CN103559489B - Palm feature extracting method under a kind of noncontact imaging mode - Google Patents
Palm feature extracting method under a kind of noncontact imaging mode Download PDFInfo
- Publication number
- CN103559489B CN103559489B CN201310589834.6A CN201310589834A CN103559489B CN 103559489 B CN103559489 B CN 103559489B CN 201310589834 A CN201310589834 A CN 201310589834A CN 103559489 B CN103559489 B CN 103559489B
- Authority
- CN
- China
- Prior art keywords
- palm
- point
- centre
- inscribed circle
- image
- 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
Landscapes
- Collating Specific Patterns (AREA)
- Image Analysis (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The present invention relates to palm feature extracting method under a kind of noncontact imaging mode, the palm inscribed circle for referring to root is extracted first, some radial line segments, which are done, from incenter intersects at circumference, simultaneously, construct the Grad of each point in formwork calculation inscribed circle, barycenter relative radius is calculated according to the Grad of each point on each line segment, characteristic parameter is defined as, with this constitutive characteristic vector space.The feature that the present invention is constructed has translation, rotation and proportional zoom consistency, meanwhile, the advantages of feature extraction time is very fast.
Description
Technical field:
The invention belongs to biometrics identification technology field, it is related to palm feature extraction side under a kind of noncontact imaging mode
Method.
Background technology:
With the development and the progress of society in epoch, gate control system has increasing need for efficient, reliable ID authentication mechanism
To determine legitimacy that personnel pass in and out to specific region, biometrics identification technology is that a kind of physiology intrinsic according to human body is special
Levy or behavioural characteristic recognizes the technology of identity.Because biological characteristic has the spy of " everybody possesses, people is variant, remain unchanged for a long period of time "
Point, and will not pass into silence or lose, more and more applied in gate control system.It can be seen from existing pertinent literature,
Being presently available for the biological characteristic of gate control system includes fingerprint, palmmprint, vein, face, iris, retina, human ear etc..Hand is special
Levy because more facilitating when being easily accepted by a user high degree and collection image, obtained among numerous biological characteristics widely
Using.
Feature of the human hand usually in state of partly clenching fist, palm is more difficult with respect to finger characteristic and the back of the hand feature stolen
Take, with higher security.Palm vein is the live biometric for being hidden in inside of human body, and the external texture with respect to hand is special
Levy, vein pattern is less prone to be stolen and replicated.But because the physiological structure of palm itself, some palms are under near infrared light
The image medium sized vein information of collection is almost estimated less than the texture in palm is mainly made up of palmmprint main line and some mastoid process lines.
Therefore, the palm of people is shot under near infrared light using single collecting device herein, by obtained hand image
The palm vein information and palm palmmprint dominant line information that palm area is included are defined as palm feature, use it for identity knowledge
Not, as shown in Figure 1.
The initial concept of palm vein identification is appeared in the 1990s, because Fujitsu's palm vein is known
The popularization of other instrument and be widely studied since 2006.At 2006 to 2010, hand collecting device was all based on connecing
Touch design.Contact acquisition mode requires that user's human hand in collection image process comes in contact with collecting device, or
The fixation held some external equipments of collecting device or hand is placed on into fixed putting position is fixed.Contact is gathered in health
Some problems can be brought with terms of security, meanwhile, the sensor surface of contact Acquisition Instrument is easier to be contaminated, especially door
Access control system is frequently used in the poor outdoor environment of sanitary condition, and this can cause the rate of refusing by mistake of system to rise, while also shortening
The service life of acquisition instrument.
Started the research for the palm characteristic recognition method for noncontact imaging mode occur in 2010.Pass through noncontact
Mode realizes that palm imaging can make hand identifier be got the nod in the crowd for worrying disease propagation, but hand and imaging device it
Between the uncertain of position cause hand to there is translation, rotation and proportional zoom in the picture, to extract stable hand feature extraction
Higher requirement.
The content of the invention:
Goal of the invention:
Set about palm feature extracting method in image the present invention relates to a kind of non-contact capture mode, its main purpose is
The accuracy that non-contact capture mode sets about to recognize is improved, is solved caused by the change of position between hand and imaging device and angle
Hand image in hand translate, rotation and proportional zoom problem.
Technical scheme:
The present invention is achieved through the following technical solutions:
Palm feature extracting method under a kind of noncontact imaging mode, it is characterised in that:This method step is as follows:
(1)The selection of centre of the palm stable reference point:A centre of the palm inscribed circle is constructed to select the reference point that the centre of the palm is stable;Design
Inscribed circle possesses following condition:It is tangent with palm both sides contour line respectively, and pass through middle finger, nameless finger root point;Divide successively
The volar edge point on palm contour line and on the outside of pinkie on palm contour line is not pointed on the outside of forefinger to be scanned, it is determined that
The position of incenter and the radius of inscribed circle;
(2)The extraction of palm invariant features:Choose two kinds of palmar hand features of palm vein and palm palmmprint and carry out identity knowledge
The realization of other system;Centre of the palm stable reference point is obtained by incenter, opponent's image is positioned, eliminate collection image mistake
The influence that palm is translated in journey;The point centered on the reference point of the centre of the palm, with angle, θ0Some rays are projected for unit, in centre of the palm ginseng
Some radial line segments, referred to as Eigenvector are constructed between examination point and inscribed circle contour line;Finally, according to each on each Eigenvector
The gradient intensity value of point calculates the barycenter relative radius of line segment, constructs characteristic vector.
Above-mentioned steps(2)It is middle to construct comprising the following steps that for characteristic vector:
Step 1:The generation of gradient image
Gradient intensity value is selected as the calculation basis of feature, by hand greyscale image transitions into gradient map before feature extraction
Picture:
1)Intersecting direction with palm vein or palm palmmprint main line, meeting gray scale minimum condition, therefore, pass through four
Individual direction (00,450,900,1350) judgement of gray scale minimum, palm vein or palm palmmprint candidate pixel point can be found;
And the pixel for being unsatisfactory for gray scale minimum is not necessarily vein or palmmprint, therefore, by the pixel Grad zero setting;
2)In the pixel for meeting gray scale minimum, pixel is located within vein or palmmprint, in its vertical direction
Two Grad all will be than larger;When pixel deviates vein or palmmprint, Grad will appear from small one and large one phenomenon, because
This, takes the smaller value in two Grad for meeting gray scale minimum direction as the Grad of the direction, then takes and meet gray scale
Maximum in all direction gradient values of minimum as the pixel Grad, the Grad can reflect along vein or
Palmmprint vertical gradient variation tendency, i.e. maximum of gradients correspond to the center of vein or palmmprint, with off-center
Point, gradient is gradually reduced;
Step 2:The extraction of Eigenvector center of mass point relative radius
From the zero degree direction of centre of the palm reference point, inscribed circle is divided into H sector region in the counterclockwise direction, thus
Obtain H bar line segments K1, K2... ..., KH;Eigenvector barycenter and the distance of centre of the palm reference point, i.e. barycenter radius, are represented, jth with r
The barycenter radius of bar Eigenvector is:
Wherein, Eigenvector is inscribed circle radius, for convenience of calculating, by inscribed circle radius round numbers, i.e. formula (1)
I from 1 to M, M be the inscribed circle pixel number that radius is included in the horizontal direction;
In addition to both horizontally and vertically, the unit point i on Eigenvector will not typically fall on pixel, its gradient
Value is calculated using interpolation method, using bilinear interpolation method, is counted using the Grad of 4 nearest neighbor pixels of unit point
Calculate the Grad of unit point;
In order to solve the problems, such as proportional zoom, to the relative radius δ of center of mass point and incenterjCalculated, calculate public
Formula is as follows:
Become using the translation of relative radius construction, rotation and the proportional zoom of H bar Eigenvector center of mass point and incenter
The invariant features changed, characteristic vector is:
F={ δ1,δ2,…,δH};
Step 3:The matching of characteristic vector
By above-mentioned steps, in palm construct H bars can reflect the line segment of palm textural characteristics, if from registration hand image
The characteristic vector of middle extraction is:
Fa={ δ1a,δ2a,…,δHa,
The characteristic vector extracted from hand image to be identified is:
Fb={ δ1b,δ2b,…,δHb,
Judge whether registration hand image matches with hand image to be identified, how many matching line segment pair between the two first judged,
Line segment is in matching, calculating the characteristic vector between Eigenvector apart from djIt is used as correspondence line segment barycenter relative radius in 2 width images
Similarity criteria, registration hand image and hand image middle conductor j to be identified characteristic vector it is as follows apart from calculation formula:
dj=| δja-δjb| (3)
Work as djLess than given threshold T1When, then it is assumed that two correspondence line segments are matching line segment, if matching line segment is to there is g,
Then matching rate is:
When matching rate G is more than given threshold T2When, then it is assumed that two width hand image sources are in same person.
Advantage and effect:
The feature that the present invention is constructed has translation, rotation and proportional zoom consistency, meanwhile, the feature extraction side provided
The method speed of service is very fast, is more suitable for the noncontact hand identifying device higher to requirement of real-time.
Brief description of the drawings:
Fig. 1 is the palm picture that collects under cordless;
Fig. 2 is palm inscribed circle detects schematic diagram;
Fig. 3 is palm feature extracting method flow chart under noncontact imaging mode;
Fig. 4 is gradient value schematic diagram;
Fig. 5 is gradient calculation template, Fig. 5(a)For gradient calculation template schematic diagram, Fig. 5(b)Subtemplate center when for l being 2
Point position view, Fig. 5(c)Subtemplate center position schematic diagram when for l being 3;
Fig. 6 is the palm invariant features structural map based on vena metacarpea and palmmprint specified point position, Fig. 6(a)For in inscribed circle
Construct Eigenvector figure, Fig. 6(b)It is characterized line segment schematic diagram, Fig. 6(c)Pixel gray value schematic diagram on line segment is characterized, is schemed
6(d)It is characterized line segment center of mass point schematic diagram.
Embodiment:
The present invention is described further with specific embodiment below in conjunction with the accompanying drawings:
The present invention is palm feature extracting method under a kind of noncontact imaging mode, it is characterised in that:Using noncontact into
When image space formula gathers hand image, palmistry is uncertain to the direction, position and shooting distance of imaging device, and this causes for same
Size, direction and the position of hand may be differed in one people, the hand image gathered every time.This method of the present invention is directly extracted
With translation, rotation and the unrelated feature of proportional zoom, it is to avoid original to that may change during being normalized in image
The some properties of image and add run time.This method step is as follows:
(1)The selection of centre of the palm stable reference point:In order to join using each point on palm vein and palmmprint main line and the centre of the palm are stable
The relative space position of examination point constructs the feature unrelated with palm rotation, translation and proportional zoom, by constructing in a centre of the palm
The circle of contact selects the reference point that the centre of the palm is stable;In order to ensure that inscribed circle has uniqueness, design inscribed circle possesses following condition:Point
It is not tangent with palm both sides contour line, and pass through middle finger, nameless finger root point;On round geometric properties, round edge circle
The perpendicular bisector of any two points must by the center of circle, and circle tangent line must perpendicular to the radius by the point of contact, according to this two properties,
The volar edge point on palm contour line and on the outside of pinkie on palm contour line is pointed on the outside of forefinger respectively successively to be swept
Retouch, determine the position of incenter and the radius of inscribed circle;Step is as follows:
1)Palm contour line is extracted using the hand contour tracing method based on direction gradient extreme value.
2)Extract middle finger, the third finger and refer to root point P1, concrete methods of realizing is as follows:For left hand imaging, its thumb is in image
In upward, the third finger is located under middle finger, from middle fingertip, along step 1) contour line that extracts finds the change of first curvature
Change big point, the point is middle finger and nameless finger tip intersection C, then in corresponding gray level image, from C points, use
Gray scale minimum tracking finds middle finger and nameless webs line;Any pixel point being set on webs line is P11, find
Its two point P on vertical webs direction12And P13, this 2 points and point P11Distance for 3 pixels it is wide;Point P is calculated respectively11
With point P12, point P13Grad, regard the average value AVG of two Grad as evaluation point P11The gradient on vertical webs direction
Variable quantity;Using C points as starting point, each pixel being pointed on webs line calculates its graded amount AVG successively, works as AVG
When being obviously reduced, tracking stops, and former point is middle finger, nameless finger root point P1, and the point is marked at into handwheel profile
In bianry image.
3)Refer to root inscribed circle to extract, it is as follows that palm inscribed circle automatic testing method implements step:A) inscribe is determined
Circle and the scope at palm contour line point of contact, regard the marginal point in the range of this as sweep object.First, middle finger, the third finger are passed through
Finger root point P1It is straight line L3, webs fitting a straight line of the straight line between middle finger and the third finger.Straight line L3Intersect at palm wheel
Profile, it is P to remove first point that side starts2, second point is P3.Secondly, from point P3Start to do a series of to palm wrist direction
Parallel lines, parallel lines are parallel to line L3, every parallel lines intersect at contour line, remove second point that side starts, and constitute point system
Row, when the ordinate of point jumps, it is P to take the point4.Finally, two point set A are constructed1And A2.First point set A1
By from point P3To point P4Between edge line on point constitute;Second point set A2By wrist marginal point Q to point P2Between edge line
On point constitute.Point set A1And A2The as scope at inscribed circle and palm contour line point of contact, using the marginal point in the range of this as
Sweep object, performs step b) and step c) respectively.B) incenter candidate point region is extracted.First, set is chosen
A1Or set A2In any point, be set to point P5.Pass through middle finger, nameless finger root point P1With marginal point P5Line is done, if line
Section is L2, to line segment L2It is perpendicular bisector L '2;Secondly, along palm contour line in point P5Left and right respectively takes at 2 points, is fitted by these points
Straight line L1, cross P5Point is straight line L1Vertical line L '1.Straight line L '1With straight line L '2Intersect at point O1.To set A1And A2Interior each point minute
Do not make aforesaid operations, extract its joining O1, respectively obtain point set A '1And A '2.Set of computations A '1In point and A '2Midpoint
Euclidean distance, the point that distance is less than Δ A is retained, set A ' is constituted(The present invention takes Δ A=6), in set A ' each point it is horizontal,
Ordinate is compared, and obtains minX, minY, maxX, maxY, constitutes each pixel in a rectangular area, the rectangular area
For the candidate point of incenter.C) incenter and inscribed circle radius are determined.Center of circle candidate point region is calculated first
The distance of each interior center of circle candidate point and palm contour line on the outside of palm contour line on the outside of forefinger and pinkie, computational methods are such as
Under:Any center of circle candidate point in the candidate point region of the center of circle is chosen, O ' is set to, center of circle candidate point O ' and set are calculated respectively
A1、A2The Euclidean distance of middle each point, takes minimum value, is set to d1And d2, apart from d1And d2As pixel O ' arrives both sides palm profile
The distance of line.Then calculate candidate point O ' and refer to root point P with middle finger, the third finger1Distance be d3.Finally calculate apart from d1、d2And d3
Standard deviation, come evaluate three distance difference degree.Aforesaid operations are done respectively to each center of circle candidate point, in obtained mark
Minimum is selected in quasi- difference series, corresponding candidate point is incenter O (x0,y0), the point and middle finger, nameless and hand
Slap interfaces point P1Distance be inscribed circle radius R.D) with point O (x0,y0) it is the center of circle, R is radius, draws palm inscribed circle,
Reject circle external information, as palm vein feature extraction region.
(2)The extraction of palm invariant features:In the hand image gathered under near infrared light, palm vein is presented netted, with
Palm palmmprint main line shows native texture state together, therefore a width hand image can regard a width texture image, palm as
Vein is the important component of palm textural characteristics.But because the physiological structure of palm itself, some palms are near infrared light
The image medium sized vein information of lower collection is almost estimated less than the texture in palm is main by palmmprint main line and some mastoid process line groups
Into.Palmmprint main line and mastoid process line not only have very strong directionality, and thickness is deep mixed, therefore, and the present invention chooses palm
Two kinds of palmar hand features of vein and palm palmmprint carry out the realization of identification system;The centre of the palm is obtained by incenter stable
Reference point, opponent's image is positioned, and eliminates the influence that palm is translated in collection image process;Centered on the reference point of the centre of the palm
Point, with angle, θ0Project some rays for unit, constructed between centre of the palm reference point and inscribed circle contour line some it is radial
Line segment(Line segment length is equal to inscribed circle radius length), referred to as Eigenvector;Finally, according to the gradient of each point on each Eigenvector
Intensity level calculates the barycenter relative radius of line segment(That is the ratio between absolute distance of barycenter and incenter and inscribed circle radius), structure
Make characteristic vector.
Above-mentioned steps(2)It is middle to construct comprising the following steps that for characteristic vector:
Step 1:The generation of gradient image
Palm vein and the extraction not a duck soup of palmmprint locus, also can bring larger even if less mistake to identification
Error.Therefore, present invention selection gradient intensity value turns hand gray level image before feature extraction as the calculation basis of feature
Change gradient image into, transfer principle is as follows:
1)Intersecting direction with palm vein or palm palmmprint main line, meeting gray scale minimum condition, therefore, pass through four
Individual direction (00,450,900,1350) judgement of gray scale minimum, palm vein or palm palmmprint candidate pixel point can be found;
And the pixel for being unsatisfactory for gray scale minimum is not necessarily vein or palmmprint, therefore, by the pixel Grad zero setting;
2)In the pixel for meeting gray scale minimum, pixel is located within vein or palmmprint, in its vertical direction
Two Grad all will be than larger;When pixel deviates vein or palmmprint, Grad will appear from small one and large one phenomenon, because
This, takes the smaller value in two Grad for meeting gray scale minimum direction as the Grad of the direction, then takes and meet gray scale
Maximum in all direction gradient values of minimum as the pixel Grad, the Grad can reflect along vein or
Palmmprint vertical gradient variation tendency, i.e. maximum of gradients correspond to the center of vein or palmmprint, with off-center
Point, gradient is gradually reduced;
Step 2:The extraction of Eigenvector center of mass point relative radius
From the zero degree direction of centre of the palm reference point, inscribed circle is divided into H sector region in the counterclockwise direction, thus
Obtain H bar line segments K1, K2... ..., KH;Eigenvector barycenter and the distance of centre of the palm reference point, i.e. barycenter radius, are represented, jth with r
The barycenter radius of bar Eigenvector is:
Wherein, Eigenvector is inscribed circle radius, for convenience of calculating, by inscribed circle radius round numbers, i.e. formula (1)
I from 1 to M, M be the inscribed circle pixel number that radius is included in the horizontal direction;
In addition to both horizontally and vertically, the unit point i on Eigenvector will not typically fall on pixel, its gradient
Value is calculated using interpolation method, and the present invention uses bilinear interpolation method, utilizes the gradient of 4 nearest neighbor pixels of unit point
Value carrys out the Grad of unit of account point;
Because hand imaging is not of uniform size, the inscribed circle radius calculated is not unique, and center of mass point is whipped with respect to the distance in the center of circle
It is imaged size and changes, in order to solve the problems, such as proportional zoom, to the relative radius δ of center of mass point and incenterj(That is center of mass point
With the barycenter radius r of incenterjThe ratio between with inscribed circle radius R)Calculated, calculation formula is as follows:
Become using the translation of relative radius construction, rotation and the proportional zoom of H bar Eigenvector center of mass point and incenter
The invariant features changed, characteristic vector is:
F={ δ1,δ2,…,δH};
Step 3:The matching of characteristic vector
By above-mentioned steps, H bars can be constructed in palm can reflect the line segment of palm textural characteristics, if from registration hand
The characteristic vector extracted in image is:
Fa={ δ1a,δ2a,…,δHa,
The characteristic vector extracted from hand image to be identified is:
Fb={ δ1b,δ2b,…,δHb,
How many judge when whether registration hand image matches with hand image to be identified, it is necessary to which matching line segment between the two first judged
It is right, in line segment in matching, calculating the characteristic vector between Eigenvector apart from djIt is relative as correspondence line segment barycenter in 2 width images
The similarity criteria of radius, registration hand image and hand image middle conductor j to be identified characteristic vector are as follows apart from calculation formula:
dj=| δja-δjb| (3)
Work as djLess than given threshold T1When, then it is assumed that two correspondence line segments are matching line segment, if matching line segment is to there is g,
Then matching rate is:
When matching rate G is more than given threshold T2When, then it is assumed that two width hand image sources are in same person.
The present invention relates to palm feature extracting method under a kind of noncontact imaging mode, palm localization method in opponent's image
And palm feature extracting method is studied, its main purpose be improve non-contact capture mode set about identification it is accurate
Property, solve, because hand is translated in the hand image caused by the change of position between hand and imaging device and angle, to rotate and proportional zoom
Problem.
The hand image that gathers in a non contact fashion of the present invention is when being matched, it is necessary to hand is normalized, so that hand
It is unrelated that palm feature is imaged translation, rotation and the proportional zoom caused with hand.Because normalization process may change original image
Some properties, and change this some properties may effect characteristicses extraction, meanwhile, normalization process can also increase operation
Time.Therefore, present invention selection directly constructs a kind of palm feature unrelated with translation, rotation and proportional zoom.
Specific steps are described as follows with reference to accompanying drawing:
(1)Fig. 1 is the palm image gathered under near infrared light by cordless.By to substantial amounts of near infrared light
The palm image of lower collection is observed, and palmmprint dominant line information can be clearly seen in palm image, can be utilized completely
Same collecting device is extracted to palm vein and palmmprint main line.From the vena metacarpea feature of same width hand image zooming-out human hand and
Palmmprint major line features, can avoid the problem of some multi-modal biological characteristics are merged, and such as system complexity, cost, user receive
Degree, data base administration etc..
(2)Fig. 2 is palm inscribed circle detects schematic diagram.For contactless hand identifying system, the people collected
Hand image suffers from the influence of the factors such as distance of camera lens, hands movement, misoperation, causes there is very big difference between image
It is different.In order to eliminate influence of the image difference to recognition effect, suitable datum mark is extracted from palm, new reference is set up and sits
Mark system, opponent's image is positioned, to reduce the shadow of the factors such as the rotation introduced in image acquisition process, translation, proportional zoom
Ring, improve the robustness of match cognization algorithm.
(3)Fig. 3 is palm feature extracting method flow chart under noncontact imaging mode, and this method proposes that one kind can reflect
Palm arteries and veins and palm print characteristics and the palm feature not influenceed by rotation and proportional zoom.The palm inscribed circle for referring to root is extracted first,
Some radial line segments, which are done, from incenter intersects at circumference, meanwhile, the gradient of each point in construction formwork calculation inscribed circle
Value, according on each line segment each point Grad calculate barycenter relative radius, be defined as characteristic parameter, with this constitutive characteristic to
Quantity space.
(4)Fig. 4 is gradient value schematic diagram.In palm image, intersect with palm vein or palm palmmprint main line
Direction, meets gray scale minimum condition, therefore, passes through four direction (00,450,900,1350) judgement of gray scale minimum, can
Find palm vein or palm palmmprint candidate pixel point.And the pixel for being unsatisfactory for gray scale minimum be not necessarily vein or
Palmmprint, therefore, by the pixel Grad zero setting.In the pixel for meeting gray scale minimum, pixel be located at vein or
, all will be than larger in two Grad of its vertical direction within palmmprint;When pixel deviates vein or palmmprint, Grad
Small one and large one phenomenon is will appear from, therefore, the smaller value in two Grad for meeting gray scale minimum direction is taken as the direction
Grad, then take the maximum in all direction gradient values for meeting gray scale minimum as the Grad of the pixel, should
Grad, which can reflect along vein or palmmprint vertical gradient variation tendency, i.e. maximum of gradients, corresponds to vein or palmmprint
Center, with off-center point, gradient is gradually reduced.
(5)Fig. 5 is gradient calculation template.Can be by hand greyscale image transitions into gradient map using the gradient template of construction
Picture.
(6)Fig. 6 is the schematic diagram constructed based on palmmprint and the palm invariant features of vena metacarpea specified point position.Palm feature
The basic thought of construction is as follows:Some radial line segments, every line segment are constructed between centre of the palm reference point and inscribed circle contour line
It may pass through some palm veins or palmmprint streakline.When removing to cover the hand image in inscribed circle with enough line segments, and
Palmmprint and vena metacarpea in the hand image in inscribed circle can be just depicted with these line segments with reference to corresponding grey scale change information
Distributed architecture.
It can be seen from the constant principle of moment of single order, Eigenvector each point Grad multiplies with the relative centre of the palm reference point distance of point
Product sum is equal to the product of this feature line segment each point Grad sum centre of the palm reference point distance relative with center of mass point, thus, it is possible to
Distance of the Eigenvector center of mass point with respect to centre of the palm reference point is obtained using the Grad of each point on Eigenvector, matter is defined as
The relative radius of heart point.
In hand image, at relatively non-streakline there is bigger gray scale to be mutated at palm vein and palmmprint main line, that is, have
Higher Grad, the influence to line segment centroid position plays a decisive role.It therefore, it can utilize the barycenter of Eigenvector relative half
The grey scale change of each point on Expressive Features line segment is carried out in footpath, and then embodies point of vena metacarpea and palmmprint in the hand image in inscribed circle
Cloth structure.
Select line segment barycenter and the relative radius of incenter as palm feature, this feature ensures that hand image scaled contracts
Put consistency.Selection centre of the palm reference point refers to the line direction of root point for zero degree direction to middle finger, the third finger, as first
The direction of Eigenvector.Due to the point position conveniently position, direction and size in the picture and change, therefore, first
The direction of bar Eigenvector has uniqueness.Analyzed more than, every Eigenvector barycenter relative radius can reflect palm
Vein and palmmprint have translation, rotation and proportional zoom consistency in the position distribution in hand space in this, as palm feature.
Claims (1)
1. palm feature extracting method under a kind of noncontact imaging mode, it is characterised in that:This method step is as follows:
(1)The selection of centre of the palm stable reference point:A centre of the palm inscribed circle is constructed to select the reference point that the centre of the palm is stable;Design inscribe
Circle possesses following condition:It is tangent with palm both sides contour line respectively, and pass through middle finger, nameless finger root point;It is right respectively successively
It is scanned positioned at the volar edge point on the outside of forefinger on palm contour line and on the outside of pinkie on palm contour line, determines inscribe
The position of the round heart and the radius of inscribed circle;
(2)The extraction of palm invariant features:Choose two kinds of palmar hand features of palm vein and palm palmmprint and carry out identification system
The realization of system;Centre of the palm stable reference point is obtained by incenter, opponent's image is positioned, eliminated in collection image process
The influence of palm translation;The point centered on the reference point of the centre of the palm, projects some rays in units of angle, in the centre of the palm reference point with
Some radial line segments, referred to as Eigenvector are constructed between inscribed circle contour line;Finally, according to the ladder of each point on each Eigenvector
The barycenter relative radius that intensity level calculates line segment is spent, characteristic vector is constructed.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310589834.6A CN103559489B (en) | 2013-11-19 | 2013-11-19 | Palm feature extracting method under a kind of noncontact imaging mode |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310589834.6A CN103559489B (en) | 2013-11-19 | 2013-11-19 | Palm feature extracting method under a kind of noncontact imaging mode |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103559489A CN103559489A (en) | 2014-02-05 |
CN103559489B true CN103559489B (en) | 2017-11-07 |
Family
ID=50013730
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310589834.6A Expired - Fee Related CN103559489B (en) | 2013-11-19 | 2013-11-19 | Palm feature extracting method under a kind of noncontact imaging mode |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103559489B (en) |
Families Citing this family (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103870831A (en) * | 2014-03-13 | 2014-06-18 | 上海云享科技有限公司 | Method and device for extracting palm print features and obtaining palm print images |
CN104933432B (en) * | 2014-03-18 | 2020-04-14 | 北京细推科技有限公司 | Processing method of finger abdomen crease and finger vein image |
CN105095849B (en) * | 2014-05-23 | 2019-05-10 | 财团法人工业技术研究院 | object identification method and device |
US9400924B2 (en) | 2014-05-23 | 2016-07-26 | Industrial Technology Research Institute | Object recognition method and object recognition apparatus using the same |
CN104318213A (en) * | 2014-10-21 | 2015-01-28 | 沈阳大学 | Method for using human body palm biology information to identify identities |
CN104392225B (en) * | 2014-12-09 | 2017-06-16 | 西安电子科技大学 | Online palmprint effective coverage fast partition method based on gradient direction template |
CN104636721B (en) * | 2015-01-16 | 2018-09-21 | 青岛大学 | A kind of palm grain identification method based on profile Yu Edge texture Fusion Features |
CN105426853A (en) * | 2015-11-24 | 2016-03-23 | 成都四象联创科技有限公司 | Human body characteristic identification method based on image |
CN107358185B (en) * | 2017-07-03 | 2019-10-25 | 上海奥宜电子科技有限公司 | Palm print and palm vein image-recognizing method and device based on topological analysis |
CN110879956A (en) * | 2019-07-12 | 2020-03-13 | 熵加网络科技(北京)有限公司 | Method for extracting palm print features |
CN110930562B (en) * | 2019-11-29 | 2021-07-13 | 浙江威欧希科技股份有限公司 | Intelligent lock and palm vein recognition optimization method applied to intelligent lock |
CN114612951B (en) * | 2022-05-11 | 2022-12-23 | 北京圣点云信息技术有限公司 | Palm feature recognition method |
CN116798077B (en) * | 2023-08-18 | 2023-11-07 | 江苏圣点世纪科技有限公司 | Palm photo detection method |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102156888A (en) * | 2011-04-27 | 2011-08-17 | 西安电子科技大学 | Image sorting method based on local colors and distribution characteristics of characteristic points |
US8229178B2 (en) * | 2008-08-19 | 2012-07-24 | The Hong Kong Polytechnic University | Method and apparatus for personal identification using palmprint and palm vein |
-
2013
- 2013-11-19 CN CN201310589834.6A patent/CN103559489B/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8229178B2 (en) * | 2008-08-19 | 2012-07-24 | The Hong Kong Polytechnic University | Method and apparatus for personal identification using palmprint and palm vein |
CN102156888A (en) * | 2011-04-27 | 2011-08-17 | 西安电子科技大学 | Image sorting method based on local colors and distribution characteristics of characteristic points |
Non-Patent Citations (3)
Title |
---|
《一种快速自适应边缘提取方法的研究》;蔡小琼等;《计算机测量与控制》;20060420;第14卷(第3期);第307-309页 * |
《基于局部灰度极小值的掌脉图像分割方法》;苑玮琦等;《光电子.激光》;20110731;第22卷(第7期);第1091-1096页 * |
《基于特征参量空间的掌脉识别方法研究》;苑玮琦等;《仪器仪表学报》;20130527;第34卷(第4期);第853-859页 * |
Also Published As
Publication number | Publication date |
---|---|
CN103559489A (en) | 2014-02-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103559489B (en) | Palm feature extracting method under a kind of noncontact imaging mode | |
Kanhangad et al. | Contactless and pose invariant biometric identification using hand surface | |
Kumar et al. | Human identification using finger images | |
Kanhangad et al. | A unified framework for contactless hand verification | |
US20150347833A1 (en) | Noncontact Biometrics with Small Footprint | |
Han et al. | Palm vein recognition using adaptive Gabor filter | |
CN102542281B (en) | Non-contact biometric feature identification method and system | |
JP4027118B2 (en) | User authentication method, program, and apparatus | |
JP6005750B2 (en) | Authentication apparatus and authentication method | |
Malassiotis et al. | Personal authentication using 3-D finger geometry | |
WO2017059591A1 (en) | Finger vein identification method and device | |
US20150186716A1 (en) | Learning-based estimation of hand and finger pose | |
Huang et al. | Local feature approach to dorsal hand vein recognition by centroid-based circular key-point grid and fine-grained matching | |
CA3199703A1 (en) | Systems and methods for performing fingerprint based user authentication using imagery captured using mobile devices | |
CN101980243B (en) | Binocular vision-based finger vein three-dimensional identification system | |
TW202038133A (en) | System and method for rapidly locating iris using deep learning | |
US20020090146A1 (en) | Hand recognition with position determination | |
Iula et al. | Multimodal biometric recognition based on 3D ultrasound palmprint-hand geometry fusion | |
CN106971131A (en) | A kind of gesture identification method based on center | |
Chen et al. | Contactless multispectral palm-vein recognition with lightweight convolutional neural network | |
Ma et al. | Contactless finger-vein verification based on oriented elements feature | |
Galbally et al. | 3D-FLARE: A touchless full-3D fingerprint recognition system based on laser sensing | |
Amayeh et al. | Hand-based verification and identification using palm–finger segmentation and fusion | |
Duan et al. | Estimating 3D finger pose via 2D-3D fingerprint matching | |
Morales et al. | Synthesis and evaluation of high resolution hand-prints |
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 | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20171107 Termination date: 20211119 |