WO2016023582A1 - Procédé de détection d'une présentation falsifiée à un système de reconnaissance vasculaire - Google Patents

Procédé de détection d'une présentation falsifiée à un système de reconnaissance vasculaire Download PDF

Info

Publication number
WO2016023582A1
WO2016023582A1 PCT/EP2014/067290 EP2014067290W WO2016023582A1 WO 2016023582 A1 WO2016023582 A1 WO 2016023582A1 EP 2014067290 W EP2014067290 W EP 2014067290W WO 2016023582 A1 WO2016023582 A1 WO 2016023582A1
Authority
WO
WIPO (PCT)
Prior art keywords
entity
images
recognition system
binary descriptors
presented
Prior art date
Application number
PCT/EP2014/067290
Other languages
English (en)
Inventor
Sébastien MARCEL
Original Assignee
Fondation De L'institut De Recherche Idiap
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 Fondation De L'institut De Recherche Idiap filed Critical Fondation De L'institut De Recherche Idiap
Priority to PCT/EP2014/067290 priority Critical patent/WO2016023582A1/fr
Priority to EP14752602.4A priority patent/EP3180736A1/fr
Publication of WO2016023582A1 publication Critical patent/WO2016023582A1/fr

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
    • G06V40/1382—Detecting the live character of the finger, i.e. distinguishing from a fake or cadaver finger
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/10—Image acquisition
    • G06V10/12—Details of acquisition arrangements; Constructional details thereof
    • G06V10/14—Optical characteristics of the device performing the acquisition or on the illumination arrangements
    • G06V10/143—Sensing or illuminating at different wavelengths
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/42—Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation

Definitions

  • the present invention concerns method of detecting a falsified presentation to a vascular recognition system and in particular to such a method in which images of an entity which is presented to the system are obtained when the entity is illuminated at different wavelengths, and using those image to determine the probability that the entity is falsified.
  • the present invention also concerns the use of hand palm and finger as biometric characteristics.
  • vascular network pattern i.e. the pattern of the person's veins and arteries, typically 3mm below the surface of the skin
  • vascular recognition This practice of recognition of an individual through their vascular network pattern is commonly known as vascular recognition.
  • the network pattern of the persons hand palm or finger is used in the practice of vascular recognition.
  • the vascular image of veins is typically captured under near- infrared (NIR) illumination using a reflection method or a transmission method.
  • NIR near- infrared
  • the reflection method the reflected NIR light that is emitted from the object is captured using a CCD camera with an infrared filter.
  • the transmission method the NIR light is captured after transmission through the object (finger for example).
  • the principle of each method is that veins that carry deoxygenated hemoglobin absorb light within the wavelength from 760nm to 880nm (NIR range). Hence, under NIR illumination veins appear as dark lines that can be extracted and used for user identification or authentication.
  • Vascular recognition is becoming more popular as it has very low error rates due to the distinct differences which exist between individual's vascular network patterns. A further advantage is that the method of recognition leaves no trace (unlike fingerprints) hence making it more difficult to make a copy to forge a fake.
  • vascular recognition systems and methods are not without limitations. For example it is possible for an unrecognized person to fool existing systems by obtaining an image of the vascular network pattern of a recognised person and to present a fake, such as a print-out on paper, to the system. Presenting a falsified biometric on the sensor of a biometric recognition system is known as "spoofing" or presentation attack.
  • a common way to build a fake for vascular recognition is to print solely the vein image on a uniformly white paper. Since the toner from most laser printer absorbs the near infrared wavelength, a typical vascular recognition system based on images captured in NIR field could easily be spoofed by this attack.
  • Patent application US201 1304720 describes a device and a method of automated personal identification that uses finger vein and finger surface images acquired at the same time to extract near-infrared (NIR) and visible features. These features are computed to provide matching scores and are combined to determine whether the person is genuine or an imposter.
  • NIR near-infrared
  • Patent application US2009232362 describes a biometric
  • an anti- spoofing method for a biometric vascular recognition system comprising the steps of:
  • the biometric vascular recognition system may use an image of the vascular network in the hand palm, in the wrist, in the dorsal hand, in a finger and/or in the retina for example.
  • the method of the present invention uses both the texture of an image of the entity and the manner in which some elements (such as pixel or features) change between successive images of the entity in the sequence, in order to detect falsification; thus providing for a more reliable detection of a spoof.
  • the entity could be simultaneously illuminated with light which has wavelengths in a first wavelength range only and with light which has wavelengths in a second wavelength range only.
  • the entity could be consecutively illuminated with light which has wavelengths in a first wavelength range only and then
  • this method of the present invention uses a combination of both multi-spectrum (i.e. use of images captured with illumination in first and second wavelength ranges) and a multi-algorithm (features of static images and dynamic texture of the sequence of images) to detect falsification, thus providing for a more reliable detection of a spoof.
  • the step of determining if the entity which is presented to the vascular recognition system is a real biometric characteristic or a spoof, based on the plurality of first dynamic binary descriptors and second binary descriptors may comprise:
  • the step of determining if the entity which is presented to the vascular recognition system is a real biometric characteristic or a spoof, based on the plurality of first dynamic descriptors, may comprise,
  • the step of determining if the entity which is presented to the vascular recognition system is a real biometric characteristic or a spoof, based on the plurality of binary descriptors, may comprise,
  • the first and second wavelength ranges are non- overlapping ranges.
  • the first wavelength range may be in the near infrared (NIR) spectrum and used for revealing the vein pattern as well as changes of the vein pattern between frames.
  • the second wavelength may be in the visible spectrum and reveal the skin structure.
  • This multi-spectrum method has the advantage that it becomes more difficult to build a fake that is realistic both in the visible spectrum and in the NIR spectrum. For example, a print-out of a vein pattern on a white paper will miss the texture of the skin in the visible spectrum range.
  • an anti-spoofing method for a biometric vascular recognition system comprising the steps of:
  • the plurality of different wavelength ranges may comprise the ranges 350nm-740nm (visible), 740nm-1400nm (NIR), 760nm-880nm (sub- NIR) and 8000nm-1 5000nm (LWIR).
  • step (c) recognising the user if step (a) indicates that the entity which is presented is a real biometric characteristic (for example a real hand palm) and if user has been identified in step (b).
  • a real biometric characteristic for example a real hand palm
  • steps (a) and (b) are performed in parallel.
  • the step of identifying the user may comprise, comparing said first and second binary descriptors with one or more reference binary descriptors each representative of texture of a reference image. Other identification methods may be used.
  • biometric vascular recognition system comprising:
  • a light source configured to respectively illuminate an entity which is presented to the system, with light which has exclusively first and second wavelength ranges only;
  • image capturing means which can capture an image and a sequence of images of the entity when the entity is illuminated by the light source to provide a sequence of images and at least one static image;
  • a processor which is configured to compute a plurality of first binary descriptors representative of the texture of at least one of said images in the sequence of images and of the manner in which some elements of the image change between successive images in a sequence of images of the sequence, and to compute second binary descriptors for each of said at least one static images, and to determine if the entity which is presented to the vascular recognition system is a spoof, based on the plurality of first and second binary descriptors.
  • the present invention is also related to a computer readable storage medium having recorded thereon a computer program, the computer program comprising an algorithm capable of any of the methods which are herein described.
  • Figure 1 is a flow chart illustrating the steps involved in an anti- spoofing method for a biometric vascular recognition system, according to an embodiment of the present invention
  • Figure 2 is a block diagram illustrating a method of vascular recognition according to a further aspect of the present invention.
  • Figure 3 shows a block diagram illustrating the features of a biometric vascular recognition system
  • Figure 4 illustrates the computation of a binary descriptor based on a Local Binary Pattern (LBP).
  • LBP Local Binary Pattern
  • Figure 5 illustrates a vascular anti-spoofing system based on a multi-algorithm and multi-spectral framework. Detailed Description of possible embodiments of the Invention
  • Figure 1 is a flow chart illustrating the steps involved in an anti- spoofing method for a biometric vascular recognition system, according to an embodiment of the present invention.
  • the vascular recognition system may use for example vein patterns in the hand palm, in the fingers, etc.
  • the method comprises the steps of:
  • the method of the present invention uses both the texture of an image of the entity and the manner in which some elements of texture change between successive images in the sequence, in order to detect falsification; thus providing for a more reliable detection of a spoof.
  • the system will capture an image of the fake and will identify if the texture of that image corresponds to expected textures (the expected textures will be textures of images of real biometric characteristics).
  • the texture of the image will not correspond to an expected texture, so it can be determined that the presentation is a spoof.
  • the method of the present invention takes a second measure to detect spoofing: since the fake produces a still image of a biometric vascular system there would be no difference in texture of successive images in the sequence (or only changes due to a move of the fake between frames); this is in contrast to if a real hand-palm has been presented to the recognition system; because of the flow of blood through the vascular system in the biometric characteristic (for example in the hand palm), the successive images in the sequence of images will differ, thus providing for a changes in texture of successive images in the sequence.
  • the first dynamic binary descriptors are representative of the manner in which some elements of texture change between successive images in the sequence, over several frames and/or over the whole sequence; thus a spoof will be detected if the plurality of first dynamic binary descriptors indicate that there is no difference in texture between successive images in the sequence, or if the changes do not correspond to some expected patterns of change.
  • the fake is a still image of a hand-palm vascular system
  • the first dynamic binary descriptors will indicate that there is no difference in texture between successive images in the sequence (or only differences corresponding to a move of the whole fake), thus spoofing will be detected.
  • the user may move the fake when presenting it to the
  • the method may involve determining if the plurality of first dynamic binary descriptors correspond with a plurality of predefined expected dynamic binary descriptors, and determining that the entity which is presented to the vascular recognition system is a spoof if the plurality of first dynamic binary descriptors fails to correspond with a plurality of predefined expected dynamic binary descriptors.
  • the first dynamic binary descriptors will indicate that there is some differences in texture between successive images in the sequence, the first dynamic binary descriptors will fail to correspond to any predefined expected dynamic binary descriptors so that the fake presented will still be recognised as a spoof.
  • the method uses a combination of both multi-spectrum (first and second wavelength ranges) and a multi-algorithm (features of static images and dynamic texture of the sequence of images) to detect a spoof. This provides for a more reliable detection of a spoof.
  • the first and second wavelength ranges may be non-overlapping ranges; for example the first wavelength range may be 741 nm-1400nm (near-infra red spectrum) and the second wavelength range may be 350nm-740nm (visible spectrum). It is nearly impossible to provide a spoof which will appear realistic to the recognition system under the two different light conditions.
  • the print-out may appear to be realistic under the near infrared light in the range 741 nm-1400nm since the ink on the print-out will absorb the light with wavelength in this range.
  • this print-out is likely to miss the skin texture that will be seen when a real biometric characteristic is illuminated with light in the visible range, i.e. light with wavelengths in the range 350nm-740nm as all the features of the print out will reflect the light with wavelengths in the range 350nm-740nm.
  • the print-out will be revealed as a spoof when illuminated with light which has wavelengths in the visible range only.
  • the entity could be simultaneously illuminated with light which has wavelengths in a first wavelength range only and light which has wavelengths in a second wavelength range only.
  • the entity could be illuminated simultaneously with light which has wavelengths in the range 350nm-740nm (visible spectrum), and with light which has wavelengths in the range 741 nm-1400nm (near-infra red spectrum).
  • the entity could be consecutively illuminated with light which has wavelengths in a first wavelength range only and then illuminated with light which has wavelengths in a second wavelength range only.
  • the entity could be illuminated consecutively with light which has wavelengths in the range 350nm-740nm (visible spectrum), and with light which has wavelengths in the range 741 nm-1400nm (near- infra red spectrum), so that the entity is exclusively illuminated with light which has wavelengths in a 350nm-740nm (visible spectrum) only for a first time period, and is then exclusively illuminated with light which has wavelengths in the range 741 nm-1400nm (near-infra red spectrum) only for a second time period.
  • the computation of the second binary descriptors which represent the spatial texture of an image may be done using a texture descriptor.
  • a Local Binary Pattern is used as texture descriptor.
  • a LBP descriptor converts a pixel into a decimal code taking into account the neighbourhood of this pixel in the image plane.
  • the computation of a Local Binary Pattern involves a binary comparison of each pixel with the surrounding ones; a 0 value is determined for all pixels having a brightness value lower than the brightness value at the center, and a " 1 " when the brightness is higher.
  • the resulting binary value (001 1 1001 in the example) corresponds to a decimal value (57) which depends on the direction and sense of the brightness change at each pixel.
  • a histogram may then be formed from a plurality of values over the whole image or over several point of interests.
  • the second binary descriptor may be represented with a series of decimal values, each value depending on a pixel value and the values of the surrounding pixels in the spatial plane.
  • the pixel value may be the pixel brightness and/or colour/ and/or saturation, etc, or any combination.
  • the element of image might be pixels, or groups of pixels, or other features.
  • the change may reflect change in brightness, and/or change of intensity, colour, hue, etc, of the pixels.
  • the first dynamic binary descriptors may be represented with binary numbers representative of spatial and temporal changes of elements of image. For example, all pixel values (in space and time) over a series of frames are transformed into binary descriptors representing variations of those values with values of (spatially) surrounding pixels and of previous and next corresponding pixels.
  • a spatio-temporal texture descriptor named LBP-TOP is used for this purpose.
  • the LBP-TOP operator converts each pixel of a series of frames into a value, by considering the sequence of images as a three dimensional volume XYT and operating the above-described LBP on each plane ⁇ , ⁇ and YT.
  • each element of the image for example each pixel, might be represented by a value representing the difference between this pixel and the spatially surrounding pixels, as well as the temporal differences between this pixel and the previous and next corresponding pixels.
  • the first binary descriptors are integers
  • the step of determining if the entity which is presented to the vascular recognition system is a real biometric
  • characteristic or a spoof may involve a comparison between a plurality of first dynamic binary descriptors with one or more pluralities of reference dynamic binary descriptors.
  • said second binary descriptors may be compared with one or more reference binary descriptors.
  • the reference dynamic binary descriptors are representative of the manner in which some elements of texture change between successive images in one or more sequences of reference images, and, the one or more reference binary descriptors with which said second binary descriptors are compared, are each representative of texture of a reference image. It will be understood that the reference dynamic binary descriptors and the reference binary descriptors may be formed from images of valid biometric characteristics retrieved by the recognition system.
  • the reference dynamic binary descriptors to which the first dynamic binary descriptors are compared and reference binary descriptors with which said second binary descriptors are compared provide a measure of the authenticity of the entity presented to the recognition system. If the first dynamic binary descriptors are too dissimilar to all of the reference dynamic binary descriptors then it can be determined that the entity presented to the recognition system is a spoof. Likewise, if the second binary descriptors are too dissimilar to all of the reference binary
  • the recognition system is a spoof. Accordingly, in the present invention there may be a dissimilarity threshold set for each of the dynamic binary descriptors and binary descriptors; exceeding this dissimilarity threshold will indicate that the entity presented is a spoof.
  • the step of determining if the entity which is presented to the vascular recognition system is a real biometric characteristic or a spoof, based on the plurality of first dynamic binary descriptors and said second binary binary descriptors may comprise using at least one classifier to differentiate between real access and spoof attempts.
  • classifiers are known in the art.
  • the classifier may have been previously built or trained with data (both real-access and spoof attempts) and with an appropriately chosen machine learning algorithm such as, typically, a discriminative method or a generative method.
  • the classifier typically minimizes a loss function to determine the degree of similarity between the plurality of first dynamic binary
  • a second classifier will use a discriminative method to minimize a loss function to determine the degree of similarity between the second binary descriptors and the reference binary descriptors.
  • a discriminative method to minimize a loss function to determine the degree of similarity between the second binary descriptors and the reference binary descriptors.
  • the classifier can use a generative technique which maximizes a likelihood function (e.g. Expectation-Maximization of a statistical model) to determine the degree of similarity between the plurality of first dynamic binary descriptors and said plurality of reference dynamic binary
  • the classifier can use a generative technique which maximizes a likelihood function to determine the degree of similarity between the second binary descriptors and the reference binary
  • the output of this step is composed of a single highly- discriminative score that maps directly to the probability of an attack, given the perceived data (a priori information).
  • a first classifier might be used for computing a first score based on the first binary descriptors.
  • a second classifier might be used for computing a second score based on the second binary descriptors.
  • the output of the first and second and classifiers may be combined in order to compute a consolidated score and to differentiate between real-access and spoof attempts.
  • the spatial and temporal texture changes in images captured at a first wavelength are represented by a single combined binary descriptor.
  • the first classifier may then be used to compute a score and to differentiate between real-access and spoof attempts, based on this combined binary descriptor.
  • the score output of this first classifier may be combined with the output of a second classifier used to differentiate between real-access and spoof attempts, based on the second binary descriptor.
  • the discrimination then depends both on the output of the first classifier and on the output of the second classifier.
  • a first binary descriptor is used for representing variations of images over time at the first wavelength
  • a distinct, third binary descriptor is used for representing spatial variations of images at the first wavelength.
  • Two distinct classifiers might then be used for classifying the first and third binary descriptors.
  • descriptor-extraction modules ei to e 8 based on both types of algorithms are used in four different wavelength ranges w, (visible, near-infrared, sub-near infrared, and thermal spectrum LWIR); each set of descriptors generated by each algorithm in each wavelength is then input to a classifier ci to Cs .
  • the decision taken by each classifier (real or spoof; possibly associated with a probability) are then combined by a multi-spectral fusion system F.
  • FIG. 2 is a block diagram illustrating an exemplary method of vascular recognition according to the present invention.
  • the method of anti-spoofing in vascular recognition comprises the step of performing the multi-spectrum method illustrated in Figure 1, to determine if an entity presented to the vascular recognition system by a user is a real biometric characteristic or a spoof (30). In parallel to this step, a step of identifying the user is performed, based on NIR image only. Finally the step of recognising that the entity presented to the recognition system is valid, if it has been determined that the entity which is presented in a real biometric characteristic and if user has been identified (32).
  • FIG. 3 is a diagram illustrating the features of a vascular recognition system 40 according to the present invention.
  • the vascular recognition system 40 comprises, a light source 41 configured to
  • an image capturing means 42 which can capture an image and a sequence of images of the entity when the entity is illuminated by the light source to provide a sequence of images and at least one static image; and a processor 43.
  • the processor 43 is configured to compute a plurality of first binary descriptors representing the texture of at least one of said images in the sequence of images, and simultaneously representative of the manner in which some elements of image change between successive images in a sequence of images over the whole sequence, and to compute second binary descriptors for each of said at least one static images, and to determine if the entity which is presented to the vascular recognition system is a real biometric characteristic or a spoof, based on the plurality of first dynamic binary descriptors and said second binary descriptors.
  • the processor is configured to perform any of the steps mentioned in the above-described methods.
  • the invention may be used in the authentication of a person, in which case the biometric characteristic which is presented to the recognition system must match a single reference hand-palm which the recognition system has stored; or the invention may be used in an identification of a person, in which case the biometric characteristic presented to the recognition system must match one of a plurality of reference biometric characteristics which the
  • the first descriptors depending on temporal changes of patterns in the near infrared range, or in any range is used to determine the heartbeat of the person.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Human Computer Interaction (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Collating Specific Patterns (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

L'invention concerne un procédé anti-usurpation pour un système de reconnaissance vasculaire, le procédé comprenant les étapes consistant à: illuminer une entité qui est présentée au système de reconnaissance vasculaire, ceci avec une lumière qui présente des longueurs d'onde dans une première plage de longueurs d'onde uniquement (1 ); capturer une séquence d'images de l'entité au cours d'une période de temps lorsque l'entité est illuminée avec une lumière qui présente des longueurs d'onde dans une première plage de longueurs d'onde uniquement (2); calculer une pluralité de premiers descripteurs binaires représentatifs de la texture d'au moins une desdites images dans la séquence d'images (3) et de la manière dans laquelle certains éléments d'image de changent entre des images successives dans la séquence sur toute la séquence (4); illuminer l'entité qui est présentée au système de reconnaissance vasculaire avec une lumière qui présente des longueurs d'onde dans une seconde plage de longueurs d'onde uniquement (5), capturer une image de l'entité lorsque l'entité est illuminée par la lumière qui présente des longueurs d'onde dans une seconde plage de longueurs d'onde uniquement afin de fournir au moins une image statique (6); calculer des seconds descripteurs binaires pour chacune desdites au moins une images statiques (7); déterminer si l'entité qui est présentée au système de reconnaissance vasculaire est une usurpation en se fondant sur la pluralité des premiers descripteurs binaires et seconds descripteurs binaires dynamiques (8).
PCT/EP2014/067290 2014-08-13 2014-08-13 Procédé de détection d'une présentation falsifiée à un système de reconnaissance vasculaire WO2016023582A1 (fr)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/EP2014/067290 WO2016023582A1 (fr) 2014-08-13 2014-08-13 Procédé de détection d'une présentation falsifiée à un système de reconnaissance vasculaire
EP14752602.4A EP3180736A1 (fr) 2014-08-13 2014-08-13 Procédé de détection d'une présentation falsifiée à un système de reconnaissance vasculaire

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/EP2014/067290 WO2016023582A1 (fr) 2014-08-13 2014-08-13 Procédé de détection d'une présentation falsifiée à un système de reconnaissance vasculaire

Publications (1)

Publication Number Publication Date
WO2016023582A1 true WO2016023582A1 (fr) 2016-02-18

Family

ID=51357928

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/EP2014/067290 WO2016023582A1 (fr) 2014-08-13 2014-08-13 Procédé de détection d'une présentation falsifiée à un système de reconnaissance vasculaire

Country Status (2)

Country Link
EP (1) EP3180736A1 (fr)
WO (1) WO2016023582A1 (fr)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105975936A (zh) * 2016-05-04 2016-09-28 广东顺德中山大学卡内基梅隆大学国际联合研究院 一种纯光学指纹活体检测方法
CN111008550A (zh) * 2019-09-06 2020-04-14 上海芯灵科技有限公司 基于Multiple loss损失函数的指静脉验证身份的识别方法
WO2020205055A1 (fr) * 2019-03-29 2020-10-08 Alibaba Group Holding Limited Identification biométrique à l'aide d'images de main composites
CN112001240A (zh) * 2020-07-15 2020-11-27 浙江大华技术股份有限公司 活体检测方法、装置、计算机设备和存储介质
US20220398820A1 (en) * 2021-06-11 2022-12-15 University Of Southern California Multispectral biometrics system

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5737439A (en) * 1996-10-29 1998-04-07 Smarttouch, Llc. Anti-fraud biometric scanner that accurately detects blood flow
US20030016345A1 (en) * 2001-07-19 2003-01-23 Akio Nagasaka Finger identification apparatus
EP1353292A1 (fr) * 2002-04-12 2003-10-15 STMicroelectronics Limited Appareil et procédés de saisie biométrique
US20050271258A1 (en) * 2004-06-01 2005-12-08 Lumidigm, Inc. Multispectral imaging biometrics
US20060110015A1 (en) * 2003-04-04 2006-05-25 Lumidigm, Inc. Systems and methods for improved biometric feature definition
EP1805690A1 (fr) * 2004-10-22 2007-07-11 Koninklijke Philips Electronics N.V. Procede et appareil d'identification basee sur la biometrie

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5737439A (en) * 1996-10-29 1998-04-07 Smarttouch, Llc. Anti-fraud biometric scanner that accurately detects blood flow
US20030016345A1 (en) * 2001-07-19 2003-01-23 Akio Nagasaka Finger identification apparatus
EP1353292A1 (fr) * 2002-04-12 2003-10-15 STMicroelectronics Limited Appareil et procédés de saisie biométrique
US20060110015A1 (en) * 2003-04-04 2006-05-25 Lumidigm, Inc. Systems and methods for improved biometric feature definition
US20050271258A1 (en) * 2004-06-01 2005-12-08 Lumidigm, Inc. Multispectral imaging biometrics
EP1805690A1 (fr) * 2004-10-22 2007-07-11 Koninklijke Philips Electronics N.V. Procede et appareil d'identification basee sur la biometrie

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105975936A (zh) * 2016-05-04 2016-09-28 广东顺德中山大学卡内基梅隆大学国际联合研究院 一种纯光学指纹活体检测方法
WO2020205055A1 (fr) * 2019-03-29 2020-10-08 Alibaba Group Holding Limited Identification biométrique à l'aide d'images de main composites
US11495041B2 (en) 2019-03-29 2022-11-08 Jumio Corporation Biometric identification using composite hand images
US11854289B2 (en) 2019-03-29 2023-12-26 Jumio Corporation Biometric identification using composite hand images
CN111008550A (zh) * 2019-09-06 2020-04-14 上海芯灵科技有限公司 基于Multiple loss损失函数的指静脉验证身份的识别方法
CN112001240A (zh) * 2020-07-15 2020-11-27 浙江大华技术股份有限公司 活体检测方法、装置、计算机设备和存储介质
US20220398820A1 (en) * 2021-06-11 2022-12-15 University Of Southern California Multispectral biometrics system

Also Published As

Publication number Publication date
EP3180736A1 (fr) 2017-06-21

Similar Documents

Publication Publication Date Title
Chen et al. Attention-based two-stream convolutional networks for face spoofing detection
Hernandez-Ortega et al. Time analysis of pulse-based face anti-spoofing in visible and NIR
Zhang et al. Online joint palmprint and palmvein verification
Barra et al. A hand-based biometric system in visible light for mobile environments
Kashem et al. Face recognition system based on principal component analysis (PCA) with back propagation neural networks (BPNN)
KR20090087895A (ko) 생체인식정보의 추출과 대조를 위한 방법 및 장치
US20080298642A1 (en) Method and apparatus for extraction and matching of biometric detail
Connell et al. Fake iris detection using structured light
JP2009523265A (ja) 画像中の虹彩の特徴を抽出する方法
JP2007188504A (ja) 画像中の画素強度をフィルタリングする方法
WO2016023582A1 (fr) Procédé de détection d'une présentation falsifiée à un système de reconnaissance vasculaire
Raghavendra et al. Hand dorsal vein recognition: Sensor, algorithms and evaluation
Spinoulas et al. Multi-modal fingerprint presentation attack detection: Evaluation on a new dataset
Seal et al. Minutiae based thermal face recognition using blood perfusion data
Anthony et al. A review of face anti-spoofing methods for face recognition systems
Gomez-Barrero et al. Towards multi-modal finger presentation attack detection
Sujatha et al. Biometric authentication system with hand vein features using morphological processing
Donida Labati et al. A scheme for fingerphoto recognition in smartphones
Song et al. Face liveness detection based on joint analysis of rgb and near-infrared image of faces
Estacio et al. A rotation invariant algorithm for bimodal hand vein recognition system
Zayed et al. A comprehensive survey on finger vein biometric
Venkatesh et al. A new multi-spectral iris acquisition sensor for biometric verification and presentation attack detection
Schuiki et al. Extensive threat analysis of vein attack databases and attack detection by fusion of comparison scores
Dronky et al. A review on iris liveness detection techniques
Bartuzi et al. Multispectral hand features for secure biometric authentication systems

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 14752602

Country of ref document: EP

Kind code of ref document: A1

REEP Request for entry into the european phase

Ref document number: 2014752602

Country of ref document: EP

WWE Wipo information: entry into national phase

Ref document number: 2014752602

Country of ref document: EP

NENP Non-entry into the national phase

Ref country code: DE