CN110383283A - Method and apparatus for fingerprint of classifying - Google Patents

Method and apparatus for fingerprint of classifying Download PDF

Info

Publication number
CN110383283A
CN110383283A CN201880016174.3A CN201880016174A CN110383283A CN 110383283 A CN110383283 A CN 110383283A CN 201880016174 A CN201880016174 A CN 201880016174A CN 110383283 A CN110383283 A CN 110383283A
Authority
CN
China
Prior art keywords
fingerprint
image
feature
template
overlapping region
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.)
Pending
Application number
CN201880016174.3A
Other languages
Chinese (zh)
Inventor
大卫·廷达尔
肯尼斯·琼森
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Fingerprint kaana Kadun Intellectual Property Co.,Ltd.
Original Assignee
Fingerprint Cards AB
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Fingerprint Cards AB filed Critical Fingerprint Cards AB
Publication of CN110383283A publication Critical patent/CN110383283A/en
Pending legal-status Critical Current

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
    • G06V40/12—Fingerprints or palmprints
    • G06V40/1347—Preprocessing; Feature extraction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G06V10/757—Matching configurations of points or features
    • 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
    • G06V40/12—Fingerprints or palmprints
    • G06V40/1365—Matching; Classification
    • 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
    • G06V40/40—Spoof detection, e.g. liveness detection

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Human Computer Interaction (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Collating Specific Patterns (AREA)
  • Image Input (AREA)

Abstract

This disclosure relates to a kind of for determining that the verifying fingerprint image (31) obtained is genuine or personation method compared with registered fingerprint.This method comprises: being obtained the verifying fingerprint image of finger by means of fingerprint sensor；Feature is extracted from verifying fingerprint image；Obtain the storage template of the feature of at least one registered images (32) of registered fingerprint；The feature of extraction is matched with the geometric transformation of the feature of template, wherein the positive match between the feature and template characteristic of extraction is confirmed as matching internal (34)；Determine the overlapping region (33) between verifying fingerprint image and registered images；And it will verify what fingerprint image was classified as really or palmed off based on the number of matching internal related with identified overlapping region.

Description

Method and apparatus for fingerprint of classifying
Technical field
This disclosure relates to for determining that the verifying fingerprint image obtained is genuine or false compared with registered fingerprint The method and apparatus emitted.
Background technique
Various types of biological recognition systems are used, more and more to provide the use of the safety and/or enhancing that improve Family convenience.
Particularly, fingerprint sensing system is since their low-profile, high-performance and high user's acceptance are for example It is used in consumer electronic devices.
When being authenticated by means of fingerprint to user, the verifying fingerprint image of user's finger is obtained by means of fingerprint sensor Picture, and be compared with the fingerprint image of the registration of storage.By the fingerprint of the feature and registration extracted from verifying fingerprint image The feature templates of image are matched, to determine that the matching number of feature, feature-based matching number, verifying fingerprint are classified To be genuine or personation, and user is certified or is rejected.
Summary of the invention
Matching between the verifying fingerprint image feature extracted and the template of at least one registered images of registered fingerprint Feature can be classified as interior point or exceptional value, this depend on they whether meet template characteristic selection identical geometry become It changes.The number of the matching internal of correspondence candidate characteristic set in feature set and authentication image in registered images is that authentication image is It is no to be the index of genuine (that is, from real user) or personation (that is, user from personation), and threshold can be applied Value carries out final categorised decision.Therefore, matching internal (or interior point) with enough accuracy validations and can register symbol Close the feature of geometric transformation.Each matching internal is always related to two features just, and one from authentication image and one is come From registered images, and convert a Feature Mapping to another feature with enough accuracy.When point threshold value in setting, It is traded off between desired false acceptance rate (FAR) and false rejection rate (FRR).This is also referred to as the operating point of classifier. For lesser fingerprint sensor size, the number of the feature in registered images and authentication image will be inevitably less, and And it is likely difficult to reach desired operating point by counting classify according only to interior point.
It, can be by estimating the geometry between registered images and authentication image simultaneously when being matched to the feature of extraction (for example, by means of random sample common recognition (RANSAC) algorithm) is converted to determine matching internal collection.The transformation can be used for calculating Lap between registered images and authentication image, for example, being indicated with the number of pixel or with the relative percentage of image size It indicates.Embodiment according to the present invention, using the number of overlapping and interior point (that is, the matching internal of feature (count in also referred to as by point Number)) between expection correlation, will verify fingerprint image to create and be classified as genuine or personation improved procedure.
According to an aspect of the present invention, a kind of method executed in the electronic equipment for including fingerprint sensor is provided, This method is for determining that the verifying fingerprint image obtained is genuine compared with registered fingerprint or personation.This method includes borrowing Help the verifying fingerprint image for the finger that fingerprint sensor is obtained close to the detection surface of fingerprint sensor.This method further includes borrowing Help image procossing and extracts feature from verifying fingerprint image.This method further includes at least one the registration figure for obtaining registered fingerprint The template of at least one storage of the feature of picture.This method further includes the geometric transformation of the feature by extracted feature and template It is matched, wherein the positive match between extracted feature and template characteristic is confirmed as matching internal.This method further includes The overlapping region between verifying fingerprint image and at least one registered images is determined based on the geometric transformation of template characteristic.The party Method further includes that will be verified fingerprint image based on the number of matching internal related with identified overlapping region and be classified as really Or personation.
According to another aspect of the present invention, a kind of computer program product that component can be performed including computer is provided, It is used to make electronic equipment execute when the computer can be performed when component is run on including processing circuit in the electronic device The embodiment of the method for present disclosure.
According to another aspect of the present invention, a kind of fingerprint sensing system is provided, comprising: fingerprint sensor；Processing circuit； And data storage device for storing instruction, described instruction can be executed by the processing circuit, the thus fingerprint sense Examining system is manipulated into the verifying fingerprint image that the finger close to the detection surface of fingerprint sensor is obtained by means of fingerprint sensor Picture.The system, which is also manipulated into, extracts feature from verifying fingerprint image by means of image procossing.The system, which is also manipulated into, to be obtained Obtain template of at least one storage of the feature of at least one registered images of registered fingerprint.The system, which is also manipulated into, to be mentioned The feature taken is matched with the geometric transformation of the feature of template, wherein between extracted feature and template characteristic just With being confirmed as matching internal.The system be also manipulated into the geometric transformation based on template characteristic determine verifying fingerprint image with Overlapping region between at least one registered images.The system is also manipulated into based in related with identified overlapping region The matched number in portion will verify what fingerprint image was classified as really or palmed off.
According to another aspect of the present invention, a kind of embodiment of fingerprint sensing system including present disclosure is provided Electronic equipment.
It has now been realized that when authentication image is genuine, the usual number when interior point (that is, internal feature matching) When big, it is contemplated that authentication image and registered images have relatively large overlapping, and when the number of interior point is smaller, it is contemplated that proof diagram Picture and registered images have relatively small overlapping.This is only due to be directed to genuine authentication image, it is contemplated that the density of interior point is whole It is substantially homogeneous in a overlapping region.
On the contrary, the interior number put is usually opposite if authentication image is personation (that is, being the finger of unauthorized) It is low, and matching threshold should being arranged so as to, such authentication image is less likely to be authorized to.However, overlapping is not necessarily It is small, this is because point in vacation (for example, as it is in the region for concentrating on relatively small overlapping, in authentication image and registration figure As similar feature cluster in the two) it is likely to occur any position in the picture, thus point in generating.
Therefore, inventors have realised that for genuine authentication image, overlapping increases with the increase of the number of interior point Add, and for the authentication image of personation, the size of overlapping is usually uncorrelated to the number of interior point.
It follows that according to the present invention, if the relationship between the number of matching internal and identified overlapping region is dark Show interior point density be in entire overlapping region it is substantial uniform, then verifying fingerprint image can be classified as really, or If the relationship between the number of person's matching internal and identified overlapping region implies identified overlapping region and internal The number matched is uncorrelated, then verifies fingerprint image and be classified as personation.
It should be noted that any feature of any aspect can be applied to any other side in any situation appropriate Face.Similarly, any advantage of any aspect can be applied to any other aspect.According to content disclosed in detail below, appended Dependent claims and attached drawing, other purposes, the feature and advantage of disclosed embodiment will be apparent.
In general, unless the context otherwise clearly defined, otherwise all terms used in claims will according to they Ordinary meaning in the technical field is explained.Unless explicitly stated otherwise, otherwise to " one (a)/mono- (an)/be somebody's turn to do (the) member The all references of part, equipment, component, device, step etc. " is explained as referring to element, equipment, component, device, step to be disclosedly Suddenly at least one equal example.Unless expressly stated, otherwise any method disclosed herein the step of not necessarily press it is disclosed Exact sequence executes." first " that uses for different characteristic/component of present disclosure, " second " etc. are only intended to the spy Sign/component is distinguished with other similar features/components, and does not assign any sequence or level to this feature/component.
Detailed description of the invention
Embodiment is described by way of example with reference to the accompanying drawings, in the accompanying drawings:
Fig. 1 is according to the present invention include the fingerprint sensor interacted with the finger of user electronic equipment embodiment Schematic diagram.
Fig. 2 is the schematic block diagram of the embodiment of the fingerprint sensing system of electronic equipment according to the present invention.
Fig. 3 a shows the matching internal in overlapping between verifying fingerprint image and registered images according to the present invention Embodiment.
Fig. 3 b shows the matching internal in overlapping between verifying fingerprint image and registered images according to the present invention Another embodiment.
Fig. 4 a show it is according to the present invention verifying fingerprint image and multiple registered images between it is overlapping in inside The embodiment matched.
Fig. 4 b show it is according to the present invention verifying fingerprint image and multiple registered images between it is overlapping in inside Another embodiment matched.
Fig. 5 shows the decision boundary of the two-dimensional coordinate system of the number of overlapping (%) according to the present invention and matching internal.
Fig. 6 is the schematic flow chart of the embodiment of method of the invention.
Specific embodiment
Embodiment is described more fully with hereinafter with reference to the attached drawing for showing certain embodiments now.However, Scope of the present disclosure interior, many various forms of other embodiments are feasible.On the contrary, mentioning by way of example For following implementation, so that present disclosure will be comprehensive and complete, and will sufficiently be passed to those skilled in the art Up to scope of the present disclosure.Identical appended drawing reference refers to identical element throughout the specification.
Fig. 1 is shown herein as the electronic equipment 1 (for example, smart phone) of the form of mobile phone, the electronic equipment 1 packet Include display body 2 (e.g., including touch function (that is, touch display)) and fingerprint sensor 2.Fingerprint sensor 2 includes example It is such as used for the fingerprint sensor circuit of output gray level image, wherein the inspection of the different brightness instruction fingerprint sensor 2 in image The contact surveying surface and being placed between finger 4 above, for example, as the finger print identifying for using fingerprint sensor to carry out or leading A part of boat.
Fingerprint sensor 2 can be operated according to any detection technology.For example, fingerprint sensor can be condenser type, optics, Calorifics or ultrasonic sensor.Herein, discuss that for certain applications may be preferred capacitance type fingerprint as example Sensor.Fingerprint sensor may include the two-dimensional array of fingerprint sensing element, and each fingerprint sensing element corresponds to by fingerprint The pixel of the image of sensor output, for example, pixel is indicated by gray value.It is shown as shown in Figure 1, fingerprint sensor can be located at Show at the side of device 3, except the display area of display or fingerprint sensor can be located in the display area.It is defeated Image out is referred to herein as authentication image, this is because it be used for by the registered images of the registered fingerprint with storage into The verifying that row relatively carries out.For example, authentication image can be the two dimension of such as gray value or the form of one-dimensional pixel array.
Each image pixel can provide brightness of image, with gray value or other values.For example, for capacitance type fingerprint Sensor, high pixel intensity (for example, gray scale is white) mean low capacitive coupling and therefore, it is intended that detection surface and finger Biggish sensing distance between line pattern.Because finger does not cover corresponding to the detection surface where the sensing element of pixel Part, it is thus possible to lead to high pixel intensity.On the contrary, low pixel brightness (for example, gray scale is black) means high capacitance coupling It closes, and therefore, it is intended that detects the lesser sensing distance between surface and fingerprint pattern.Because of corresponding sensing element position In the spine of fingerprint pattern, it is thus possible to lead to low pixel brightness.Moderate pixel intensity can indicate that sensing element is opened up by finger It flutters covering but is located at the valley of fingerprint pattern.
Feature, such as the details of such as bifurcated and spine end, and other tables of verifying fingerprint are extracted from authentication image Levy feature.The other positions of each feature and the one group of x and y coordinates or restriction of specifying their corresponding positions in authentication image It is associated.The feature extracted from authentication image can be collected in validation template together with their own associated coordinate.
Similarly, the feature of registered images or multiple registered images can store in enrollment.As discussed herein , template, verifying or enrollment can be considered as the container for information associated with image, and the information includes feature (associated with coordinate), user identifier (ID), finger ID and/or other information.
Since the fingerprint of authentication image may not have size identical with the fingerprint in registered images, gradient, rotation Deng, or not same position in the picture, therefore the geometry of the feature for the feature and enrollment extracted from authentication image becomes Change matching.In other words, passed through according to two feature sets (feature set of the feature set and enrollment extracted from authentication image) They are mutually aligned (compensation such as relative translation and rotation and (optionally) scaling) and carrys out computational geometry transformation, in obtaining Point maximizes the number of interior point.In order to reduce complexity, such as three spies extracted from authentication image can be used first (for example, any three features) is levied to be matched with the geometric transformation of the registration feature of corresponding number (such as three).Geometry Transformation generally includes any one or all in the translation, rotation and scaling of template characteristic.Herein, translation is image sky Between in displacement/movement, that is, in registered images can for verifying feature find matched position；Rotation relates to how to revolve Turn authentication image and registered images (by the coordinate representation of their individual features), to find internal feature matching；And scaling The distance between zoom feature in a coordinated fashion is related to how, to find internal feature matching.RANSAC algorithm can be used for The suitable transformation of selection, it is therefore preferable to which affine transformation, for example, rigid affine transformation (only rotation and translation), this results in higher than pre- Determine the interior point amount of threshold value, authentication image for being potentially classified as really by the predetermined threshold.However, according to the present invention, testing Before card image is classified, it is also contemplated that according to the overlapping region between selected transformation, authentication image and registered images.When When feature from authentication image and registered images is compared to each other, some features are to (from authentication image and registration figure As each of a pair) may form interior point when meeting the first transformation, and other features are to may meet the second change Interior point is formed when changing, however other other features may not meet any common transformation at all.In this case, work as root The number of number and/or interior point related with overlapping region according to matching step of the invention for example based on interior point is determined with first When which of transformation and the second transformation carry out, makes a choice and (be similar in this hair between the first transformation and the second transformation Situation in bright classifying step).
Fig. 2 is the schematic block diagram of the embodiment of the fingerprint sensing system 10 of electronic equipment 1 for example as shown in figure 1.System System 10 includes processing circuit 11 (for example, central processing unit (CPU)).Processing circuit 11 may include the one of microprocessor form A or multiple processing units.However, other suitable equipment with computing capability are (for example, specific integrated circuit (ASIC), existing Field programmable gate array (FPGA) or Complex Programmable Logic Devices (CPLD)) it may include in processing circuit 11.Processing electricity Road 11 is configured to run one in the data storage device 14 for being stored in one or several storage units (for example, memory) Or several computer programs or software (SW) 15.By running SW 15, at least part including processing circuit 11 can be formed Application, for example, be configured for being obtained by means of the fingerprint sensor 2 in system 10 for example verify fingerprint image finger Print image obtain circuit 12, and be configured for such as acquisition verifying fingerprint image image procossing (for example, for from In authentication image extract feature and this feature is matched with the transformation of the feature in enrollment) image processing circuit 13.Storage unit is considered as computer readable device 14 as discussed herein, and may, for example, be random access memory (RAM), flash memory or other solid-state memories or hard disk, or combinations thereof form.Processing circuit 11 can also be configured At the information for example about the enrollment of registered images 16 is stored in storage device 14 as needed.
Electronic equipment 1 can be any equipment including fingerprint sensor 2, and is capable of handling and is passed by means of the fingerprint The image that sensor obtains.For example, equipment 1 can be mobile phone (such as smart phone), smart card, tablet computer, portable In computer (for example, laptop computer) or fixed computer (for example, desktop computer, server or mainframe computer) Any one.
Fig. 3 a and Fig. 3 b show the matching internal 34 in overlapping 33 between verifying fingerprint image 31 and registered images 32 Embodiment.For true authentication image 31, the situation in Fig. 3 a is typical, wherein interior point 34 is (that is, in overlapping region Positive match feature in 33) it is dispersed substantially uniformly in entire lap.In contrast, in the case where in fig 3b, Interior point concentrates on only limited, the relatively small part of overlapping.The number of interior point 34 in Fig. 3 b itself can be higher than for true The classification thresholds of authentication image, but in view of relatively large overlapping, it is contemplated that have the interior point of greater number, therefore authentication image quilt It is classified as personation.However, if overlapping it is smaller, report by mistake may be instead classification result.
In certain embodiments of the present invention, multiple registered images 32 related with same finger 4 can be used and carry out shape The enrollment 16 stored at least one.In some embodiments, the enrollment 16 of storage is based on multiple registered images 32.In some embodiments, multiple registered images 32 can be stitched together to form single splicing registered images.At it In his embodiment, the enrollment 16 of storage includes the spy of each of multiple registered images 32 of independent (non-splicing) Sign, it means that for example, if registered images itself are overlapped, then there may be phases in the template from more than one registered images Same feature.In some other implementations, multiple enrollments, each of multiple registered images 32 corresponding one are stored with A enrollment.
Fig. 4 a and Fig. 4 b show overlapping 33 example between authentication image 31 and multiple registered images 32.Fig. 4 a's In example, there are three registered images 32 for being expressed as 32a, 32b and 32c.As in Fig. 4 a as it can be seen that registered images 32 are opened up It opens, i.e. the different piece of the registered fingerprint all including finger 4.Therefore, each registered images 32 its own with authentication image 31 Different overlappings 33.Herein, registered images 32a generates overlapping 33a, registered images 32b generates overlapping 33b and registration Image 32c generates overlapping 33c.The case where Fig. 4 a, may be better than situation shown in Fig. 4 b, this is because overlapping 33a, 33b and 33c Combined overlapping region be greater than it is in Fig. 4 b for being formed by four registered images 32a, 32b, 32c and 32d, fingerprint is shown The overlapping region 33 of same section.However, the registered images of Fig. 4 b can be for example with different quality, so that generating has The combination image of quality is improved, therefore, here it is compared with only one registered images, possessing multiple registered images can be still Right advantageous reason.
It can reasonably believe that the matching in matching ratio Fig. 4 b in Fig. 4 a conveyed more information, this is because registration figure Picture 32 is Chong Die with the major part of authentication image 31.It can include arriving feature by overlapped 33 to distinguish both of these case In vector:
X=[s (v, t1), s (v, t2) ..., s (v, tN), o (v, t1), o (v, t2) ..., o (v, tN),
o(t1, t2), o (t1, t3) ..., o) (tN-1,tN)].
Here, s (a, b) and o (a, b) are score (number of interior point) between image a and image b respectively and Chong Die.It is right Registered images tiPre-sorting is carried out, so that t1It is the image with the interior point 34 of highest number, with authentication image v.N is us Consider the amount for the registered images 32 attempted for one-time authentication.
In the case where multiple registered images 32, score can be obtained for each registered images (for example, probability or interior point Count), and the interior number for classification can be calculated from these scores.For example, the highest score observed can be used Make the final score in a manner of winner overwhelm the market.Other than the registered images with highest score, check that other candidates can also It can be suitable.For example, we, which can check, has time high score if the image with highest score is immediately lower than classification thresholds Several images.If the score is also close to classification thresholds, it is genuine for may having stronger evidence checking image.
As discussed herein, in the case that the number especially put inside is few, for example, the case where close to classification thresholds Under, the number (also referred to as score or counting) of interior point 34 may be not enough to for authentication image being correctly classified as genuine or personation 's.Certainly, it is also contemplated that the adeditive attribute of overlapping region 33.
User or the number and the threshold value of overlapping 33 for interior point 34 that manufacturer defines can be used, but can be more excellent Selection of land is using machine learning (for example, neural network model, such as Multilayer Perception (MLP) model, convolutional neural networks (CNN) model Or support vector machines (SVM) model) determine decision boundary 51 as shown in Figure 5 with for classifying.
If training MLP classifier, to find best decision boundary 51, a feature can indicate that subtemplate matches, and And by being formed the number of interior point 34 is associated with overlapping 33: x=[interior point is overlapped].
In Fig. 5, the oblique line in two-dimensional coordinate system is shown as by means of the decision boundary 51 that MLP model obtains, wherein The number of interior point and overlapping are used as two dimensions.Therefore, classification can include determining that the point in two-dimensional coordinate system, wherein internal The number of matching 34 corresponds to the first coordinate of the point and overlapping region 33 corresponds to the second coordinate of the point.It is then possible to Be located at according to the first coordinate of the point and the second coordinate according to the point the predetermined decision boundary 51 in coordinate system which side 52 or 53 classify.In the example of hgure 5, if the point is located in the right regions 52 of coordinate system, authentication image is classified To be genuine, and if the point is located in the lower left region 53 of coordinate system, authentication image is classified as personation.It can see Out, as the number of interior point 34 increases, large range of overlapping is acceptable for being classified as really, and subtracts for the number of interior point Few, received overlapping range narrows, this is because, if the number of interior point is smaller, we are only pre- for genuine classification Count relatively small overlapping.
Fig. 6 is the flow chart for showing the different embodiments of method of the invention.This method can include fingerprint sensing Executed in the electronic equipment 1 of device 2, for determine obtain verifying fingerprint image 31 with register fingerprint compared be it is genuine or Personation.This method includes for example obtaining (S1) by using image acquisition circuit 12 by means of fingerprint sensor to pass close to fingerprint The verifying fingerprint image of the finger 4 on the detection surface of sensor.This method further includes by means of image procossing for example by using figure As processing circuit 13 extracts (S2) feature from verifying fingerprint image.This method further includes for example by means of using image procossing Circuit 13 come obtain at least one registered images 32 in (S3) registered fingerprint feature at least one storage (for example, depositing In storage device 14) template 16.This method further includes for example by means of using image processing circuit 13 that will extract the feature of (S2) (S4) is matched with the geometric transformation of the feature of the template of acquisition (S3), wherein between the feature and template characteristic of extraction Positive match is confirmed as matching internal 34.This method further includes that template spy is for example based on by means of using image processing circuit 13 The geometric transformation of sign determines the overlapping region 33 between (S5) verifying fingerprint image and at least one registered images.This method is also wrapped Include the number based on matching internal relevant to identified overlapping region will verify fingerprint image classification (S9) be it is genuine or Personation.
In certain embodiments of the present invention, classification S9 includes number and the overlapping region 33 of determining matching internal 34 Whether ratio is higher than predtermined category threshold value.
In certain embodiments of the present invention, this method further includes before classification (S9) by means of machine learning model (for example, neural network model, such as MLP model, CNN model or SVM model) determines S6 decision boundary 51, at classification (S9) Use the decision boundary.In some embodiments, classification (S9) includes the point in determining two-dimensional coordinate system, wherein internal Number with 34 corresponds to the first coordinate of the point, and overlapping region 33 corresponds to the second coordinate of the point, and according to this The first coordinate and the second coordinate of point determine the which side 52 or 53 of predetermined decision boundary 51 that the point is located in coordinate system.
In certain embodiments of the present invention, the template 16 of at least one storage includes based on comprising at least one registration The single template of multiple registered images of image 32.
In some other embodiments of the invention, the template 16 of at least one storage includes the list based on stitching image A template, the stitching image is by being stitched together to form multiple notes comprising at least one registered images 32 of stitching image Volume image composition.
In some other embodiments of the invention, the template 16 of at least one storage includes to contain at least one One template of each of multiple registered images of registered images 32, the template are also referred to as subtemplate.
In certain embodiments of the present invention, this method further includes that (S7) matching internal 34 is determined before classification (S9) Number be higher than predetermined first threshold, first threshold instruction verifying fingerprint may be genuine.
In certain embodiments of the present invention, this method further includes that (S8) matching internal 34 is determined before classification (S9) Number be lower than predetermined second threshold, second threshold instruction verifying fingerprint may be personation.
In certain embodiments of the present invention, the template 16 of at least one storage is from including in the electronic device 1 It is obtained in data storage device 14.
Instruction usually as the SW 15 being stored in storage device 14 is the non-transient meter to form computer program product The form of component can be performed in computer in calculation machine readable medium (for example, storage device 14), and computer program product is configured At for making equipment (for example, electronic equipment 1 and/or system 10) when instructing and running on 15 processing circuit 11 in a device It is able to carry out the embodiment of this method as described above.
Therefore, computer program product 14 includes that component 15 can be performed in computer, and computer can be performed component 15 and be used for Component can be performed in computer makes electronic equipment 1 execute public affairs herein when running on the processing circuit 11 being included in electronic equipment The embodiment for the method opened.
Therefore, fingerprint sensing system 10 may include: fingerprint sensor 2；Processing circuit 11；And for storing instruction 15 Data storage device 14, instruction 15 can execute by the processing circuit, and thus the fingerprint sensing system is manipulated into: by In fingerprint sensor obtain (S1) close to fingerprint sensor detection surface finger 4 verifying fingerprint image 31；By means of figure As processing extracts (S2) feature from verifying fingerprint image；Obtain the feature of at least one registered images 32 of (S3) registered fingerprint At least one storage template 16；The feature of extraction is matched into (S4) with the geometric transformation of template characteristic, wherein extract Feature and template characteristic between positive match 34 be confirmed as matching internal；It is determined based on the geometric transformation of template characteristic (S5) overlapping region 33 between fingerprint image 31 and at least one registered images 32 is verified；And based on it is identified overlapping It is genuine or personation that the number of the related matching internal 34 in region, which will verify fingerprint image classification (S9),.In some embodiment party In formula, electronic equipment 1 discussed in this article includes the embodiment of fingerprint sensing system 10.
Above referring especially to some embodiments describe present disclosures.However, as those skilled in the art holds Intelligible, the other embodiments other than embodiments disclosed above can equally be limited by the appended claims It is fixed that scope of the present disclosure interior.

Claims (14)

1. a kind of method executed in the electronic equipment (1) including fingerprint sensor (2), the method is for determining acquisition Verifying fingerprint image (31) is genuine compared with registered fingerprint or personation, which comprises
By means of the fingerprint sensor, obtain (S1) close to the fingerprint sensor the finger (4) for detecting surface verifying Fingerprint image；
By means of image procossing, (S2) feature is extracted from the verifying fingerprint image；
Obtain the template (16) of at least one storage of the feature of at least one registered images (32) of (S3) registered fingerprint；
Extracted feature is matched into (S4) with the geometric transformation of the feature of the template, wherein extracted feature with Positive match between template characteristic is confirmed as matching internal (34)；
Determined based on the geometric transformation of template characteristic (S5) described verifying fingerprint image and at least one described registered images it Between overlapping region (33)；And
It is by verifying fingerprint image classification (S9) based on the number of matching internal related with identified overlapping region It is genuine or personation；
Wherein, if the relationship between the number of the matching internal and identified overlapping region implies the interior density put whole Be in a overlapping region it is substantial uniform, then it is genuine that the verifying fingerprint image, which is classified (S9), or if institute The relationship stated between the number of matching internal and identified overlapping region implies identified overlapping region and the inside The number matched is uncorrelated, then it is personation that the verifying fingerprint image, which is classified (S9),.
2. according to the method described in claim 1, wherein, the classification (S9) comprises determining that the number of the matching internal (34) Whether the ratio of mesh and the overlapping region (33) is higher than predtermined category threshold value.
3. method according to claim 1 or 2, before the classification (S9), further includes: by means of machine learning mould Type, for example, neural network model determines (S6) decision boundary such as Multilayer Perception MLP model or support vector machines model (51), the decision boundary is used in the classification.
4. according to the method described in claim 3, wherein, the classification (S9) comprises determining that the point in two-dimensional coordinate system, in institute It states in two-dimensional coordinate system, the number of the matching internal (34) corresponds to the first coordinate of the point, and the overlapping region (33) correspond to the second coordinate of the point；And determine that the point is located at according to the first coordinate of the point and the second coordinate The which side (52/53) of predetermined decision boundary (51) in coordinate system.
5. method according to any preceding claims, wherein the template (16) of at least one storage includes being based on The single template of multiple registered images comprising at least one registered images (32).
6. the method according to any one of preceding claims 1 to 4, wherein the template (16) of at least one storage Including the single template based on stitching image, what the stitching image formed the stitching image by being stitched together includes institute State multiple registered images composition of at least one registered images (32).
7. the method according to any one of preceding claims 1 to 4, wherein the template (16) of at least one storage A template including each of multiple registered images to contain at least one registered images (32).
8. method according to any preceding claims, before the classification (S9), further includes: in determining that (S7) is described The number that portion matches (34) is higher than predetermined first threshold, and the predetermined first threshold indicates that the verifying fingerprint may be genuine.
9. method according to any preceding claims, before the classification (S9), further includes: in determining that (S8) is described The number that portion matches (34) is lower than predetermined second threshold, and the predetermined second threshold indicates that the verifying fingerprint may be personation 's.
10. method according to any preceding claims, wherein template (16) of at least one storage is from including It is obtained in data storage device (14) in the electronic equipment (1).
11. one kind includes the computer program product (14) that component (15) can be performed in computer, it is used for when the computer can Before executive module makes the electronic equipment (1) execution any when running on the processing circuit (11) being included in electronic equipment State method described in claim.
12. a kind of fingerprint sensing system (10), comprising:
Fingerprint sensor (2)；
Processing circuit (11)；And
The data storage device (14) of (15), described instruction (15) can be executed by the processing circuit for storing instruction, by This described fingerprint sensing system operation at:
By means of the fingerprint sensor, the verifying fingerprint of the finger (4) close to the detection surface of the fingerprint sensor is obtained Image (31)；
By means of image procossing, feature is extracted from the verifying fingerprint image；
Obtain the template (16) of at least one storage of the feature of at least one registered images (32) of registered fingerprint；
Extracted feature is matched with the geometric transformation of the feature of the template, wherein extracted feature and template Positive match (34) between feature is confirmed as matching internal；
The verifying fingerprint image (31) and at least one described registered images are determined based on the geometric transformation of template characteristic (32) overlapping region (33) between；And
The verifying fingerprint image is classified as based on the number of matching internal (34) related with identified overlapping region It is genuine or personation；
Wherein, if the relationship between the number of the matching internal and identified overlapping region implies the interior density put whole Be in a overlapping region it is substantial uniform, then the verifying fingerprint image is classified as really, or if it is described in Relationship between the matched number in portion and identified overlapping region implies identified overlapping region and the matching internal Number is uncorrelated, then the verifying fingerprint image is classified as personation.
13. a kind of electronic equipment (1), including the fingerprint sensing system (10) described in claim 12.
14. electronic equipment according to claim 13, wherein the electronic equipment (1) is mobile phone, for example, intelligence Phone；Smart card；Tablet computer；Portable computer, for example, laptop computer；Or fixed computer, for example, desk-top Computer, server or mainframe computer.
CN201880016174.3A 2017-12-18 2018-12-10 Method and apparatus for fingerprint of classifying Pending CN110383283A (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
SE1751569A SE1751569A1 (en) 2017-12-18 2017-12-18 Method and device for classifying fingerprint
SE1751569-3 2017-12-18
PCT/SE2018/051264 WO2019125270A1 (en) 2017-12-18 2018-12-10 Method and device for classifying fingerprint

Publications (1)

Publication Number Publication Date
CN110383283A true CN110383283A (en) 2019-10-25

Family

ID=66998028

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201880016174.3A Pending CN110383283A (en) 2017-12-18 2018-12-10 Method and apparatus for fingerprint of classifying

Country Status (3)

Country Link
CN (1) CN110383283A (en)
SE (1) SE1751569A1 (en)
WO (1) WO2019125270A1 (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6778685B1 (en) * 1999-07-08 2004-08-17 Nanyang Technological University Two-stage local and global fingerprint matching technique for automated fingerprint verification/identification
WO2012138086A2 (en) * 2011-04-06 2012-10-11 조선대학교산학협력단 Apparatus and method for hierarchically verifying fingerprints using similarity distribution
CN104463129A (en) * 2014-12-17 2015-03-25 浙江维尔科技股份有限公司 Fingerprint registration method and device
US20160042247A1 (en) * 2014-08-11 2016-02-11 Synaptics Incorporated Multi-view fingerprint matching
CN105447436A (en) * 2014-12-19 2016-03-30 比亚迪股份有限公司 Fingerprint identification system, fingerprint identification method and electronic device

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105205439B (en) * 2015-02-13 2017-05-03 比亚迪股份有限公司 Method for calculating area of fingerprint overlapping region and electronic device

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6778685B1 (en) * 1999-07-08 2004-08-17 Nanyang Technological University Two-stage local and global fingerprint matching technique for automated fingerprint verification/identification
WO2012138086A2 (en) * 2011-04-06 2012-10-11 조선대학교산학협력단 Apparatus and method for hierarchically verifying fingerprints using similarity distribution
US20160042247A1 (en) * 2014-08-11 2016-02-11 Synaptics Incorporated Multi-view fingerprint matching
CN104463129A (en) * 2014-12-17 2015-03-25 浙江维尔科技股份有限公司 Fingerprint registration method and device
CN105447436A (en) * 2014-12-19 2016-03-30 比亚迪股份有限公司 Fingerprint identification system, fingerprint identification method and electronic device

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
ARCHANA C. LOMTE: "Biometric fingerprint authentication with minutiae using ridge feature extraction", 《2015 INTERNATIONAL CONFERENCE ON PERVASIVE COMPUTING (ICPC)》 *

Also Published As

Publication number Publication date
SE1751569A1 (en) 2019-06-19
WO2019125270A1 (en) 2019-06-27

Similar Documents

Publication Publication Date Title
US20220165087A1 (en) Systems and methods for performing fingerprint based user authentication using imagery captured using mobile devices
Alvarez-Betancourt et al. A keypoints-based feature extraction method for iris recognition under variable image quality conditions
CN106663204B (en) Device for the certification based on fingerprint and computer implemented method
US10108858B2 (en) Texture features for biometric authentication
Silva et al. An approach to iris contact lens detection based on deep image representations
Guo et al. Contact-free hand geometry-based identification system
Agarwal et al. Local binary hexagonal extrema pattern (LBHXEP): A new feature descriptor for fake iris detection
CN109661670A (en) For relative to template rather than the system and method for the fraud detection according to absolute measure
US20220327189A1 (en) Personalized biometric anti-spoofing protection using machine learning and enrollment data
US11238271B2 (en) Detecting artificial facial images using facial landmarks
US11341348B2 (en) Hand biometrics system and method using digital fingerprints
Labati et al. Touchless fingerprint biometrics
Krishneswari et al. A review on palm print verification system
Kamboj et al. CED-Net: context-aware ear detection network for unconstrained images
Giełczyk et al. Lightweight Verification Schema for Image‐Based Palmprint Biometric Systems
Ibsen et al. Digital face manipulation in biometric systems
CN116884045B (en) Identity recognition method, identity recognition device, computer equipment and storage medium
Wang et al. Sphere-spin-image: A viewpoint-invariant surface representation for 3D face recognition
Wojciechowska et al. The overview of trends and challenges in mobile biometrics
CN110383283A (en) Method and apparatus for fingerprint of classifying
Harris et al. An Improved Signature Forgery Detection using Modified CNN in Siamese Network
Bhavani et al. A multi-dimensional review on handwritten signature verification: strengths and gaps
Sehgal Palm recognition using LBP and SVM
Faez et al. Personal verification using ear and palm-print biometrics
Ramalingam et al. Fundamentals and advances in 3D face recognition

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
TA01 Transfer of patent application right

Effective date of registration: 20211206

Address after: Gothenburg

Applicant after: Fingerprint kaana Kadun Intellectual Property Co.,Ltd.

Address before: Gothenburg

Applicant before: FINGERPRINT CARDS AB

TA01 Transfer of patent application right
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20191025

WD01 Invention patent application deemed withdrawn after publication