Security document verification apparatus and method for inspecting security documents

Abstract

The present invention relates to a security document verification device comprising: a document scanning unit configured to scan a security document; a smart inspection unit configured to automatically inspect at least one provided image; and an output unit configured to provide one or more security document feature levels and / or document forgery probabilities. The present invention further relates to a method for inspecting a security document using the security document verification device, and a computer-readable medium for causing a programmable processor to perform the method.

Description

Technical Field

[0001] The present invention relates to a security document verification device, a method for inspecting a security document using the security document verification device, and a computer-readable medium for causing a programmable processor to execute the method.

[0002] Background of the Invention WO 2009 / 075987 provides a technique for identifying and verifying security documents by applying a dynamic document identification framework. A security document authentication device includes an image capture interface that receives a captured image of a document to be identified and / or verified. The security document authentication device further includes a memory unit that stores a plurality of document types within a data structure provided by the dynamic document identification framework. The security document authentication device also includes a document processing engine that traverses the captured image data using a data structure selectively invoked by one or more of a plurality of processes. The processing engine identifies the captured image(s) as one of a plurality of document type objects. This identification method is performed by traversing a data structure stored according to the dynamic document identification framework, and a given data structure has to be traversed element by element, which is not a very efficient way of providing an identification result as it consumes computing time and memory.

[0003] U.S. Patent No. 11594053 describes a deep learning-based identity card authenticity verification device for automatically checking the authenticity of identity cards. The device includes inputting identity card data into a feature extraction model to extract feature information fragments. Furthermore, indicators for checking the authenticity of the identity card are represented from the identity card data. The extracted feature information fragments are input into a classification model for determining the authenticity of the identity card. If the identity card is determined to be tampered with, a class activation map is extracted from the identity card data using the feature information fragments. The class activation map indicates and displays the tampered areas of the identity card.

[0004] Currently, there are multiple solutions for automated document inspection, including physical devices supported by a reference document database and, additionally, authentication software that interacts with an authentication database. For example, U.S. Patent Application Publication 2022 / 0207901 discloses a border control system that enables the verification of travel documents using artificial intelligence. This system comprises a document reader, a display, a keyboard, a database, and a control unit. The document reader comprises a processor, a communication device, and memory containing an operating system, AI, a self-learning verification module, and other functional modules. Travel documents are compared to an AI model, so the entire travel document is the input, and the AI ​​model outputs whether it is a potentially new document type or a potentially forged document. Thus, if a document type is detected and one or more new features are found, these features are captured and used for future model training to confirm that this document type is new with xyz differences.

[0005] Apply this instruction to the following example. Assume there are three actual versions of a valid passport from country A. Version 1 of the passport exists within its border control system and can be detected and verified. Version 1 is already correctly trained with slight differences as similarity. If version 2 of the passport is detected, which does not exist in the system but is within the similarity range of version 1, the system can find that version 2 is similar to version 1 and will also verify version 2. If version 3 of the passport is detected, which also does not exist in the system but is not within the similarity range of version 1, the system cannot verify version 3 because the differences are too great or too large. This is because this approach still relies on a document database.

[0006] These authentication software and associated databases require continuous, time-consuming updates whenever new security documents to be scanned are officially released, or whenever new security features are introduced or enhanced for known security documents. Automated machine scans cannot be performed clearly if a document is not yet included in the database, and therefore the authentication software cannot compare scanned features with referenced features and provide a scan result.

[0007] Therefore, there is a need for security document validation devices and methods for examining security documents that do not require a reference document database and / or authentication database.

[0008] The primary objective of the security document verification apparatus of the present invention is to enhance the verification and authentication of security documents so that the inspection process is no longer based on element-wise comparison routines that consume time resources, computational resources, and memory space resources.

[0009] Summary of the Invention The objectives identified above are addressed by detecting security features and verifying their authenticity using a predetermined trained learning module. Features of the security document verification apparatus and its corresponding method of the present invention are given by the independent claims. Further advantageous embodiments are described in the dependent claims.

[0010] In one aspect of the present invention, a security document verification device is provided, comprising a document scanning unit, a smart inspection unit, and an output unit.

[0011] The document scanning unit is configured to scan security documents to be analyzed and inspected, which means that security documents are scanned, captured, or pictured. Security documents may be (electronic) identification cards, passports, visas, residence permits, driver's licenses, social security cards, physical certificates or banknotes, or other valuable documents (documents that have a higher value than the substrate on which they are printed or placed).

[0012] The scanning unit may include a camera or another optical system for scanning security documents, such as a microscope.

[0013] According to one aspect of this disclosure, a security document has one or more security features that are machine-readable and therefore can be automatically processed further upon detection. As a prerequisite for automatic and immediate processing, at least one image of the scanned security document is provided to a smart inspection unit. Two or more images of the security document may be further provided by a document scanning unit, for example, sequentially or simultaneously.

[0014] Additionally, the electronic data stored within the inspected security document is received by a security data verification application for further processing and / or inspection.

[0015] The electronic data provided within the security document may be biometric data that can be extracted or read from an RFID chip by applying an appropriate frequency and encoding scheme. The electronic data may also be personal data stored in an RFID chip that is readable only to authorized units or users.

[0016] The stored electronic data can also be a digital representation of one or more security features (physically located) on or within a secure document, such as an attribute certificate, a digital signature, a hologram as digital data, or a picture as digital data, or a photograph as digital data, each stored on the RFID chip. These electronic data can only be read by authorized users.

[0017] The stored electronic data can be transferred from the RFID chip of the security document to a smart inspection unit. For example, the RFID chip of the security document is read by a chip reading unit, such as a smart card reader unit of a smart inspection unit, to receive the stored electronic data. This received data is then transferred for further inspection by the smart inspection unit.

[0018] The smart inspection unit is configured to automatically and / or instantly inspect at least one image provided by the scanning unit. Each of at least one of the security features is inspected by applying a corresponding trained learning module. The corresponding trained learning module is a module that is automatically applied to the (detected) security feature. For example, when this security feature is detected, a trained learning module is invoked that can inspect the security feature for its authenticity.

[0019] The approach of the present invention does not require reference documents and / or authentication databases to examine security documents. The examination process of the present invention, according to a predetermined routine, notifies a person or machine operating the security document verification device of the present invention of the presence and location of security features on the security document to be verified. The authenticity level of these security features is maintained and displayed on the output unit. The probability of a forged security document may be additionally maintained and output.

[0020] This invention does not rely on the limitations of prior art, such as time-consuming databases and the updating of such databases. In this invention, no database of documents by country or type is created; instead, all security documents are scanned with a focus solely on security features. This means that it is possible to train an AI module on any security document that possesses genuine security features. Furthermore, by using one or more trained learning modules, security document feature levels and / or security document forgery probabilities can be provided. This is advantageous because it allows for a much broader simulation of border patrol behavior.

[0021] Using the example presented in the background technology section above, it means that even if all three versions of a passport are scanned and have not been used to train an AI module in the past, it is possible to extract security features, provide document feature levels, for example in percentages, and / or give a probability of forgery, for example in percentages.

[0022] The present invention provides a device that outputs the probability that a security document or security features on / within a security document are genuine or fraudulent. The present invention also provides a quantified indicator of how secure a genuine security document is, as well as how reliable the manufacturing and personalization processes are.

[0023] These security documents may possess at least one machine-readable security feature. Such security features may also be human-detectable.

[0024] Machine-readable security features may also include the detection of specific materials used as the basis for the physical appearance of the security document. For example, the security document may consist of a given paper or a plastic such as a characteristic polycarbonate material.

[0025] The paper of the security document can have specific characteristics such as a predetermined thickness and / or brightness, and the inspection unit is configured to inspect the thickness and / or brightness using a corresponding learning module. As another example, the surface structure of a security document, such as a paper surface or a plastic surface, can also be characteristic of the validity or authenticity of the security document, and the inspection unit can be configured to inspect the surface structure using a corresponding learning module. The surface structure can include a defined number of lines per area or a defined pattern as a mechanically detectable security feature. The nature of the surface roughness of the paper used to issue the security document can also be a machine-readable security feature. Such security features can additionally be perceptible by a human who inspects and evaluates the security feature (or the security document) at a glance.

[0026] An exemplary machine-readable security feature may be a multiple laser image (MLI). The MLI includes multiple images within the same surface area, and the appearance of one or more of the images changes with the change in the viewing angle.

[0027] An exemplary machine-readable security feature may be a variable laser image (CLI). The CLI includes multiple colors, and one or more of the colors changes with the change in the viewing angle.

[0028] Such MLI or CLI machine-readable security features can be printed on the (polycarbonate) surface of the security document, and a plurality of notches can be provided. The nature of the notch can be a prominent feature suitable for the scanned security document.

[0029] The document scanning unit may be configured to scan the security document at high resolution and / or ultra-high resolution to determine and / or analyze security features, such as MLI or CLI, for example. There are several options described below.

[0030] One option is, preferably within the area of security features such as MLI or CLI, to scan the security document by taking a visible light image / picture from the security document and, without further optical improvement of the scan, analyze this visible light image. This option is considered to scan and analyze pictures at a normal resolution, for example 400 dpi or less.

[0031] Another option is, preferably within the area of security features such as MLI or CLI, to scan the security document by taking a visible light image / picture from the security document and, to obtain a high-resolution image / picture, apply an optical zoom lens, and thus, preferably within the area of security features such as MLI or CLI, apply optical magnification to analyze the security document. This option is considered to scan and analyze pictures at a high resolution higher than the normal resolution (without magnification), for example greater than 400 dpi, preferably greater than 500 dpi, more preferably 600 dpi.

[0032] Yet another option is, preferably within the area of security features such as MLI or CLI, to scan the security document by taking a visible light image / picture from the security document and, preferably within the area of security features such as MLI or CLI, apply an even stronger optical magnification to analyze the security document, for example by using a microscope for zooming to obtain an ultra-high-resolution image / picture. This option is considered to scan and analyze pictures at an ultra-high resolution higher than the high resolution (with magnification) and higher than the normal resolution (without magnification), for example the high resolution is greater than 600 dpi, preferably greater than 800 dpi, more preferably greater than 1000 dpi, and most preferably 1200 dpi or more.

[0033] In a preferred embodiment, to read MLI or CLI security features and satisfy the need to obtain high-resolution images from a document scanner, the security document may be magnified to obtain high-resolution images for further analysis, or a microscope may be used as a high-magnification device to obtain ultra-high-resolution images for further analysis.

[0034] Exemplary machine-readable security features may include, for example, (microscopic) laser-perforated patterns placed on the surface of a security document, provided as through-holes within a paper or plastic substrate that define the physical appearance of the security document.

[0035] Exemplary machine-readable security features that may also be detectable by humans could be holograms and / or watermarks, for example, which could be further examined by applying conventional light sources and / or laser light.

[0036] A security document verification device may receive electronic data and use one or more corresponding trained learning modules to examine corresponding security features. The received electronic data may be compared with one or more security features. Thus, the received electronic data may be compared with one or more of ultraviolet light images, infrared light images, and visible light images. For example, the inspection unit may attempt to match and compare an image printed on a security document with electronic image data received from an RFID chip, for example, using facial recognition technology. For example, biometric data may be detected by reading an RFID chip, triggering a smart inspection unit to automatically apply the corresponding inspection learning module to compare the received biometric data with the security features of the security document corresponding to the biometric data, such as a passport picture. This routine may be used for authentication purposes when the biometric features of a recently captured image of a passport holder are aligned with the security features of the passport holder's security document, for example, the passport holder's passport. This comparison may result in positive or negative authentication, for example, in an immigration inspection room at a border site, which may cause further steps to be handled by a border control unit.

[0037] The corresponding inspection learning modules, as requirements for smart automated inspection of security features provided with security documentation, may be one or more of the modules listed below. This list may be expanded in an unrestricted manner as future developments and the implementation of trained learning modules are desired.

[0038] A trained learning module can handle analyzing the security feature "paper" to detect the thickness of the paper or its surface roughness. For example, the presence of a hinge, e.g., the top of a security document sewn into a booklet, which holds the data pages of the security document. In the case of a paper-based security document, there is no hinge, and the top of the data page is paper, but in the case of a polycarbonate-based security document, the hinge is a special gap between the top of the data page and the binding of the booklet.

[0039] As of today, paper-based substrates for data pages in security documents can be regular paper or artificial paper such as Teslin.

[0040] A trained learning module can, for example, analyze the security feature "polycarbonate" considering its optical properties and extract its chemical composition for comparison with a learned baseline composition.

[0041] As of today, passports as security documents may have a thick or thin polycarbonate substrate for the data pages. Alternatively, cards as security documents may have a polycarbonate substrate, PET / PVC substrate and / or ABS and other multipolar polymer substrates.

[0042] The trained learning module can handle the analysis of the security feature "laser drilling." For example, laser drilling parameters such as the geometric design of the drilling or the distance between holes can be examined.

[0043] A trained learning module can handle the analysis of security features, specifically "printed information." For example, inkjet printing parameters such as the scale of characters or fonts used within a security document are observed and inspected. Printed information also includes laser-engraved information, punched information, or other techniques for providing information within or on a security document.

[0044] The trained learning module can handle the analysis of the security feature "ultraviolet image". For example, wavelength analysis is applied to identify the image. Thus, artwork such as patterns visible under ultraviolet light is extracted from the document's data pages. The quality of these patterns can then be verified by checking different colors and their properties. For example, red in older documents will be less visible than in newer documents. How the UV artwork appears on the data pages is further checked.

[0045] A trained learning module can handle analyzing the security feature "infrared image". For example, wavelength analysis can be applied to identify the image. Evidence of which printing technology was used (laser printing, inkjet printing, or both) can be found. Based on this, any differences between printed text visible under UV and printed text visible under IR can be examined.

[0046] The trained learning module can handle analyzing the security feature "holographic overlay." For example, holographic overlay parameters such as mirroring a photograph can be examined.

[0047] Further future training and learning modules may be reserved for special and / or confidential security features.

[0048] The corresponding trained learning module is automatically selected and applied when the associated security feature is detected. For example, if a hologram is detected as a security feature, the smart inspection unit selects the trained learning module "hologram" and applies it. Next, it calculates the probability of it being a genuine security feature and / or a forged security feature.

[0049] After detecting a hologram on a data page, the following method can be applied: examining the surface of the hologram to find cracks or missing elements. A document forger may attempt to reuse a genuine hologram or replace it with some kind of dummy hologram. A hologram may have several artwork designs, and it may be applied to measure the complexity of the detected hologram artwork. When checking a hologram, especially on a paper data page, wear and tear of the document should be taken into consideration.

[0050] The output unit of the security document verification device is configured to provide multiple calculated probabilities of the authentic security document feature level for each machine-readable security feature after applying the corresponding trained learning module.

[0051] Alternatively, the output unit of the security document verification device may be configured to provide multiple calculated probabilities of forged security document feature levels for each machine-readable security feature after applying the corresponding trained learning module.

[0052] The output unit of the security data verification device may be configured to provide the probability of authentic security features while simultaneously providing the probability of document forgery.

[0053] The smart inspection unit of the security document verification unit is configured to detect machine-readable security features, such as laser-perforated features, within security documents by applying image processing methods to scanned or optically analyzed images of security documents. One or more parameters are required to select the corresponding trained learning module to apply for verification of the detected security features against a learning pattern, such as that provided by the trained learning module.

[0054] Image processing methods (e.g., edge detection) are used in conjunction with computer vision techniques (e.g., object detection and / or image segmentation). This module was trained to detect specific holes (visible under IR and possibly VL) within document data pages, extract them, and merge them together to read out the "engraved" information.

[0055] The image processing method may involve calculating the image histogram, determining its entropy, calculating its brightness or the distribution of its brightness, or simply interpolating.

[0056] The trained learning modules may be based on artificial intelligence techniques such as deep learning schemes, machine learning, neural networks, convolutional neural networks, computer vision, pattern recognition, or knowledge engineering. For example, computer vision may be used.

[0057] The trained learning module is trained to capture scanned security documents or analyze scanned images based on one or more of the following parameters:

[0058] In one embodiment of a security document verification device, the document scanning unit may be configured to scan in a first wavelength region of the electromagnetic spectrum, where the first wavelength region is, for example, the visible light region that is recognizable by humans.

[0059] Additionally or alternatively, the system may be configured to scan in a second wavelength region, which is the infrared region. For example, an IR image may be visible under IR light but invisible under UV light or light in the visible spectrum.

[0060] Additionally or alternatively, the system may be configured to scan in the ultraviolet light region. For example, a UV image may be visible under UV light but invisible under IR light or light in the visible spectrum.

[0061] Therefore, different images of the security document at different wavelengths are provided in order to detect and inspect different types of security features.

[0062] Each of the different images provided by the scanning unit can be examined by its own trained learning module.

[0063] The output unit of security document verification may be configured to provide one or more security document feature levels on the display unit. Alternatively or additionally, it may also be configured to transfer the probability of document forgery to a backend unit to condition the execution of further steps, such as retrieving a forged passport and denying the holder of the forged passport the right to cross a border.

[0064] The following two parameters are provided.

[0065] Document security feature levels (e.g., %) to provide information about security features detected within the document.

[0066] Document forgery probability (e.g., in percentage) - For each security feature detected, the probability of forgery is provided in percentage.

[0067] Using a graphical user interface, two parameters are displayed in a user-friendly mode, showing the security level of the scanned document along with its probability of forgery.

[0068] The present invention further includes a method for inspecting a secure document using the security document verification apparatus of the present invention according to the embodiments described above.

[0069] This method includes the steps of scanning or capturing a security document and providing an image of the scanned security document, or detecting electronic data stored on the security document. Further steps include receiving this electronic data and analyzing it, taking into account its classification and content.

[0070] After classification, the provided image data and / or electronic data are automatically inspected, taking into account the detected security features, by using one or more trained learning modules corresponding to the detected security features that need to be further analyzed.

[0071] After evaluating security features by matching them against patterns provided by a trained learning module, one or more security document feature levels are provided as percentage values. The resulting document feature levels can be either successful or unsuccessful detections. Regardless of whether detection is successful or unsuccessful, a defined, configurable threshold is used. For example, the threshold for polycarbonate data pages is different from that for paper data pages.

[0072] The probability of document forgery can also be determined by the method of the present invention.

[0073] Each trained learning module can produce one result, and multiple such results can be organized or summarized within a vector or vector structure that can be output by an output unit.

[0074] A vector can be normalized by the number of its components. For example, each row can correspond to one separate learning module (for one security feature) and contain one result value for that security feature, for example, as a percentage. Combining all these values ​​means that the vector is normalized by the number of its components. In other words, if a passport data page is scanned and multiple security features are recognized, it is possible to verify each security feature against the inspection learning module and assign a percentage of its authenticity. Then, a threshold is established for each component value of the vector. Normalization in this sense means that if a particular security feature is not recognized in the security document (for example, because the security document does not have that security feature), that feature is not added to the vector.

[0075] If the normalized vector quantity (the quantity of the normalized vector) is greater than a defined threshold, the scanned security document is validated and / or authenticated. Vector normalization is necessary for comparing vectors with each other or comparing the quantity of a vector with a threshold.

[0076] The present invention also includes a computer-readable medium in which instructions for causing a programmable processor to perform the method steps described above are stored. [Brief explanation of the drawing]

[0077] The present invention or further embodiments and advantages of the present invention will be described in more detail below with reference to the drawings, which illustrate only embodiments of the present invention. The same reference numerals are used for the same parts in each figure.

[0078] Drawings should not be assumed to be to scale, and individual elements of the drawings may be shown in an exaggeratedly large or exaggeratedly simplified form. [Figure 1] This figure shows an example of a document inspection system using conventional technology. [Figure 2] This figure shows an exemplary embodiment of the security document verification device according to the present invention. [Figure 3A] This figure shows an exemplary embodiment of a hardware unit in a security document verification device according to the present invention. [Figure 3B] This figure shows another exemplary embodiment of a hardware unit in a security document verification device according to the present invention. [Figure 4] This figure shows an exemplary security document verified by the security document verification device of the present invention. [Figure 5] This figure shows the output screen of the output unit of the security document verification device according to the present invention. [Figure 6a] This figure shows an exemplary embodiment of a hinge for polycarbonate-based security documents. [Figure 6b] This figure shows an exemplary embodiment of a hinge for polycarbonate-based security documents. [Figure 6c] This figure shows an exemplary embodiment of a paper-based security document. [Figure 6d] This figure shows an exemplary embodiment of a paper-based security document. [Figure 7a] This figure shows exemplary embodiments of visible light, infrared, and ultraviolet images of a passport picture provided using inkjet printing technology. [Figure 7b] This figure shows exemplary embodiments of visible light, infrared, and ultraviolet images of a passport picture provided using laser printing technology. [Figure 8a]This figure shows exemplary embodiments of visible light, infrared, and ultraviolet images of a passport picture provided using inkjet printing technology. [Figure 8b] This figure shows exemplary embodiments of visible light, infrared, and ultraviolet images of a passport picture provided using laser printing technology. [Figure 9a] This figure shows an exemplary embodiment of a portion of the ultraviolet images of a passport picture being inspected for authenticity. [Figure 9b] This figure shows an exemplary embodiment of some ultraviolet images from passport pictures that are being inspected for fraud. [Figure 10] This figure shows exemplary embodiments of the learning module vector and its corresponding result, indicated by percentages.

[0079] Detailed description of embodiments of the present invention Figure 1 shows an example of a document inspection system using conventional technology.

[0080] An identity document 3a, as an example of a security document 3, includes a data page 3b as its top surface. One or more security features 4 are placed on the data page 3a. The identity document 3a is placed on a scanning device 18 to perform a document scan 18a at three different wavelengths to scan the data page 3b. The document scan 18a provides, for example, a visible light image 5v, an ultraviolet light image 5u, and an infrared light image 5i of the data page of the identity document 3a. If an RFID chip 16 is placed on the identity document 3a, chip reading 17 is also performed by the scanning device 18.

[0081] As the next step, authentication software 19 interacting with the reference document database 20 analyzes the provided visible light image 5v, ultraviolet image 5u, and / or infrared image 5i in terms of their truthfulness or authenticity, and then provides a result 14. For this purpose, the reference document database 20 includes a reference identity document 21 that is compared with the provided visible light image 5v, ultraviolet image 5u, and / or infrared image 5i, taking into account specific document characteristics. The result 14 may be a "verification success" or "verification failure" of the truthfulness / authenticity of the identity document 3a. The result can be displayed as "Result OK," "Result NOT OK," or "No Result." This result 14 may further be a prerequisite for authentication of the owner of the identity document, taking into account the already verified identity document 3a.

[0082] The existing automated document inspection system shown in Figure 1 uses authentication software 19, which includes a dynamic document identification framework that interacts with a reference document database 20. The data structure of a reference identity document 21 of known identity documents 3a previously stored in the reference document database 20 is compared element by element with the provided visible light image 5v, ultraviolet light image 5u, and / or infrared light image 5i.

[0083] A known drawback of the system is that if identity document 3a is not known to the system, the reference document database 20 must first be updated so that the authentication software 19 can provide a valid result 14. If the unknown identity document 3a is not inserted into the reference document database 20, the result 14 will simply show "No result," and verification and / or tampering will not be possible.

[0084] Figure 2 shows an exemplary embodiment of a security document verification device 1 according to the present invention. The security document verification device 1 comprises a document scanning unit 2 configured to scan a security document 3 having one or more machine-readable security features 4. The document scanning unit 2 comprises three different image capture units and their corresponding processing units, each interfaced with a smart inspection unit 7. The security document 3 may be a passport, ID card, visa, residence permit, driver's license, social security card, physical certificate, or banknote. The document scanning unit 2 provides at least one of a visible light image 5v, an ultraviolet light image 5u, and / or an infrared light image 5i of the scanned security document 3. If electronic data 6 is available and stored in the security document 3 and read by a chip readout 17, the electronic data 6 may be received by the document scanning unit 2 configured accordingly. The storage unit 15 for the electronic data 6 may be an RFID chip 16. A security document 3 without electronic data 6 may be additionally scanned by the document scanning unit 2. These security documents 3 may also be passports issued by countries that do not apply biometric data as a security feature 4.

[0085] The security document verification device 1 further includes a smart inspection unit 7 configured to automatically inspect a provided image 5. More than one image may also be provided. The provided image(s) 5, 5v, 5u, 5i and / or the received electronic data 6 are immediately and / or automatically analyzed after being scanned via the scan unit 2 by using one or more trained learning modules 8. Each trained learning module 8 (indicated here as different learning modules 8a to 8g) is configured to inspect one or more corresponding security features 4 of the security document 3. Each security feature 4 is machine-readable. Each security feature 4 is detected by the smart inspection unit 7. Upon detection, the corresponding trained learning module 8 is applied, and an output unit 9 configured accordingly provides one or more security document feature levels 10 for each detected machine-readable security feature 4 after the corresponding trained learning module 8 has been applied (executed). Additionally or alternatively, the document forgery probability 11 for each machine-readable security feature 4 is provided after the corresponding trained learning module 8 has been applied.

[0086] The trained learning module 8a is configured to analyze the security feature “paper” to detect the thickness or surface roughness of the paper. This may include analyzing the color of the paper using an infrared image 5i, an ultraviolet image 5u, and a visible light image 5v. This may include analyzing other visible elements in (or on) the paper. This may include analyzing the paper to find security features that may only be present on paper.

[0087] This may include analyzing the color of the paper using infrared, ultraviolet, and visible light images. This may include analyzing other visible elements in or on the paper. This may include analyzing the paper to find security features that may only be present in paper. For security documents, paper with different components(s) compared to conventional paper may be used. For example, security document paper may contain cellulose and cotton. Therefore, security paper may exhibit different reflection or absorption behavior than other paper materials, particularly under ultraviolet or infrared light. In counterfeit documents, the paper appears brighter under UV light.

[0088] Additionally, it is possible to detect watermarks in paper and / or particles within the paper, such as UV fibers or metallic threads, that are visible under one or more of the following: ultraviolet light 5u, infrared light 5i, or visible light 5v.

[0089] This could involve analyzing paper to find security features that may only exist on polycarbonate.

[0090] This may include, for example, analyzing the paper surface to detect surface roughness by analyzing images under ultraviolet light 5u or infrared light 5i to detect the properties of the paper material used.

[0091] For example, as shown in Figures 6a to 6d, the presence of a hinge 26 is analyzed. The hinge 26 points to the top of the data page 3b of the security document 3 sewn into the booklet and holds the data page 3b of the security document 3. Figures 6a and 6b show a hinge 26 for a polycarbonate-based security document 3. Here, the hinge 26 is a special element adjacent to the data page 3b. Figures 6c and 6d do not show a hinge, and therefore, this security document 1 can be determined to be a paper-based security document 3. There is no gap or step between the top of the data page 3b and the binding of the booklet.

[0092] Further applicable techniques for checking whether paper-based security document 1 or polycarbonate security document 1 is used are listed in the table below.

[0093] [Table 1]

[0094] As of today, paper-based substrates for data pages in security documents can be ordinary paper or artificial paper such as Teslin. As of today, passports as security documents can have thick or thin polycarbonate substrates for data pages. Alternatively, cards as security documents may have polycarbonate substrates, PET / PVC substrates and / or ABS and other multipolar polymer substrates.

[0095] The trained learning module 8b is configured to analyze the security feature "polycarbonate" considering its optical properties and extract its chemical composition for comparison with a learned baseline composition. This may include analyzing security documents to find security features that can only exist on polycarbonate.

[0096] The trained learning module 8c is configured to analyze the security feature "laser perforation". For example, laser perforation parameters such as the geometric design of the perforations or the distance between holes are examined. For example, data page 3b, visible under IR light, provides an opportunity to see laser-perforated holes. These holes may be arranged in a specific pattern to provide user-readable information such as letters, numbers, etc. Alternatively, it is a machine-readable (encoded) pattern. Using the trained learning module 8c, it is possible to find the holes in images 5i, 5v, 5u of data page 3b, extract these holes, and classify and decode arbitrary information by merging all the holes together. Classification means that the holes may have different shapes such as squares, circles, or triangles.

[0097] The trained learning module 8d is configured to analyze the security feature “Printed Information.” For example, inkjet printing parameters such as the scale of characters or fonts used within a security document are observed and inspected. Printed information also includes laser-engraved information, punched information, or other techniques that provide information within or on a security document.

[0098] The trained learning module 8d can, for example, detect offset printing technology. Offset printing is widely used for all types of secure documents 1.

[0099] The trained learning module 8d inspects image quality under visible light, UV light, and IR light, yielding images 5v, 5u, and 5i, for example, as shown in Figures 7a to 9b. In the case of laser printed images, as shown in Figures 7b and 8b, the motif can be identified very well in 5v and 5i. In contrast, in inkjet printed images, as shown in Figures 7a and 8a, visibility is highly dependent on the black ink, which is only visible as other inks cyan, magenta, and yellow, and cannot be seen under UV in 5u. Additionally, in the photographic domain, patterns differ considerably due to the technology used; for example, laser printing yields solid areas and very detailed images, while inkjet printing yields lower resolution compared to laser printing, and specific patterns are produced because the print head of an inkjet printer drops ink whenever needed; this technology is called Drop on Demand (DoD).

[0100] When documents are forged, inkjet printing technology is typically used. The difference is clearly visible in print quality, which can be categorized using module 8d.

[0101] The trained learning module 8e is configured to analyze the security feature "holographic overlay". For example, holographic overlay parameters such as mirroring of a photograph are examined. Since the holographic overlay is visible under visible light, it can be extracted from image 5i using the trained module 8e. In the case of paper-based data page 3b, personalized data needs to be protected. In the case of polycarbonate-based security document 3, a holographic logo partially covers the main photograph. This is to protect this photograph from having another photograph added to this layer. The holographic overlay can be extracted from the visible light image 5i. If the pattern was created using a machine, and this pattern covers the entire data page 3b, it can be measured by module 8e.

[0102] The trained learning module 8f is configured to analyze special and / or confidential security features. Some of the security features 4 are confidential and require specialized software to decrypt them. These features, such as Jura IPI or IAI ImagePerf, allow encrypted data to be read and decrypted using either a dedicated lens or a dedicated software solution.

[0103] A trained learning module (not shown in Figure 2) can analyze the security feature "ultraviolet image". For example, wavelength analysis is applied to identify the image. It is possible to extract UV patterns, colors, and their presence in certain areas, such as the MRZ or on the main passport holder's photograph. This provides information on whether the passport is adequately protected using UV functionality. Additionally, the age of Document 1 can be estimated.

[0104] Furthermore, for example, how the UV artwork appears on the data page is checked.

[0105] If UV light is layered on top of the printed area, it means that the UV light comes from a holographic overlay that is attached to the document surface after personalization.

[0106] If a UV light artwork element matches a visible light artwork element, it means that one of the colors used during offset printing was UV active.

[0107] If the UV brightness is too high (result of the paper module), and at the same time, UV fibers are not visible in security document 3, then data page 3b is not authentic.

[0108] Some technologies allow for the printing of passport holder information under UV light. When a UV module is used, it is possible to verify whether this data (personal data and / or photographs) is present.

[0109] As can be seen from image 5u in Figure 7a, image 5u provides a star-shaped UV pattern 4h-2 and a diamond-shaped pattern 4h-1. As can be seen from image 5u in Figure 7b, image 5u provides a first color line UV pattern 4h-3 and a first color line UV pattern 4h-4. As can be seen from image 5u in Figure 8a, image 5u provides a stamp UV pattern 4h-5 and a snow star UV pattern 4h-6. As can be seen from image 5u in Figure 8b, image 5u provides a stamp UV pattern 4h-5. As can be seen from image 5u in Figure 9a, image 5u provides a wing pattern 4h-8 and a dot UV pattern 4h-7. As can be seen from image 5u in Figure 9b, image 5u provides only the wing pattern 4h-8. Since the dot UV pattern 4h-7 is missing in Figure 9b, this security feature is considered invalid, and the result is "not authenticated".

[0110] A trained learning module (not shown in Figure 2) can analyze the security feature "infrared image". For example, wavelength analysis can be applied to identify the image. Evidence of which printing technology was used, i.e., laser printing, inkjet printing, or both, can be found. Based on this, any difference between printed text visible under UV and printed text visible under IR can be examined. This helps to understand which technology was used to personalize the passport. In some countries, the wrong black ink may be used to personalize passports. Neither data was visible under IR light. This module can be used to find out which printing technology was used. The IR wavelength allows reading of passport data page 3b - personalized text only - without interruption of the artwork. Additionally, since laser images are fully visible under IR, it is possible to double-check whether the main photograph was printed using a laser (see image 5i in Figure 7b or Figure 8b, compare with image 5i in Figures 7a and 8a).

[0111] As can be seen from image 5i in Figure 7a, image 5i does not provide a photograph but provides some artifacts 4i. As can be seen from image 5i in Figure 7b, image 5i provides a photograph. As can be seen from image 5i in Figure 8a, image 5i provides a stripe IR pattern 4i. As can be seen from image 5i in Figure 8b, image 5i provides a photograph.

[0112] The present invention is not limited to modules 8a to 8g shown in Figure 2. It is included in the present invention that module 8 may include more or fewer modules than those shown in Figure 2.

[0113] Figure 3A shows an exemplary embodiment of a hardware unit 23 in a security document verification device 1 according to the present invention, the device 1 including three different image processing units 22v, 22u, and 22i, an RFID chip reader 17, and a data processing unit 17a. Each of these units 22v, 22u, 22i, 17, and 17a, as well as a smart inspection unit 7, an output unit 9, and a plurality of trained learning modules 8, can be implemented as software on one or more hardware units 23.

[0114] Figure 3B shows another exemplary embodiment of a hardware unit 23' in a security document verification device 1 according to the present invention. In contrast to Figure 3A, only the smart inspection unit 7, the output unit 9, and the multiple trained learning modules 8 may be implemented as software on one or more hardware units 23'. The hardware unit 23' includes an interface 24 for the hardware unit 23' to receive one or more images 5i, 5u, 5v from each of the one or more image processing units 22i, 22u, 22v.

[0115] In both Figure 3A and Figure 3B, the output unit 9 is part of a display device 13 having a screen that displays the security document feature level 10 and the document forgery probability 11.

[0116] Figure 4 shows an exemplary security document 3 to be verified by the security document verification device 1 of the present invention. Here, the security document 3 is an electronic ID card. This eID card is read by the security document verification device 1 of the present invention and further processed for verification. The eID card includes several different security features 4, for example, a picture 4a showing a photograph of the eID card holder, a hologram 4b obtained by mirroring the picture 4a, an RFID chip 16, and at least one MLI (or CLI) image 4d. The material of the eID card is a predetermined polycarbonate 4c material or material composition. Another security feature 4 is an area 4e having microfonts, which can only be read by a document scanning device 2, for example, a camera or another optical device. Another security feature 4 is printed information 4f that can be detected by an optical character recognition method.

[0117] Figure 5 shows a display device 13 showing an exemplary security document 3 processed by the security document verification device 1 of the present invention. The display 13 displays the results 14 of the method of inspecting the security document 3 using the security document verification device 1 described above.

[0118] In the example in Figure 5, the tampering result 14 is output. The image of the analyzed ID document 3 is also displayed in combination with the tampered area 25. In this case, the security document verification device 1 recognizes the mirrored hologram 4b as forged, which is indicated by "No" enclosed in a box. Alternatively (not shown in Figure 5), the security document verification device 1 may recognize the mirrored hologram 4b as valid, which may be indicated by "Yes" enclosed in a box.

[0119] Figure 10 shows exemplary vectors for learning modules 8a-8g and the corresponding results for security feature levels and forgery probabilities.

[0120] Each trained learning module 8a-8g can produce one result, and multiple such results can be organized or summarized into a vector or vector structure that can be output by output unit 9.

[0121] A vector can be normalized by the number of its components. For example, each row could correspond to one separate learning module 8a through 8g (for one security feature) and contain one result value for that security feature, for example, a percentage. Combining all these values ​​means that the vector is normalized by the number of its components.

[0122] When the passport data page of security document 3 is scanned, multiple security features 4 may be recognized. Here, it is possible to verify each of the security features 4 to the inspection learning module 7 and assign it a percentage of its (unique) authenticity. For example, in Figure 10, the first column is dedicated to the detection of security features 4 in security document 3. The detection results are shown in the second column. The third column represents the probability of forgery when the document is viewed (analyzed) for the first time in the system. The fourth column represents the probability of forgery when similar documents have been viewed (analyzed) by the system in the past.

[0123] A threshold is established for each component value of the vector. Normalization, in this sense, means that if a particular security feature 4 is not recognized in security document 3 (for example, because security document 3 does not possess that security feature 4), it is not added to the vector. If the amount of the vector normalized by the number of its components (the amount of the normalized vector) is greater than the defined threshold, the scanned security document 3 is validated and / or authenticated. Vector normalization is necessary for comparing vectors with each other or comparing the amount of a vector with a threshold. [Explanation of Symbols]

[0124] 1. Security Document Verification Device 2 Document Scanning Unit 3. Security Documents 3a Identity Document 3b Data page 4. Security Features 4a Picture 4b Mirror Hologram 4c polycarbonate material 4D Multiple Laser Imaging (MLI) Area with 4e microfonts 4F Printed Information 4G infrared pattern 4h UV pattern 4i Visible Pattern 5 images 5i Infrared light image 5u ultraviolet light image 5V visible light image 6. Electronic data stored (received) within a security document 7 Smart Inspection Unit 8 Trained Learning Modules 8a-8g Individually trained learning modules 9 Output Units 10 Security Document Feature Levels 11. Probability of Document Forgery 12 detection units 13 Display Devices 14 Results 15 Memory Units 16 RFID chips 17 Chip Readout 18 Scanning Devices 18a Document Scan 19 Authentication Software 20 Reference Document Database 21. Reference Identity Document 22V Visible Light Image Processing Unit 22u Ultraviolet Light Image Processing Unit 22i Infrared Light Image Processing Unit 23,23' Hardware Unit 24 Interfaces 25. Area of ​​tampering 26 Hinge

Claims

1. Security document verification device (1), Document scanning unit (2), A document scanning unit (2) is configured to scan a security document (3) having one or more machine-readable security features (4), provide at least one image (5) of the scanned security document (3), and preferably also receive electronic data (6) stored within the security document (3), A smart inspection unit (7) configured to automatically inspect the provided at least one image (5) and the received electronic data (6) using one or more trained learning modules (8), wherein each of the at least one of the security features (4) is inspected by applying the corresponding trained learning module (8), Output unit (9), An output unit (9) is configured to provide one or more security document feature levels (10) for each machine-readable security feature (4) after applying the corresponding trained learning module (8), and / or to provide a document forgery probability (11) for each machine-readable security feature (4) after applying the corresponding trained learning module (8), A security document verification device (1) equipped with the following:

2. The one or more machine-readable security features (4) of the security document (3) are, Polycarbonate material, Laser drilling, Multiple laser imaging (MLI), Variable laser imaging (CLI), hologram, watermark, The specified paper type, Holographic overlay, UV activatable area, IR activatable region, Printed information, Microfonts, A security document verification device (1) according to claim 1, comprising at least one of the following.

3. The security document verification device (1) according to claim 1 or 2, wherein the smart inspection unit (7) comprises a detection unit (12) for determining the type of security document (3) and / or the security features (4).

4. The security document verification device (1) according to any one of claims 1 to 3, wherein the smart inspection unit (7) is expandable with a further trained learning module (8) for further security features.

5. The smart inspection unit (7) is By applying an image processing method to at least one image (5) of the scanned security document (3), one or more machine-readable security features (4) in the security document (3) are detected. The one or more trained learning modules (8) extract one or more parameters necessary to validate the detected security feature (4) against the learned patterns from the trained learning modules (8). A security document verification device (1) according to any one of claims 1 to 4, configured as described above.

6. The trained learning module (8) is a security document verification device (1) according to any one of claims 1 to 5, based on an artificial intelligence deep learning scheme.

7. The trained learning module (8) is Paper security parameters, Laser drilling parameters, Inkjet printing parameters, UV security features, Infrared security features, Holographic parameters, and / or Special security features, A security document verification device (1) according to any one of claims 1 to 6, which is trained to analyze the at least one image (5) of the scanned security document (3) based on one or more of the following:

8. The security document verification device (1) according to any one of claims 1 to 7, wherein the document scanning unit (2) is configured to scan in a first wavelength region of the electromagnetic spectrum, the first wavelength region being the visible light region, and / or the document scanning unit (2) is configured to scan in a second wavelength region of the electromagnetic spectrum, the second wavelength region being the infrared light region, and / or the document scanning unit (2) is configured to scan in a third wavelength region of the electromagnetic spectrum, the third wavelength region being the ultraviolet light region.

9. The security document verification device (1) according to any one of claims 1 to 8, wherein the document scanning unit (2) is configured to scan the security document (3) in high resolution and / or ultra-high resolution.

10. The security document verification device (1) according to any one of claims 1 to 9, wherein the output unit (9) is configured to provide the one or more security document feature levels (10) and / or the document forgery probability (11) on a display device (13) and / or transfer them to a backend unit.

11. A method for inspecting a security document (3) using a security document verification device (1) according to any one of claims 1 to 10, wherein the method is: The steps include scanning the security document (3), providing an image (5) of the scanned security document (3), and receiving electronic data (6) stored in the security document (3), The steps include: automatically inspecting the provided image (5) and the electronic data (6) in consideration of the detected security features (4) by using one or more trained learning modules (8) corresponding to the detected security features (4); A step of providing one or more security document feature levels (10) and providing a document forgery probability (11), Methods that include...

12. A method for inspecting a secure document according to claim 11, wherein the one or more security document feature levels (10) are determined as the result (14) of an evaluation processed by a corresponding inspection learning module (8) for the security feature (4), the result (14) being either a success in detecting the security feature (4) or a failure in detecting the security feature (4), and the result (14) is preferably provided as a percentage value.

13. A method for inspecting a secure document according to claim 11 or 12, wherein each trained learning module (8) provides one result (14), and the multiple results (14) evaluated by the smart inspection unit (7) are organized as a normalized vector output, the vector being normalized by the number of its components.

14. A method for inspecting a secure document according to claim 13, wherein if the amount of the normalized vector is greater than a predetermined threshold, the scanned security document (3) is verified and / or authenticated.

15. A computer-readable medium storing instructions for causing a programmable processor to perform the method steps according to any one of claims 11 to 14.

Images