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
  • G06V40/1353—Extracting features related to minutiae or pores
• • 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

Definitions

• the invention relates generally to techniques related to the automated authentication of a set of characteristic points, with a very high degree of reliability.
• the invention relates, according to a first aspect, to an authentication method implementing a cloud of points, comprising at least the operation of identifying, in this cloud of points, a set of characteristic points that we will call subsequently minutiae, in number greater than 3.
• biometrics such as fingerprints
• a recognition of the celestial bodies makes it possible to position oneself on the globe, as well as to orient oneself.
• the representation of a cloud of points is sensitive to various transformations and in particular translations, rotations and homotheties. These transformations come from changes in the position of the capture means (fingerprint reader, camera, camera, etc.) during the capture. Indeed, two consecutive catches will never be identical because the capture device will never be perfectly in the same place, the subject or the target of the capture will never be exactly in the same place either, to this we must add the environmental parameters that can alter the catch.
• the purpose of the present invention is to propose an authentication method that does not require the location of the point cloud in an orthonormal frame, and in which the representation of the point cloud is insensitive to at least one of these transformations, so that it can be implemented on weak computers.
• These authentication methods are based on a comparison of a candidate to a reference.
• the first step will therefore always be a so-called enlistment step, consisting of a registration of the value which will serve as a reference thereafter. This particular step must be done, as far as possible, in good security conditions so that it can subsequently be trusted. Enlistment is divided into
• At least one storage step (usually in non-volatile memory)
• the second step is the authentication step itself. This step consists of a comparison of a candidate value and the reference value recorded during the enrollment.
• At least one processing step • At least one storage step (usually in volatile memory)
• the authentication method according to the invention proposes a solution for automating and embedding the processing and comparison steps in devices that do not have large resources, both in memory and computing power.
• the authentication method of the invention is essentially characterized in that it further comprises the operations of:
• non-flat triangle a triangle whose three vertices are not aligned.
• the scatterplot can be represented by the list of triplets formed from the three information obtained during the two preceding steps, for each selected triangle.
• the non-flat triangles are selected according to the Delaunay triangulation method.
• a device according to the present invention further comprises:
• automatic recognition means for identifying, in the two-dimensional image, a two-dimensional distribution of points, each of which corresponds to the location of a point characteristic of the point cloud;
• programmed calculation means to select, in this distribution of characteristic points, a plurality of non-flat triangles, each of which has three points of the distribution at its vertices, to associate with each selected triangle information on the diameter of the circumscribed circle, and to store it defining, for each selected triangle ABC, two pieces of information respectively representative of two ratios involving the three angles of the triangle, taken two by two, and one nonvolatile memory for storing, as a representation of said imprint, a list of triplets formed from the three information obtained in the two previous steps, for each selected triangle.
• the device will also have a volatile memory for temporarily storing triplets or triplet lists during the various calculations.
• the programmed calculation means are furthermore programmed:
• NPC candidate point cloud
• NPR reference point cloud
• an NPC triplet (Anglel-NPC, Angle2-NPC, Diam-NPC) will be retained only if there is an NPR triplet (Anglel-NPR, Angle2-NPR, Diam-NPR), for which: Anglel-NPC and Anglel-NPR vary in proportions below a threshold Sl Angle2-NPC and Angle2-NPR vary in proportions below an S2 threshold
• Diam-NPC and Diam-NPR vary in proportions below an S3 threshold to assimilate, or not, the cloud of candidate points to the cloud of reference points, according to whether the number of previously counted triplets of NPC, relative to the first number of triplets, represents or not a proportion at least equal to a threshold Threshold-Acceptance determined.
• FIG. 1 represents a set of characteristic points (minutiae) coming from a point cloud not shown;
• FIG. 2 represents an example of a network of non-flat triangles according to the invention applied to the minutiae of FIG. 1;
• FIG. 4 represents the triangles highlighted in FIG. 4, with their respective circumscribed circles;
• FIG. 5 reduces the cloud of points from which the minutiae represented in FIG. 1 are drawn, to the three triangles highlighted in FIG. 3, and to their circumscribed circles drawn in FIG. 4;
• Figure 6 shows an enlarged view of a fingerprint
• FIG. 7 represents the minutiae resulting from the fingerprint of FIG. 6, these minutiae being connected between they form non-flat triangles, and each of these triangles has its circumscribed circumscribed circle;
• FIG. 8 shows the list of triplets resulting from the triangular data and circles of FIG. 7. This triplet list forms a signature of the fingerprint of FIG.
• the invention relates to an authentication method implementing a cloud of points.
• an authentication method of this type comprises, in its acquisition phase, an identification operation, consisting of identifying, in the cloud of points to be acquired, a set of characteristic points (called minutiae in the present document), generally in the order of a few tens.
• this acquisition phase which leads to obtaining a representation of the point cloud, further comprises assimilation operations, selection, association, definition, and representation.
• the assimilation operation consists in assimilating the set of previously identified minutiae in the imprint to a two-dimensional distribution of points such as A to K.
• Figure 1 illustrates the assimilation of all the characteristic points from a cloud of points, to a two-dimensional representation of these minutiae.
• the minutiae are conventionally constituted by line crossings.
• the brightest bodies are often used as characteristic points of a portion of the sky.
• the characteristic points will be generally isolated points, so as not to confuse them with their neighbors. Some groups of points may be chosen because of a particularly identifiable arrangement. All these point selection methods required that the scatterplot be sufficiently supplied so that the number of characteristic points is sufficient.
• any points can become characteristic points because the "choice" of points answering quality criteria is replaced by the choice of triangles allowing the calculations.
• the selection operation consists in selecting, in this distribution of points, a plurality of non-flat triangles, each of which has three points of the two-dimensional distribution of points as vertices.
• the Delaunay triangulation technique gives very good results, but many other methods can be implemented, as long as they exclude flat triangles.
• a method can consist of a refusal of any triangle sharing a vertex with a triangle already registered, and flat triangles.
• Another method may consist of an exclusion of flat triangles, and triangles inscribed partially or in their entirety, in triangles already recorded. Another method may be to exclude flat triangles and triangles containing minutiae in addition to their three vertices.
• the selection can for example be done by excluding only the flat triangles.
• Figure 2 illustrates the selection of triangles from the points of Figure 1, according to the Delaunay method.
• the association operation consists in associating with each selected triangle a DIAM information on the diameter of the circumscribed circle.
• Figure 4 illustrates the association, with three particular triangles identified in Figure 3 of their respective circumscribed circles.
• this operation can associate to the triangle ABC, the square of the diameter of the circumscribed circle CCABC, to the triangle CDE, the square of the diameter of the circumscribed circle CCCDE, to the triangle FGH, the square of the diameter of the circumscribed circle CCFGH.
• the definition operation consists in defining, for each selected triangle ABC, two pieces of information (Anglel and Angle2) respectively representative of two ratios involving the three angles of the triangle, taken two by two.
• TétaA is the angle inside the triangle, formed in A by the intersection of the segments [BA] and [CA] - TétaB is the internal angle to the triangle, formed in B by the intersection of segments [AB] and [CB ]
• TétaC is the angle inside the triangle, formed in C by the intersection of segments [BC] and [AC]
• a particular embodiment of the invention consists of its application to the world of biometrics.
• biometrics particularly needs fast and simple means to authenticate a person with a high level of reliability.
• Authentication based on biometrics begins with a capture step, and an extraction step.
• Capture is the step that consists of a reading of the biometric data at a given moment. In the case of the majority of the biometrics, this step is done by the capture of an image, this is the case, among others, for the fingerprints, the facial, retinal, iris, and good d 'other.
• this capture can take a variety of forms, such as recording for voice recognition, video recording for gait recognition, combined motion and pressure recording for fingerprint recognition.
• the extraction step makes it possible to isolate characteristic points in this capture.
• these characteristic points are inter alia line crossings, end lines, or islands in a line.
• the result of the extraction step, applied to the fingerprint capture of FIG. 6, gives a set of points comparable to that of FIG.
• a candidate fingerprint will for example be sent to a portable computing unit, and processed as previously described to represent it by a series of triplets. ⁇ 143; 62; 68 ⁇ ⁇ 155; 17; 76 ⁇ ⁇ 102; 15; 6 ⁇ ⁇ 210; 54; 88 ⁇ ⁇ 19; 91; 42 ⁇ ⁇ 70; 23; 86 ⁇ ⁇ 85; 82; 27 ⁇ ⁇ 28; 85; 95 ⁇ ⁇ 181; 29; 85 ⁇ ⁇ 394; 27; 50 ⁇ ⁇ 327; 16; 33 ⁇ ⁇ 192; 55; 14 ⁇
• the set of characteristic points which made it possible to obtain this list of triplets could come, not from the fingerprint directly, but from a representation of it made in any form, provided that it is It is possible for the portable computing unit to make a two-dimensional representation, in order to calculate the triplets.
• triplets candidates obtained, this list will be stored in volatile memory of the portable computing unit.
• the authentication system is, for example, set as follows:
• Acceptance threshold 60%: 60% of the triplets of the reference fingerprint must be found in the candidate fingerprint.
• Sl 5: the "Anglel” values of the two triplets compared must vary by less than 5.
• S2 5: the "Angle2" values of the two triplets compared must vary by less than 5.
• the portable computing unit will compare the triplets ⁇ 81; 84; 29 ⁇ and ⁇ 85; 82; 27 ⁇ : 81 and 85 range from 4, which is much lower than or equal to S3 84 and 82 range from 2, which is well below or equal to S2 29 and 27 range from 2, which is well below or equal to sl
• the counter contains the value seven, for a reference footprint consisting of 10 triplets, or 70% of the reference footprint was found in the candidate footprint. This threshold is greater than the Threshold-acceptance value set at 60%.
• the candidate fingerprint is therefore authenticated as belonging to the cardholder.
• Another particular embodiment of the invention consists in its application to the world of astronomy. Indeed, the invention, once embedded in a mobile phone with a camera, will allow, for example, the following application. It is easy to treat according to the invention, all or part of the map of the stars visible at night, from a particular territory, or a larger area of the globe. Once this information is converted into a triplet list, this map can be considered as reference data.
• the application would treat this data to obtain a series of triplets considered as candidates.
• the application would then seek to authenticate the part of the sky of the general reference map to which the "candidate" portion corresponds. This would for example identify where the picture was taken, in case the user is lost.
• Another use may be to identify the stars or groupings of stars (constellations, ...) that the user has in front of him, if these data have been added to the general reference map.
• a particular use of the invention consists in its application to point clouds already stored in a form other than that according to the invention.
• the invention will convert the point cloud into triplets.
• the solution according to the invention makes it possible to very greatly improve the performance of the comparison step (match in English) compared to the solutions known in the state of the art and the technique.
• the present invention also makes it possible to authenticate with sets of feature points less provided than in the prior art. Indeed, the fact that the present invention uses triangles from the characteristic points, and not the characteristic points directly, allows, by carefully choosing the triangles selection method, to obtain a number of triangles much higher than the number points. This feature allows the present invention to authenticate sets of points hitherto impossible to authenticate because of their low number of characteristic points.

Landscapes

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

Priority Applications (1)

Applications Claiming Priority (3)

Publications (1)

Family

ID=38441599

Family Applications (2)

Family Applications Before (1)

Country Status (4)

Families Citing this family (6)

Family Cites Families (4)

• 2007
  • 2007-05-11 EP EP07301034A patent/EP1990757A1/de not_active Withdrawn
• 2008
  • 2008-04-14 US US12/598,639 patent/US20100135538A1/en not_active Abandoned
  • 2008-04-14 WO PCT/EP2008/054492 patent/WO2008141872A1/fr active Application Filing
  • 2008-04-14 EP EP08749554A patent/EP2147394A1/de not_active Ceased
• 2009
  • 2009-10-30 ZA ZA200907648A patent/ZA200907648B/xx unknown

Non-Patent Citations (3)

Also Published As

Similar Documents

Legal Events