US 2026/0212643 A1Published application

Authentication of physical objects based on reliable parts of copy detection patterns

OwnerSvyatoslav Voloshynovskiy, Joakim Tutt +3inventors

InventorsSvyatoslav Voloshynovskiy, Joakim Tutt, Roman Chaban, Olga Taran, Taras Holotyak

Life of this patent

Estimating the term…

Loading patent…

Abstract

A method for authentication of physical objects, by sampling and verifying of a probe image of a copy detection pattern representing an object to be authenticated, including, at a training stage, acquiring a set of reference templates, acquiring a set of training images, pre-processing the training images, and, at an authentication stage, acquiring a probe image of a copy-detection pattern representing an object to be authenticated, performing pre-processing of the probe image, and producing a decision about authenticity of the physical object. The method also includes, at the training stage, producing estimates of reference templates, determining a pattern reliability, and, at the authentication stage, producing an estimate of probe reference template, applying the determined pattern reliability, by use of a mask reflecting the pattern reliability, for a weighted score computation, the step of producing a decision about authenticity being based on a score produced by the weighted score computation.

A method for authentication of physical objects, by sampling and verifying of a probe image of a copy detection pattern (CDP) representing an object to be authenticated, the method comprising,

at a training stage, the steps of:
acquiring a set of reference templates
{
t
n
}
n
=
1
N
[
[
(
2
⁢
5
)
]
]
adapted for production of corresponding original physical objects,
acquiring a set of training images
{
x
n
}
n
=
1
N
[
[
(
1
⁢
0
)
]
]
of copy detection patterns (CDP) representing original objects produced by use of said set of reference templates
{
t
n
}
n
=
1
N
[
[
(
2
⁢
5
)
]
]
,
performing pre-processing of the training images in the set of training images
{
x
n
}
n
=
1
N
by alignment of the training images with the reference templates, and,
at an authentication stage, the steps of:
acquiring a probe image y of a copy detection pattern (CDP) representing an object to be authenticated,
performing pre-processing of the probe image by alignment of probe image y with the reference template associated with the corresponding original object, and
producing a decision about authenticity of the physical object represented by the probe image y,
wherein the method further comprises:
at said training stage, the steps of:
producing a set of estimates of reference templates
{
t
˜
n
}
n
=
1
N
[
[
(
4
⁢
0
)
]
]
from a set of pre-processed training images representing original objects,
determining a pattern reliability by measuring the level of correspondence between pre-defined regions of the copy detection pattern (CDP) of at least part of the estimates of reference templates within said set of estimates of reference templates
{
t
˜
n
}
n
=
1
N
[
[
(
4
⁢
0
)
]
]
and corresponding reference templates within said set of reference templates
{
t
n
}
n
=
1
N
[
[
(
2
⁢
5
)
]
]
,
and,
at said authentication stage, the steps of:
producing an estimate of probe reference template {tilde over (t)} from said pre-processed probe image,
applying the determined pattern reliability, by use of a mask m reflecting said pattern reliability, in order to weigh regions of the copy detection pattern (CDP) according to its reliability and to allow its weighted use for a score computation between the estimate of probe reference template {tilde over (t)} corresponding to probe image y and a reference template corresponding to an original object,
wherein the step of producing a decision about authenticity of the physical object represented by the probe image y is based on at least a score s produced by said weighted score computation, said score s allowing to produce a decision d on authenticity of the physical object.

133 more elements

Facts

Application

19/137,986Filed 30 Nov 2023

Priority

EP 22212904.112 Dec 2022

Published

23 Jul 2026

Term ends

Not estimatedThe source records don’t support an estimate.

Classification28 codes

G06V 10/761
Image or video pattern matching; Proximity measures in feature spaces: proximity, similarity or dissimilarity measures
G06V 20/80
Scenes; Scene-specific elements: recognising image objects characterised by unique random patterns
G03G 21/046
Preventing copies being made of an original: by discriminating a special original, e.g. a bank note
G06F 18/211
Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation: selection of the most…

Abstract

Claims17