EA046392B1 - METHOD FOR DETECTING SIGNS OF Falsification of Documents and Banknotes from IMAGES OBTAINED IN UV - Google Patents

Description

The invention relates to the field of authenticating images of various types of documents.

Various methods are known from the prior art for detecting signs of falsification of documents and banknotes from images obtained in UV.

For example, in [1] (see Kumar R, Kumar V, Sharma V. Discrimination of Various Paper Types Using Diffuse Reflectance Ultraviolet-Visible Near-Infrared (UV-Vis-NIR) Spectroscopy: Forensic Application to Questioned Documents. Applied Spectroscopy . 2015;69(6):714-720. doi:10.1366/14-07663) analyze the reflection and absorption spectra of paper of different types of radiation with different wavelengths (250-260 nm, 305-315 nm, 385-395 nm, 575-585 nm ). This approach is focused only on checking the properties of paper and does not analyze the luminosity of luminescent elements.

A similar method is described in [2] (see J. Mantec'on D. Bret'on J.F. Garc'ia A. Vila, N. Ferrer. Development of a fast and nondestructive procedure for characterizing and distinguishing original and fake euro notes. 2006), here an analysis of the absorption properties of €50 banknotes is carried out. The analysis is carried out only in the infrared range and can guarantee the absence of falsification.

Verifying the authenticity of banknotes using IR radiation of different wavelengths is disclosed in the information source [3] (see patent RU 2274900 C2, published 04/20/2006). This approach is focused on a specific feature - the presence of a special image of two parts, each of which is made with its own paint; the shades of the paints are the same, but the reflectance coefficients in the IR range are different for them. This approach also does not cover the protective features that appear when using UV illumination.

The objective of the claimed invention is to eliminate the disadvantages of the prior art. The technical result consists in providing a method for detecting signs of falsification of documents and banknotes from images obtained in UV, which allows for automatic verification of the correspondence of luminescence in the input image assigned to a given type of document and identifying signs of document authenticity and ensuring increased accuracy in determining the presence/absence of falsifications .

The stated problem is solved, and the stated technical result is achieved through the claimed method of detecting signs of falsification of documents and banknotes from images obtained in UV.

The claimed method for detecting signs of falsification of documents and banknotes from images obtained in UV consists of sequential analysis of UV luminescence in the input image, while considering images in the RGB color space containing a document obtained using UV illumination, and at the first stage producing analysis of the image for the presence of luminescence in those parts of the document where it is not provided; at the second stage of analysis, luminescent protective fibers from various materials embedded in the surface layer of paper are detected in the input image; at the third stage, the location of all luminescent areas of the document, except for fluorescent fibers, is analyzed , relative to each other, while for analysis they first construct a proper description of the location of the luminescent elements on the document, then check whether the input image corresponds to this description, at the fourth stage, a detailed analysis of the key signs of the authenticity of elements luminescent in UV is carried out for signs of falsification, at the fifth stage the shapes of the luminescent elements are clarified, a description of the shape of the luminescent element is first constructed using a feature vector, then the same feature vector is calculated on the input image in the zone where this element should be located, at the sixth stage of the algorithm the color of the luminescent elements is analyzed and compared with the one provided for this document type.

The figures show:

fig. 1: block diagram of the algorithm for analyzing images of documents and banknotes obtained in UV;

fig. 2: examples of the third page of genuine and fabricated passports of a citizen of the Russian Federation in ultraviolet light;

fig. 3: examples of luminescent security fibers;

fig. 4: image of a foreign passport of Uzbekistan in UV and a mask of luminescent elements for this type of document;

fig. 5: an example of a protective luminescent wave on a general passport of the Russian Federation and a mask of a single element of this structure;

fig. 6: example of superimposing a luminescent wave on the photo area on a general passport of the Russian Federation;

fig. 7: example of a fragment for analyzing the shape in the image of an Armenian passport;

fig. 8: example of different colors of luminescent ink in images of general passports of the Russian Federation.

To verify the authenticity of documents, small-sized full-page document scanners are used. In them, the work area is illuminated using LEDs of various types. The use of various LEDs allows you to obtain images of documents traditionally in the following

-

Claims (6)

blowing ranges: visible optical, infrared (dominant wavelength of the IR illuminator is 870 nm), ultraviolet (dominant wavelength of the UV illuminator is 365 nm). Such images allow for a detailed analysis of the verifiable features of the authenticity of documents, including in a fully automatic mode. UV imaging is one of the key elements when verifying the authenticity of a document using a multispectral scanner. Let an image of a document obtained from such a scanner using UV illumination be given. It is necessary to automatically check the correspondence of the luminescence in the input image assigned to a given type of document and identify signs of the authenticity of the document. Images in the RGB color space containing a document obtained using UV illumination are considered. The algorithm consists of a sequential analysis of UV luminescence in the input image according to various parameters (see Fig. 1). As a rule, paper used for the manufacture of security printing products absorbs UV radiation. While ordinary office paper reflects such radiation (see Fig. 2). Therefore, at the first stage, the image is analyzed for the presence of false luminescence, i.e. luminescence in those parts of the document where it is not provided. If such luminescence was detected, the system will issue a fault. The second stage of the analysis consists of detecting luminescent protective fibers from various materials embedded in the surface layer of the paper in the input image (see Fig. 3). Such fibers have certain properties: they are thin, short, multi-colored threads, randomly located on the document image. The number of fibers found on the document must be greater than the threshold specified for this type of document. If the fibers were detected in insufficient quantities, the system will also issue a fault. At the third stage, the location of all luminescent areas of the document, except for fluorescent fibers, relative to each other is analyzed. Here, the luminosity recorded for a document in a certain area of the document is perceived as a light object on a dark background. For analysis, some ideal description of the arrangement of luminescent elements on the document is first constructed (see Fig. 4), then it is checked whether the input image matches the given description. If the location of the luminescent elements in the input image does not correspond to the ideal one, the system will fail. This stage provides two approaches to implementation: description of the template constructed using machine learning methods; comparison with a reference (model) image. The fourth stage is a detailed analysis of the key features of the authenticity of elements luminescent in UV for the presence of signs of falsification. An example of such a feature is a protective periodic wave on a general passport of the Russian Federation. When analyzing this element, it is necessary to control the presence of a wave, its structure (see Fig. 5), as well as its integrity. When producing a document, a wave is applied over a photograph (see Fig. 6), therefore, any damage to the luminosity in the wave area may indicate that the photograph has been re-taped for the purpose of falsification. If such inconsistencies are detected, the system will issue a refusal. The fifth stage is to clarify the shape of the luminescent elements. A description of the shape of the luminosity element is first constructed using a feature vector. Next, the same vector of features is calculated on the input image in the area where this element should be located. If the distance between the ideal and the features calculated in the input image exceeds a specified threshold, the system issues a failure. When producing documents, luminescent inks of different colors can be used. The sixth stage of the algorithm involves analyzing the color of luminescent elements and comparing them with those provided for this type of document. If the luminescence color differs from the specified one, the system will issue a failure. CLAIM A method for detecting signs of falsification of documents and banknotes from images obtained in UV, which consists in sequential analysis of UV luminescence in the input image, characterized in that images in the RGB color space containing a document obtained using UV illumination are examined; in this case, at the first stage, the image is analyzed for the presence of luminescence in those parts of the document where it is not provided; at the second stage of the analysis, luminescent protective fibers from various materials embedded in the surface layer of the paper are detected in the input image; at the third stage, the location of all luminescent areas of the document is analyzed, except -