-08-04 | 南昌航空大学 | Method for recognizing partially shielded cambered object |
-
2015
- 2015-07-14 CN CN201510410862.6A patent/CN104933448B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AU2007254666A1 (en) * | 2007-12-24 | 2009-07-09 | Canon Kabushiki Kaisha | Recognition of overlapping shapes from document images |
| CN101794379A (en) * | 2010-02-04 | 2010-08-04 | 南昌航空大学 | Method for recognizing partially shielded cambered object |
Non-Patent Citations (1)
| Title |
|---|
| "基于轮廓的形状匹配方法研究";余江;《中国优秀硕士学位论文全文数据库 信息科技辑》;20131115(第11期);第I138-904页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104933448A (en) | 2015-09-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101315698B (en) | Characteristic matching method based on straight line characteristic image registration | |
| CN111080684B (en) | Point cloud registration method for point neighborhood scale difference description | |
| CN101980250B (en) | Method for identifying target based on dimension reduction local feature descriptor and hidden conditional random field | |
| CN106355577B (en) | Rapid image matching method and system based on significant condition and global coherency | |
| CN107424161B (en) | Coarse-to-fine indoor scene image layout estimation method | |
| CN105046694B (en) | A kind of point cloud rapid registering method based on surface fitting coefficient characteristics | |
| CN102880737B (en) | Based on the workpiece method for registering in flexible assembly and system | |
| CN110866934B (en) | Normative coding-based complex point cloud segmentation method and system | |
| CN111415379B (en) | Three-dimensional point cloud data registration method based on cuckoo optimization | |
| CN110009680B (en) | Monocular image position and posture measuring method based on circle feature and different-surface feature points | |
| CN105976397B (en) | A kind of method for tracking target | |
| CN111145228A (en) | Heterogeneous image registration method based on local contour point and shape feature fusion | |
| CN111815686A (en) | Coarse-to-fine point cloud registration method based on geometric features | |
| Xu et al. | A robust fingerprint matching approach: Growing and fusing of local structures | |
| CN112364881B (en) | Advanced sampling consistency image matching method | |
| CN109766903B (en) | Point cloud model curved surface matching method based on curved surface features | |
| CN104933448B (en) | The curve matching algorithm unrelated with Location Scale in a kind of image recognition | |
| CN111652070A (en) | Face sequence collaborative recognition method based on surveillance video | |
| Ouyang et al. | Fingerprint pose estimation based on faster R-CNN | |
| CN112652003A (en) | Three-dimensional point cloud registration method based on RANSAC measure optimization | |
| Bai et al. | Graph matching using spectral embedding and alignment | |
| CN108122202A (en) | A kind of map Linear element multi-scale information derived method based on Shape context | |
| CN114463425A (en) | Workpiece surface featureless point positioning method based on probability Hough linear detection | |
| CN104268502A (en) | Recognition method for human body after vein image feature extraction | |
| CN104298980B (en) | A kind of finger print matching method and device based on smart card |
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: 20180918 Termination date: 20210714 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |