LV15413B - Method for detection of coloured counterfeits - Google Patents

Abstract

A method for detecting colored counterfeits by authenticating a test object, characterized in that illuminating the test object with a single discrete spectral line in the sensitivity range of the light source produces a monochromatic image at a fixed wavelength, then comparing this image with a reference monochromatic image of the corresponding wavelength and after adjusting the size of both images and normalizing the brightness, using the authentication criterion α> / A-BV determined in the analysis, where Ai and Bi - the signal values of the corresponding pixels or groups of pixels of the reference and object images, the object is considered to be falsified. The method according to claim 1, wherein the object is illuminated with two spectral lines λ1 and λ2 in the sensitivity range of the camera and to determine its authenticity further compares the direct and / or inverse relations of the two monochromatic images using the analysis authentication criteria β and γ: / ( AVAJHBVBJ / <β and / or ΛΑλ ^ λΛ- ^ λ ^ Βλ!) / <Γ. The method according to claim 1, wherein the object is illuminated with three spectral lines λ1, λ2 and λ3 in the sensitivity range of the light source and to further authenticate the object and reference images formed by the first and second and / or first and third, and / or the direct or inverse relationship between the images of the second and third spectral lines. A method according to any one of claims 1 to 3, wherein instead of normalizing the brightness of the image, the brightness and / or contrast of the image of the object is gradually reduced by a certain step and all newly acquired images Cj by pixels or groups of pixels are compared with the corresponding image of the reference original. for false, if at no time is the condition satisfied: / Ai-Cji / <σ, where σ is the critical cut-off value of the pixels or groups of pixels determined by the analysis. A method according to any one of claims 1 to 4, wherein the object is further irradiated in the ultraviolet spectral range and the fluorescent color images F1 and F2 of the reference and the object are compared by applying an authentication criterion for signal differences of pixels or groups of pixels σ> / F1-F2. /. A method according to any one of claims 1 to 5, wherein a digital RGB camera of the smartphone is used to display the object. A method according to any one of claims 1 to 6, wherein scattered laser radiation is used to illuminate the object. A method according to any one of claims 1 to 7, wherein the object to be inspected is a possibly counterfeit banknote. A method according to any one of claims 1 to 7, wherein the object to be checked is a possibly forged document, such as an identity card. A method according to any one of claims 1 to 7, wherein the object to be examined is a possibly forged work of art, such as a painting.

Description

DESCRIPTION OF THE INVENTION

The invention relates to display technologies, in particular to banknotes, documents, works of art, etc. detection of colored counterfeits using monochromatic images in narrow (<1 nm) spectral bands.

Analysis of the prior art

[002] To distinguish genuine from counterfeit banknotes, there are various commercially available counterfeit detectors (see, for example, https://www.accubanker.com/collections/counterfeit-detectors, https://fitsmallbusiness.com/how-to- detect-counterfeit-money /). Their operating principles are most often based on the acquisition of images of banknotes using transmitted, reflected or fluorescent radiation, and the comparison of these images with images of authentic banknotes. Special watermark fluorescent lamps, ultraviolet fluorescent lamps, etc. are often used for this purpose. specific light sources. In addition, such devices use magnetic ink scanners and other accessories that provide a real distinction between banknotes and counterfeits. Similar principles are used to detect forged documents and works of art (https://en.wikipedia.org/wiki/ldentity_document_forgery; http://orionanalytical.com).

[003] There are a number of patented solutions for detecting colored counterfeits with imaging technologies. For example, US2017287252 (A1) proposes a method of processing images of special areas of interest (Rol) by additionally "smearing" them and applying a threshold algorithm. CN106846602 (A), on the other hand, offers to use the database of accumulated counterfeit banknote image features in the image analysis process. CN 106934914 (A) offers the analysis of transmitted infrared images, while KR20170044502 (A) offers the selective analysis of transmitted light signals of light emitting diodes (LEDs) of different spectral composition. Several patent applications propose the use of mobile phone cameras with special light sources (eg CN106850917 (A), WO2017083841 (A1)) for the detection of counterfeit notes. Imaging technologies (JP2002178609 (A)) are also used to detect forged documents.

Due to the fact that counterfeit technologies are rapidly advancing, adapting to the already known counterfeit detection solutions, it is necessary to constantly supplement and update the range of such solutions. The perspective in this respect is the use of monochromatic spectral images. Monochromatic images in very narrow (<1nm) spectral bands can be obtained, for example, by separating them from a single digital RGB image data array at one, two or three fixed light wavelengths (WO2013135311 (A1); WO2015071786 (A1)). Because the reflected images of any colored object at different fixed wavelengths are specifically different, such monochromatic spectral images can be used to identify surface color pigments and determine their distribution. For example, it has been successfully implemented for mapping the distribution of human skin pigments (chromophores) using a smartphone and an attached three spectral line laser illuminator (WO 2017/012675 A1; J. Spigulis et al., J. Biomed. Op., 22 (9), 091508 (2017)). A similar approach is useful for the authenticity analysis of banknotes, documents and works of art, in which the analysis of monochromatic spectral images has not been used so far.

Purpose and essence of the invention

The aim of the proposed method is to ensure rapid detection of colored counterfeits by using monochromatic images of an object and a reference (original) taken with a digital RGB camera for illumination in the illumination of one, two or three discrete spectral lines.

[006] In addition, the authenticity of the object under test is determined by analyzing its monochromatic spectral images at fixed wavelengths. For example, a digital RGB camera is used to capture the subject and the subject is illuminated with one, two, or three discrete spectral lines in the camera's sensitivity range. From the obtained RGB image data array, one, two or three monochromatic images of the respective spectral lines are extracted, for example, using the method described in WO2013135311 (A1).

[007] Brief description of the drawings:

Fig. 1 is a block diagram of the proposed method in one of its variants.

Fig. 2 illustrates the difference between monochrome images and a traditional color (RGB) image for a genuine and counterfeit banknote.

Fig. 3 illustrated the possibility of the method to increase the selectivity against counterfeiting by using monochromatic image ratios.

[008] In the first step of one variant of the proposed method (Fig. 1) n monochromatic images (n = 1, 2 or 3) of the reference and test object are taken at the respective selected wavelengths, then the dimensions and brightness of the compared images are adjusted (normalized). , then the pixel or groups of pixels are determined to meet the above authenticity criteria and as a result it is concluded whether the object is authentic or counterfeit.

FIG. Compared to a smartphone camera, genuine (a) and counterfeit (b) 20 EUR banknotes in single fragment color (RGB) and monochrome images at 448 nm and 532 nm were compared. Visually, there are several differences between the images of the counterfeit banknotes and the images of the aristas banknotes.

FIG. Comparison of traditional (a) and counterfeit (b) traditional color (RGB) images of the same fragment (Fig. 2) of the EUR 20 banknote with spectral ratio images obtained by dividing one monochromatic image by another: 448 nm / 532 nm and 659 nm / 448 nm. Visually, there are more significant differences between the two images than in Figure 2, which indicates that the relationship images are better suited for detecting counterfeits.

The obtained monochromatic image of one spectral line is further used to determine the authenticity of the object compared to the monochromatic image of the same wavelength reference (original). When an object is illuminated with two spectral lines, the pixel value ratios of the monochromatic images of the two wavelengths are additionally calculated to determine its authenticity, and then the division images thus formed are compared with each other. When an object is illuminated simultaneously with three spectral lines, it is additionally compared for images to be authenticated by direct or inverse ratios of the first and second, first and third, and / or second and third wavelength images.

When comparing monochromatic object and reference images, the authentication criterion α> / Ai-Bi / determined in the analysis is used to detect counterfeiting, where Ai and Bi are the signal values of the corresponding pixels or groups of pixels of the reference and object images. If the object is illuminated simultaneously with two spectral lines λ1 and λ2 in the sensitivity range of the camera, its direct and / or inverse relationship images of the two monochromatic images shall be additionally compared using the authentication criteria β and γ defined in the analysis:

/ (Αλΐ / Αλ ₂ ) ί- (Βλΐ / Βλ2) ί / <β and / or / (Αλ2 / Αλΐ) ί- (Βλ2 / Βλΐ) ί / <γ.

[013] Alternatively, the method does not normalize the images, but gradually degrades the brightness and / or contrast of the monochrome image of the object by a certain step, and compares all newly acquired images Cj by pixels or groups of pixels with the corresponding reference monochrome image A. The object is considered counterfeit. at the same time the condition: / A-Cji / <σ is not satisfied for the whole image or a part of it, where σ is the critical limit value determined by the analysis.

[014] In the implementation of the method, a digital RGB camera of a smartphone can be used to display the object and scattered one or more laser radiation to illuminate the object. In this case, distortions of the laser spectra are possible in monochromatic images, which can be prevented, for example, by mechanically oscillating the scattering optical element with a period shorter than the exposure time of the image capture (PCT / EP2017 / 063565, 2017).

[015] In order to increase the recognition efficiency of counterfeit objects, the object can be additionally irradiated in the ultraviolet range by inducing specific fluorescence of its fragments and further comparing the fluorescent images F1 and F2 of the reference / original and test object to meet the criterion / F1ī-F2 </ < where kritthe critical value of the pixel signal difference is determined before the analysis.

[016] In all of the above cases, the thresholds set out in the analysis refer to the scattering of maximum values for authentic / original objects, such as genuine banknotes of the same denomination produced at different locations and times.

[017] The suspected counterfeit object to be checked by this method may be a banknote (eg a EUR banknote), a document (eg an identity card) or a work of art (eg a painting).

The proposed method includes the following steps:

1) obtain monochrome reference images and / or their ratios from the original / authentic object at all planned light wavelengths and, if necessary, also reference fluorescent images under ultraviolet light;

2) obtain one, two or three monochromatic images of the object to be examined and / or their relations at the same fixed wavelengths and, if necessary, also a fluorescent image of the object under ultraviolet light;

3) perform mutual alignment of the captured images, ensuring compatible and reference images of the same scale at each fixed wavelength, as well as normalized image brightness in a certain area of the object;

4) determine and compare the signal values of all or separately selected pixels or groups of pixels of the matched object and reference monochrome images, checking compliance with the authentication criteria:

α> / Ai-Bi / and / or / (Αλι / Αλ2) ί— (Βλι / Βλ ₂ ) ί / <β, and / or / (Αλ2 / Αλι) - (Βλ2 / Βλι) Α <γ, and / or / F1i-F2i / <φ, and in case of non-compliance the object is recognized as counterfeit; 5) as an alternative to step (4) or in addition, the brightness and / or contrast of the monochrome image of the object is gradually reduced by a predetermined step and each image Cj thus obtained is compared with the corresponding monochrome image A of the reference original and the object is found to be counterfeit. the condition / Ar — Cji / <σ, where σ is a predetermined critical limit, is not satisfied for all or part of the object.

Claims (10)

A method for detecting colored counterfeits by authenticating a test object, characterized in that illuminating the test object with a single discrete spectral line in the sensitivity range of the light source produces a monochromatic image at a fixed wavelength, then comparing this image with a reference monochromatic image of the corresponding wavelength and after adjusting the size of both images and normalizing the brightness, using the authentication criterion α> / Ai — Bi / determined in the analysis, where A and Bi are the signal values of the corresponding pixels or groups of pixels of the reference and object images, it is concluded that the object is possible. The method according to claim 1, wherein the object is illuminated with two spectral lines λ1 and λ2 in the sensitivity range of the camera and to determine its authenticity further compares the direct and / or inverse relations of the two monochromatic images using the analysis authentication criteria β and γ: / ( Αλΐ / Αλ2) ί- (Βλΐ / Βλ2) ί / <β and / or / (Αλ2 / Αλΐ) ι- (Βλ2 / Βλΐ) ί / <γ. The method of claim 1, wherein the object is illuminated with three spectral lines λ1, λ2 and λ3 in the sensitivity range of the light source and to further authenticate the object and reference images formed by the first and second and / or first and third and / or second and a direct or inverse relationship between the images of the third spectral line. A method according to any one of claims 1 to 3, wherein instead of normalizing the brightness of the image, the brightness and / or contrast of the image of the object is gradually reduced by a certain step and all newly acquired images Cj by pixels or groups of pixels are compared with the corresponding image of the reference original. for false, if at no time is the condition satisfied: / Ai — Cj · / <σ, where σ is the critical cut-off value for the pixels or groups of pixels determined by the analysis. A method according to any one of claims 1 to 4, wherein the object is further irradiated in the ultraviolet spectral range and the reference and fluorescent color images F1 and F2 of the object are compared by applying an authentication criterion for signal differences of pixels or groups of pixels σ> / F1i-F2i /. A method according to any one of claims 1 to 5, wherein a digital RGB camera of the smartphone is used to display the object. A method according to any one of claims 1 to 6, wherein scattered laser radiation is used to illuminate the object. A method according to any one of claims 1 to 7, wherein the object to be inspected is a possibly counterfeit banknote. A method according to any one of claims 1 to 7, wherein the object to be checked is a possibly forged document, such as an identity card. A method according to any one of claims 1 to 7, wherein the object to be examined is a possibly forged work of art, such as a painting.