Method for authenticating a document

Abstract

An authentication method for authenticating a document from images obtained from the document, determining an indicator of document authenticity by verifying whether a meta-characteristic satisfies a predefined threshold associated with the meta-characteristic, the meta-characteristic being a measure of a difference.of a parameter of a detected element with respect to a reference parameter of a reference model; and also an authentication method for authenticating the identity of a person comprising performing the authentication method to authenticate a document of the person, the document being an identification document, and determining a similarity indicator by comparing a first face image and a second face image, using an image processing method; and authenticating the identity of the person depending on the result of the authentication method performed to authenticate the identification document and whether the determined similarity indicator is within a similarity threshold.

Description

description Method for authenticating a document FIELD OF INVENTION This description refers to methods, devices, and software products for authenticating a document. This description also refers to an authentication method for verifying a person's identity. BACKGROUND In recent years, digital registration and payment have increased worldwide. Portable devices, such as mobile phones, are replacing desktop computers as a payment tool. At the same time, criminals are developing new types of fraud to access personal data online. The authentication process is the most sensitive stage users face. In short, this process involves verifying that the person is who they claim to be and checking that the user's data is correct. These two types of validation are called personal validation and ID validation (e.g., an identity document). There are four rules for control: 1. If a DI is validated and the bearer is also validated, then the authentication process is correct. 2. If an ID is validated but the bearer is not, then the authentication process is incorrect. This occurs, for example, when a criminal steals a genuine ID and attempts to use it. 3. If an ID is not validated but the bearer is, then the authentication process is incorrect. This occurs, for example, when a forger modifies an ID and attempts to use it. 4. If neither the DI nor the bearer is validated, then the authentication process is incorrect. The authentication process can be performed using any of the three types of security measures present in the document (DI). Type one security measures are those that can be detected by the human eye. Type two security measures are those that can be detected by a specific sensor. Finally, type three security measures are those that are implemented but are only known to the manufacturer. Regarding the validation of a document, for example, an identity document or a banknote, it is a really difficult matter due to the following factors: • Differences between countries. • Differences in shape. The cut edges of the document have large tolerances. • Differences in printing. Each document (DI) is completed in different sections. • Physical condition: torn DI documents, printing errors (ink removed), folded documents, missing chip, etc. Identity documents such as ID cards are typically manufactured using various processes carried out on different machines located in different government buildings. The final stage is the printing of the user's data, such as name, surname, or date of birth. The variability of these different manufacturing processes results in significant differences between different documents of the same type. Regarding personal validation, the most common procedure involves taking a photo or video of a user and comparing it to a previously obtained photo or video. However, this personal validation procedure could be circumvented by showing a photo or video of the user (victim) obtained from social media in front of the camera. Therefore, it is necessary to obtain an authentication process to validate both the document and the person, which is more secure and more robust to implement than known authentication processes. US2003 / 099379 discloses an authentication method for authenticating a document comprising: receiving one or more images obtained from the document; detecting one or more elements of the document; obtaining at least one parameter of the detected elements by processing the one or more images obtained from the document using an image processing method; determining at least one meta-feature by comparing at least one parameter of the detected elements with at least one reference parameter of one or more pre-obtained elements; and determining an indicator of document authenticity by verifying whether the at least one meta-feature satisfies at least one predefined threshold associated with the at least one meta-feature. object of the invention Firstly, an authentication method is provided to authenticate a document. This authentication method for authenticating the document may include: receive one or more images obtained from the document; detect one or more elements of the document; to obtain at least one parameter of the detected elements, by processing one or more images obtained from the document using an image processing method; determine at least one meta-characteristic by comparing at least one parameter of the detected elements with at least one reference parameter of one or more pre-obtained elements; determine an indicator of document authenticity by verifying whether the at least one meta-characteristic satisfies at least one predefined threshold associated with the at least one meta-characteristic, in which A meta-feature is at least a measurement of a difference of at least one parameter of a detected element relative to at least one reference parameter of at least one reference model; the reference model comprising one or more pre-obtained elements. This authentication method for authenticating a document, for example a DI document or similar, is more secure and robust than the solutions of the previous technique since it is based on analyzing meta-characteristics that are a stronger security measure than the security measures typically implemented. A "meta-characteristic" is a measurement of the deviation or difference of at least one parameter of a detected element relative to the same parameter (at least one reference parameter) of a reference model. This deviation may be inherent in a random manufacturing tolerance of the process used to produce the element to be detected in the document, or it may be intentionally created. Meta-characteristics refer to all the parameters of the elements that make up a document, measuring deviations (differences) of these elements with respect to a reference model. These tolerances will be larger or smaller depending on the element and the manufacturing process used. Thus, any element subjected to a manufacturing process to be included in the document can be "meta-characterizable," that is, it can contain meta-characteristics. According to the present invention, certain parameters are used as a document security measure that are not, strictly speaking, security features. For example, differences in character spacing, alignment between text fields, color intensity, etc., have not been implemented as printed security features. Checksums, the presence of the CLI, OVI ink, kinegrams, etc., are security features that have been explicitly defined as such. This makes it more difficult for a counterfeiter, who can obtain information about explicitly developed security features, but has greater difficulty obtaining information about a non-security feature unless they know the manufacturing tolerances, which is simply very difficult. Another advantage of using the present method based on analyzing meta-features is that these meta-features can be a security measure whose working order of magnitude is a millimeter. The relative distances between characters, or the difference in height between them, can only be explained using measurements smaller than one millimeter. This differs from other security measures whose working order is based on millimeters, or even centimeters, such as the previously mentioned CLI, the kinegram, etc. The accuracy in forging the text of a counterfeit document should be on the order of one-tenth of a millimeter; this is undoubtedly a very complex task and may be virtually impossible unless the forger has obtained sophisticated machines similar to those used by government agencies. Furthermore, meta-characteristics can serve as a security measure to detect common alterations made by counterfeiters. These include altered birthdates or identity theft through scraping and subsequent writing of false identification. In some examples of the authentication method, comparing at least one parameter of the detected elements with at least one reference parameter of pre-obtained elements may include: to obtain at least one reference parameter by calculating at least one statistical element of at least one parameter of pre-obtained elements from a plurality of documents; wherein The predefined threshold comprises at least one statistical element. The benchmark can be used as a security measure that can only be determined by having a large database of genuine documents (statistical analysis). It might be easy for a forger to focus on manipulating a security measure if they can concentrate on a single document, for example, their own. But it is certainly more difficult to access a database of genuine documents to determine manufacturing tolerances. In some examples of the authentication method, comparing at least one parameter of the detected elements with at least one reference parameter of pre-obtained elements may include: Decode at least some of the detected elements to obtain a unique identifier for the document; access at least one database of reference parameters of pre-obtained elements using the unique identifier obtained from the document. Using the document's unique identifier and comparing it with a document issued with the same unique identifier (singularity analysis) allows reducing the permissible threshold for validation with much lower manufacturing tolerance threshold values. In some examples of the authentication method, at least one reference parameter of at least one pre-obtained element is obtained at or after the time of document creation by: detect one or more elements of the document; determine at least one parameter of at least one element of the document; Decode at least part of the detected elements to obtain a unique document identifier; store the unique identifier and at least one parameter of at least one document element in the database as at least one reference parameter. In some examples of the authentication method, the document to be authenticated can be, for example, an identity card, driver's license, credit card, debit card, business card, medical card, insurance card, passport, tickets, etc., or any other documentation capable of being used to, for example, identify at least one person or to be used in transactions such as banknotes, checks, etc., or similar. In some examples of the authentication method, the at least one element is at least one of: pattern and drawing element, text element, light-sensitive element, optical-sensitive element, infrared light-sensitive element, ultraviolet light-sensitive element, magnetic field-sensitive element, temperature-sensitive element. Most of the elements mentioned above can be detected using a conventional vision camera without necessarily requiring any special lighting, and only those elements sensitive to infrared light, ultraviolet light, magnetic fields, or temperature require special sensors for detection. This makes the present method extremely easy to implement. An authentication method for verifying a person's identity is also disclosed. This method may include: to carry out an authentication method to authenticate a person's document as described above, the document being a DI document; receive one or more first images of the person's face; receiving one or more second images of the person's face included in the DI document by processing one or more images of the DI document using an image processing method, said one or more images of the DI document being received by the authentication method performed to authenticate the DI document; determine a similarity indicator by comparing one or more first images of the face and one or more second images of the face using an image processing method; and authenticate the identity of the person depending on the result of the authentication method performed to authenticate the DI document and whether the similarity indicator is within a similarity threshold. With the authentication method cited above, authenticating the identity of a person by performing the authentication method described above to authenticate a digital document, it cannot be broken by showing a photo or video of the user (victim) obtained from social networks in front of the camera device as the solutions of the previous technique. This authentication method can be used, for example: - In digital banking, it's a way to acquire new customers remotely. It's also a way to update customer data that the bank already has, for example, when a customer's ID card expires. - The previous point can be extended to the registration process in other entities such as hotels, transport, notarial procedures, etc. - As a way to verify a person's identity and their ID document by relevant security authorities: police, military, private security, etc. - As a tool for controlling people at borders or airports. Specifically, it could be a tool used by an ABC (Automated Border Control System). In another respect, a software product is disclosed. The software product may comprise program instructions to cause a computer system to perform an authentication method, as illustrated by some examples disclosed herein. The product of a computer program can be materialized on a storage medium (e.g., a CD-ROM, a DVD, a USB drive, a computer memory, or a read-only memory) or carried on a carrier signal (e.g., an electrical or optical carrier signal). The computer program may be in the form of source code, object code, an intermediate source code and object code, such as in a partially compiled form, or in any other form suitable for use in implementing processes. The carrier may be any entity or device capable of carrying the computer program. For example, the carrier may comprise a storage medium, such as a ROM, for example a CD-ROM or a semiconductor ROM, or a magnetic recording medium, for example a hard drive. Furthermore, the carrier may be a transmissible medium such as an electrical or optical signal, which can be transmitted via electrical or optical cable, radio, or other means. When the computer program is materialized in a signal that can be transmitted directly by a cable or other device or medium, the carrier may consist of said cable or other device or medium. Alternatively, the carrier may be an integrated circuit in which the computer program is embedded, the integrated circuit being adapted to carry out, or to be used in the performance of, the relevant methods. BRIEF DESCRIPTION OF THE DRAWINGS Non-limiting examples of the present description will be described below, with reference to the accompanying drawings, in which: Figure 1 illustrates a flowchart of an authentication method particularly for authenticating a document (DI) according to the present invention. Figure 2 illustrates a flowchart for comparing at least one obtained parameter with at least the same reference parameter following a statistical analysis. Figure 3 illustrates a flowchart for comparing at least one obtained parameter with at least the same reference parameter following a singularity analysis. Figure 4 illustrates a flowchart of an authentication method particularly for authenticating a person's identity according to the present invention. DETAILED DESCRIPTION OF EXAMPLES Figure 1 illustrates a flowchart of an authentication method specifically for authenticating a Document (DI) 100, which can be performed using a device with a camera. This device can be any type of electronic device capable of taking photos, such as a mobile phone, an action camera, a tablet, a smartwatch, etc. The document could be, for example, an identity card, driver's license, credit card, debit card, business card, medical card, insurance card, passport, tickets, etc., or any other documentation capable of being used to identify at least one person, or even a banknote, check, etc., or similar document. Although this detailed description refers to a digital identity document (DI), it should be understood that it can be applied to any other document to be authenticated. The DI document is only one possible example for implementing the present invention. This authentication method can be performed by the device itself or by the device communicating with at least one server. In one architectural example, the device could be a mobile device such as a smartphone with a camera or similar; this option is feasible due to improvements in processor capabilities and other components developed in recent years. In a second architectural example, the device could be a mobile device with a camera equipped to establish data communication with a server. In this way, the mobile device could capture and send data to the server, which could then perform computation for the mobile device; that is, the server receives the captured / obtained data. Furthermore, the server could also have a database or be in data communication with a database.Once the data has been properly processed, it could be sent back to the device. Communication between the device and the server could be implemented using any system known to the art, whether wireless or wired. The device may be equipped with at least one UV, IR, and / or visible light source and sensors so that the elements in a digital information document (DI) can be captured and verified under different lighting conditions. Such element(s) may comprise at least one of the following classified items: • Pattern and drawing elements may include: • Logo. • Flag. • Anti-scanning patterns. • Background image patterns. • Barcodes. • Eyelets. • Rivets. • Fingerprints. • Guilloches. • Planchettes. • Cuts along the grain. • Safety fibers. • Safety thread. • Watermark. • Elements with bleeding ink. • Laser-readable camouflaged images (CLR). • OVDot holographic microparticles. • Iridescent laminates. • Overprint laminates. • Latent image mask. • Latent encoded image. • Moiré variable color (MVC). • Chips. • Text elements, depending on their quantity, may include: • Character. • Word (group of characters) . • Text field (group of words) . • Text elements, depending on their size, may include: • Text, visible to the human eye, which may include character, word, or text field. • Microtext (not visible or not clearly visible to the unaided human eye) that may comprise character, word, or text field. • Nanotext (size is smaller than microtext by at least 3 orders of magnitude) that can comprise character, word or text field. • Light-sensitive elements (sensitive to visible light) which may include: • Transparent window. • Diffractive element patch for diffractive identification device (DID). • Glossy laminate overprint. • Hologram images. • Kinegram images. • Retroreflective images. • Patterns with security print on the core (SCP). • Images of records viewed through. • Optically sensitive elements (orientation-sensitive) that may include: • Floating images. • Hologram images. • Kinegram images. • Latent filter image (LFI) . • Latent image in rotation. • PEAK latent image. • Multicolor latent image. • Laser-based image switching (CLI). • Multiple laser imaging (MLI). • Optically variable identification element (FEEL-ID & FUSE-ID). • Elements with optically variable ink (OVI). • Elements with optically variable magnetic ink (OVMI). • Optically variable Dynaprint printed image. • Infrared-sensitive elements (sensitive to infrared light) which may include: • Elements with anti-stokes ink. • Document numbers with IR characteristics. • Security fibers with IR-fluorescent ink. • Texts with IR-fluorescent inks. • Ultraviolet-sensitive elements (sensitive to UV light) which may include: • Barcodes with UV inks. • Document numbers with UV characteristics. • Elements with photochromic ink. • Hl-lites fluorescent with UV light. • Elements with fluorescent ink under UV light. • Magnetically sensitive elements (sensitive to magnetic fields) which may include: • Barcodes with ferromagnetic components. • Document numbers with magnetic components. • Thermally sensitive (temperature-sensitive) elements, which may include: • Elements with thermochromic ink. The manufacturing processes that can be used to create Dl documents and can produce the aforementioned meta-characterizable elements may comprise at least one of the following: • Offset printing. • Iris printing. • Orlov Print. • Letterpress printing. • Dot matrix printing. • Laser printing (xerography) . • Inkjet printing. • Dye sublimation printing. • Thermal transfer printing. • Thermographic printing. • Gravure printing. • Screen printing. • Lenticular impression. • Intaglio printing. • Mechanical erasure. • Mechanical drilling. • Chemical erasure. • Chemical metallization. • Laser erasing. • Laser engraving. • Laser drilling. • Laser metallization. • Folio printing. • Wet printing. • Dry printing. • Blind printing. • Electron beam lithography. • Slot pattern technology. • Additive manufacturing. Any other manufacturing process suitable for producing the meta-characterizable elements mentioned above may be foreseen. The authentication method for authenticating a Dl 100 document may comprise several stages, which will be described below: - Receive one or more images of document DI 101. The number of images may vary depending on the implementation; for example, one image of the front of document DI, or one image each of the front and back of document DI, etc. If the device includes a camera, then the one or more images are obtained by the device itself; that is, the camera takes photos and sends them to the processing media, which receives them. However, if the device includes, for example, a mobile device and a server, then the server receives the one or more images obtained by the mobile device. The obtained images may be stored in a database or similar system and may be received by the server's processing media. For example, to obtain one or more images of the digital document (DI), this method may involve determining when the camera is capturing a full view of the DI by using an AI-based image processing method. This method is trained to detect the DI and automatically trigger the camera when it determines that the full view of the DI is being displayed. This automatic detection of DI documents helps the user take the necessary photo under optimal conditions and eliminates camera shake caused by hand movement when the camera shutter is manually operated. If the server and the device are independent units, the server could send an instruction to control the device's camera. One possible solution for performing automatic detection for DI documents can be done, for example, through Local Binary Pattern (LBP) or HAAR descriptors. When one or more images of the digital document are obtained via the camera, this method may additionally include adjusting one or more focus and lighting conditions by regulating the camera's illumination unit. This can be done automatically by the device itself with the camera or by receiving the corresponding instruction from the server. The user does not need to adjust the lighting and focus conditions. Additionally, the method may involve adjusting a camera focusing unit to a fixed focus mode, so that the device can obtain a properly focused photograph of the DI document. At least one or more images of the DI document could be obtained by using different camera orientations relative to the DI document. The user may be notified by the device to change the orientation of the device and / or the camera through any means of alerting them, such as visual and / or audible signals and / or vibration. Changing the camera orientation relative to the DI document may reveal optically sensitive (orientation-sensitive) elements, as described above. To correct the document's perspective, an image processing method could be applied to both frontal and tilted images to detect the elements located on the document in a perfectly distortion-free image. The image processing method could be any method available in the prior art. Obtaining one or more images of the DI document may also include obtaining at least two photos of the DI document, where at least one is taken with flash and at least one is taken without flash to maximize the benefits of flash illumination of the DI document (in such a way as to detect light-sensitive elements). The at least two photos may be taken in a random order, separated by a random time interval, for example less than 500 milliseconds, which may be appropriately modified. To prevent fraud (replacing a counterfeit DI document with another DI document that has been artificially flashed), the photos can be taken and processed almost simultaneously, thus preventing the DI document from being replaced. To confirm that both images belong to the same DI document, various characteristics can be compared between them, including text, OCR data, and patterns. As an additional measure to prevent the substitution of the DI document between the two photos, the present authentication method may further comprise generating at least one video signal between the capture of the at least one photo taken with flash and the at least one photo taken without flash; and processing the at least one video signal using a motion detection method to detect whether the DI document has been substituted between the capture of said at least one photo taken with flash and said at least one photo taken without flash. Depending on the requirements of a particular DI document, the image could be obtained by taking into account the combination of a particular camera orientation related to the DI document and the use of flash to detect specific elements. All this makes it possible to work with sufficient image quality to study the elements of a document. - Detect one or more elements of document DI 102 and obtain at least one parameter of the detected elements 103, by processing one or more images obtained from document DI using an image processing method. This detection can be performed on the device itself or on the server that has received data from the device. An example of detecting one or more elements in the DI document may involve performing an image processing method by training one or more neural networks with a set of images using machine learning techniques in such a way that the output of those networks represents at least one parameter of the detected elements. Given a meta-characterizable element, it is possible to obtain at least one meta-characteristic of that element. For example, different types of parameters of the detected elements can be cited: alignment, spacing, width, height, color, texture, excess / deficiency of material... these differences from a reference model are called meta-characteristics. • The at least one meta-characteristic may comprise at least one measure of the difference in the parameter deviation from "alignment" between the deviation from alignment of at least two different elements of the detected elements and the deviation from alignment of the same at least two different elements of the reference model. Given an image of a digital intelligence document (DI), it is possible to measure the difference in alignment between two metacharacterizable elements that ideally would form a straight line. This reference straight line is taken from the reference model. This difference in alignment is called "meta-alignment". There is a direct relationship between distance measured in pixels and distance measured in micrometers, since the number of micrometers per pixel can be predetermined or known beforehand based on the camera's resolution. Therefore, it might be possible to determine and measure misalignment between elements by processing the number of pixels involved. The relative position of the elements is determined, and then the relative deviation is calculated. For example, a Spanish ID card 2.0 should ideally have the text fields for first name, last name, date of birth, and expiry date aligned. However, the IDESP field may be misaligned with the other fields because it was printed using a different process. Therefore, processing the pixels of the captured / obtained image can provide the information needed to determine and measure the misalignment between text fields on an ID card. • The at least one meta-characteristic may comprise at least one measurement of the difference in the parameter "spacing" between the difference in space of at least two different elements of the detected elements and the difference in space of the same at least two different elements of the reference model. Given an image of a digital intelligence document (DI), it is possible to measure the spacing between any two meta-characterizable elements. This spacing is compared against a reference value (parameter) taken from the reference model. For example, it is possible to measure the distance between text fields, the distance between each character, the distance between a chip and the Cl I, etc. This difference in spacing is called "metaspacing." As described above, it may be possible to determine and measure a space by processing the number of pixels involved. The number of micrometers per pixel can be known beforehand. • At least one meta-characteristic may comprise at least one measurement of the difference of the parameter "height" and / or "width" from a reference model of height and / or width of at least one of the detected elements. Given an image of a digital information document (DI), it is possible to measure the width and height of a meta-characterizable element by counting and determine which pixels belong to the element(s). This width and height are compared against a reference value (parameter) taken from the reference model. These differences in width / height are called "meta-width" and "meta-height," respectively. Therefore, processing the pixels of the captured image can provide the information to determine and measure the height and width of the meta-characterizable elements. • At least one meta-feature may comprise at least one measurement of the difference of the parameter "related to color and / or related to texture" from a reference color-model and / or texture-model of (a parameter of) at least one of the detected elements. These differences are called "meta-color" and "meta-texture" respectively. In some documents, it's possible to define a valid range for the ink used in printing certain elements, such as text character colors. However, working with absolute color values ​​is complicated due to their dependence on lighting. In other cases, it's possible to define the color range used in an element when it has been printed with a color gradient. This is easier to quantify since it measures differences in color, hue, or intensity in specific regions of the image. To measure these differences in color and / or texture, state-of-the-art techniques can be used. Sometimes, elements are not printed solidly or in a gradient, but contain a specific texture that has been printed intentionally or even unintentionally. In the latter case, these textures, which can be seen on some elements due to the ink printing processes carried out by the machine, are included. • The at least one meta-characteristic may comprise at least one measurement of the excess / deficiency difference of the "printing material" parameter from a reference printing model in at least one of the detected elements. The manufacturing process of the meta-characterizable element depends on the machine used and its tolerances. However, it also depends on other factors that can influence the ink deposited on the paper. The ink absorption process by the document depends on the paper's thickness, which also has tolerances. The ambient temperature and humidity at the time of printing influence the ink's drying rate. The amount of ink in the printing press's reservoir can affect the volume of ink expelled during the printing process. All these factors result in unique prints, where elements of the same type (for example, two characters of the same range, two images with a similar pattern, etc.) differ from each other due to an excess or deficiency of ink in these elements compared to the reference print model. Using image processing techniques, it is possible to evaluate these material excesses and deficiencies, allowing each element to be characterized individually and uniquely, distinguishing it from other elements of the same type. This difference between the reference print model and the actual element can be called "meta-material." The meta-material of an image element is a function defined on each of the n x m pixels of the image element and stores, in each pixel, information depending on whether the pixel corresponds to a material excess, a material deficiency, or a pixel whose value matches the element's reference print model. From the above, it follows, for example, that no two "A" characters in a digital document are identical, or that no two "1" characters in a digital document are identical. Or at least, the similarity is minimal. Characterizing an element of a digital document through the excess or deficiency of ink in relation to its reference printing model is a security measure for evaluating the similarity between elements of the same type in the digital document. For example, three characters with the value "1" printed on the same DI document. The first may have an apparent excess of ink on its right side. The second may be missing ink at the top. The third may not contain material defects as noticeable as the previous two. Therefore, using image processing techniques, ink excesses and shortages can be determined and measured in relation to a reference print pattern. The accuracy of meta-features depends on the resolution of the camera used. The higher the resolution, the greater the accuracy. The following summary table illustrates this relationship between accuracy and resolution: The table above shows a study for a text character height of 1 mm photographed at different resolutions. The higher the resolution, the fewer micrometers (micrometers) are represented in a pixel. If we consider that the ability to detect deviations using image processing techniques is, for example, 4 pixels, then the maximum achievable accuracy is obtained by multiplying the number of micrometers per pixel by four. The previous study shows that even with 3MPX resolution cameras, deviations greater than approximately 100 microns can be detected. This is an advantage over other security systems that are more dependent on camera resolution. The previous study was developed for a given distance between the camera and the digital image. If the distance decreases, the number of micrometers per pixel also decreases, increasing accuracy. Regardless of the height, the fundamental idea of ​​the study is to show the dependence of the meta-features on the camera resolution, and it is possible to obtain good accuracy, even a tenth of a millimeter, for low-resolution cameras. - Determine at least one meta-characteristic 104 by comparing at least one parameter of the detected elements with at least one reference parameter. This meta-characteristic can take any absolute value, and if there is no difference, that value will be zero. This determination can be performed by the device itself or by the server that has received data from the device. - Determine an authenticity indicator for the DI 105 document by verifying whether at least one meta-characteristic satisfies at least one predefined threshold associated with that meta-characteristic. This determination can be made by the device itself or by the server that has received data from the device. The information that determines the authenticity indicator can be sent to the device or can be available on the server itself. This information can indicate whether the DI document is authentic or not. Comparing at least one parameter of the detected elements with at least one reference parameter could be done using two different options (shown respectively in Figures 2 and 3): statistical analysis and singularity analysis. In either case, the reference parameters must be pre-obtained in order to calculate the measured meta-characteristics. Statistical analysis (Figure 2) of a set of non-forged documents makes it possible to obtain parameters for the manufacturing tolerances of each element of the document. Thus, for each element, one can calculate, for example, the deviation of the alignment relative to any other element, the dimensions (width and height), the spacing with other elements, the color and texture, and the characterization of excess and deficiency of material. The mean and standard deviation (statistical elements) of each of the aforementioned parameters (from 1 to M) of the pre-obtained elements are calculated measures that can be used as reference parameters; for example, the mean can be called avg_param and the standard deviation dev_param. The reference parameters can be obtained from 201 pre-obtained elements.The statistical element(s) used as a reference parameter(s) may comprise at least one of the following: mean, median, mode, standard deviation, or a combination thereof. Any other statistical element may be used, but in this example, the mean (or simply "mean" from now on) and standard deviation will be used. The predetermined number N of non-forged DI documents used to perform the statistical analysis may vary depending on the desired precision, but a larger number of DI documents analyzed results in greater precision. With this analysis, several reference parameters—that is, from avg_param1, dev_param1 to avg_paramM, dev_paramM—could be stored in a DB 202 database. This DB database may be stored on the device itself or on a server or similar device. Once the statistical analysis has been performed, its results are then used as a tool to validate the authenticity of any DI document. Now, for any DI document to be validated (DI_Y document), which has not been used in the statistical analysis, it is possible to calculate the meta-characteristics of at least one of its elements. That is, it is possible to calculate the difference of at least one parameter (for example, from parameter l' to parameter M') of at least one element obtained from the DI document to be validated with respect to the statistically inferred reference parameter. The meta-characteristic(s) (meta-characteristic1...meta-characteristicM) are then calculated by comparing at least one reference parameter of the previously obtained elements with at least the same parameter (parameter l' to parameter M'). These parameters are obtained from the element(s) of DI_Y document.For example, the predefined threshold associated with at least one meta-characteristic used to determine an indicator of authenticity of the DI 105 document can be a combination of the mean avg_param1...avg_paramM and the standard deviation dev_param1...dev_paramM associated with each parameter1...parameterM, or at least one of the mean and standard deviation. Figure 3 also illustrates a second, alternative option called singularity analysis. This analysis consists of several stages. First, a snapshot of a DI document (document DI_X) is taken as a reference model at time ti to obtain the parameter(s) of its 301 element(s). Then, a value is obtained that uniquely identifies this reference DI document from other DI documents of the same type, for example, an ID number, a serial number, or another unique value (unique identifier) ​​that may be ii present in the DI document. Subsequently, for each meta-characterizable element, the alignment relative to any other element, dimensions (width and height), spacing with other elements, color and texture measurements, and excess and deficiency of material from the document captured at time ti can be calculated. These calculated values ​​are used as reference parameters (parameter l to parameter M) and can be stored with the unique identifier in a DB 302 database on the device itself or on a server or similar. Subsequently, at time t2 (after time ti), the same procedure described above is performed, obtaining at least one parameter (from parameter l' to parameter M') from the DI document (DI_Y document) to be validated 303. The unique identifier of the DI document to be validated is decoded from the detected elements. This decoding can be performed by the device itself and / or the server. The set of parameters (or one of them) obtained at time points tiy t2 can be compared (this is a meta-characteristic) and verified to be close to zero and not exceed a certain threshold (the meta-characteristic is close to zero). Thus, the comparison of at least one parameter of the obtained parameters with at least the same reference parameter may include: decode at least part of the detected elements to obtain a unique identifier #ID of the DI document, for example the ID number at time t2; access at least one DB database or repository of reference parameters of pre-obtained elements using the unique identifier #ID cited above from the DI document. As described above, that DB database could be located on the server itself or it could be located separately. This uniqueness analysis allows determining an authenticity indicator of the DI document to be validated based on whether the calculated meta-characteristic(s) (meta-characteristic1...meta-characteristicM) 304 is / are within a predetermined threshold associated with the particular meta-characteristic(s) (with respect to the reference parameter of the DI document captured in ti). It can be seen from the preceding description that the need to detect a unique element such as the DI document number is motivated by the need to know what to compare it against. This occurs when a photograph of the model DI document is taken at time t1 at the police station at the moment of its manufacture or issuance. The parameters (alignment, spacing, etc.) are then stored in a database (on a device and / or server) along with the corresponding DI document number. If a user wants to perform the method of the present invention with a DI document (document Dl_Y) to be validated at time t2, an application for validating its authenticity can take the photograph, read (decode among the elements) the Dl number, and obtain the reference parameter for that Dl number. Then, the meta-characteristics are calculated, and an authenticity threshold is used to determine whether the document is valid. By carefully considering both the statistical analysis and the singularity analysis, it can be understood that in the first analysis, the differences in parameters for comparison (meta-characteristics) are of the same order as the manufacturing tolerances. However, in the second analysis, the meta-characteristics can be much smaller than the manufacturing tolerances, since the comparison is made between the reference parameters at time t1 and the parameters obtained at time t2 that characterize two documents D1 with the same unique identifier. In other words, in the second analysis, it is not the document read at time t2 that has variability limited by manufacturing tolerances. Rather, the detected variability is the same as that of the document read at time t1. In any case, the authenticity indicator can, for example, make a positive decision regarding the authentication of document Dl if the amount of deviation associated with the parameter (meta-characteristic) is less than or equal to a predetermined threshold. Conversely, the authenticity indicator can, for example, make a negative decision regarding the authentication of document Dl if the amount of deviation is higher than the predetermined threshold. Additional authenticity indicators for the Dl document can be determined, complementing the authenticity indicators calculated using meta-characteristic analysis. The authentication method can then further comprise: Determine at least one additional authenticity indicator of document Dl by verifying if at least one authenticity validation satisfies at least one predefined requirement associated with at least one authenticity validation. Authenticity validation may include at least one of the following: logical text validations, micro-writing validations, validations of at least one optically sensitive element, and validations of at least one light-sensitive element, or combinations thereof. Therefore, validation can be performed on safety elements that change their behavior under varying lighting and tilt conditions. Additionally, micro-intrusions are studied within this group. Logical validations are then performed on the information obtained from the OCR text fields. The human eye cannot detect and read the micro-writing (MW) present on a digital document without the aid of a magnifying glass. MW are widely used on security-obvious documents, as they can only be achieved through high-quality printing techniques. It is unlikely that a forgery of a digital document could have these micro-writings printed as well as the original. There could be micro-writing on both the front and back of the Di documents. All types of micro-writing (positive and negative, for example) are found on both sides. As an example, a Spanish ID card 2.0 features a kinegram. This kinegram displays the Spanish coat of arms in color or the letters "DNI" and "ESP," which change position depending on where the flash hits. The kinegram and the frontal photograph of the face are superimposed, making it more difficult to forge the image and kinegram. Therefore, photos taken with and without flash are analyzed to verify the consistency of the frontal image. Additionally, there are several raised elements that are only visible when the flash hits the document; one has the word "ESPAÑA" written on it, and another has the word "DNI" repeated several times along its edge and could be read using micro-writing techniques. Illuminating the CLI window with a flash produces three effects: the background color changes, the text appears more contrasted, and several vertical lines are revealed. The number of lines and the color variation follow certain rules, and the characters on the CLI must correspond to the expiration date (with a difference of years, depending on the bearer's age) and the first consonant of the surname and given name. Finally, but no less importantly, in the upper left corner, below the Spanish flag, there is text printed with Optically Variable Ink (OVI) that changes color. Illuminating it with a flash produces this effect, and the correct color variation is analyzed to certify the document's authenticity. Locating the fields of interest (name, surname, date of birth, etc.) in the digital document is not easy due to manufacturing tolerances (edges) and printing elements. Here, iterative procedures are used that, at each stage, can reduce the error in locating the field, similar to an image processing method. After locating the necessary items, adaptive filters can be key to character recognition. Appropriate pre-processing and cleaning can be performed to process the characters. Again, iterative operations can be run independently for each individual field until the expected conditions are met. For example, background cleaning, adaptive contrast techniques, and edge blurring and sharpening can be used at this stage to read each region. The position of each character, each word, and each group of words is determined for digital document validation. Furthermore, the authentication method may involve processing one or more images obtained from document Di using Optical Character Recognition (OCR) to extract text into a computer format. This extracted text may then undergo one or more logical validations to obtain a logical validity indicator. These logical validations may include detecting words within the extracted text and comparing them against a predefined dictionary. Names, surnames, etc., can vary greatly in length or the number of words they contain. The simplest way to determine if a read field is correct is to compare it against a dictionary. An n-gram-based algorithm is used to compare words. Each gram is evaluated, and the word with the highest metric similarity is selected as the correct one. There are several databases, each with a dictionary, one for each field (name, surname, province, and city). These databases can be dynamically updated to account for word frequency and can be stored on the device, a server, and / or a repository linked to them. Additionally, several fields are located on both the front and back of the data and can be compared with each other. Furthermore, there are also several logical sums on the back that can be used to determine the correct guess.All this prior work provides near-perfect OCR recognition. It is possible to verify whether the logical correctness indicator of the text is within a logical correctness threshold, which can be predefined for each case. This verification is done to determine the authenticity indicator of the Di document. If the document (Di) comprises a chip containing data, this authentication method may further comprise receiving or obtaining such data from or with the device itself, if the device is capable of communication. Alternatively, the device or server may be in data communication with a reader to retrieve data from the chip. This data stored on the chip may comprise at least text data, image data, fingerprint data, signature data, biometric data, and all data readable by OCR, among other valuable information. Data obtained from the document (Di) using image processing methods may be compared with at least one piece of data stored on the chip to determine the additional authenticity indicator of the document (Di). It should be noted that Near Field Communication (NFC) technology is just one non-exhaustive example of how to communicate pre-recorded data in the digital document. RFID-based technology or any other wireless technology could also be used. Additionally, an authentication method can be implemented to verify a person's identity. Figure 4 illustrates a flowchart of this method. As mentioned previously, this authentication method can be performed by the device itself, provided it has a camera, or by the device communicating with at least one server. This authentication method for verifying a person's identity may include: - Implement an authentication method to authenticate a DI 100 document from the person as described above. - To receive or obtain, via the camera, one or more initial images of the person's face. For example, receiving or obtaining these images may include a photograph or video of the person's face. If a video of the face is received or obtained, the method may also include obtaining a depth field or 3D field from the images of the person's face by processing the received / obtained video of the person's face using an image processing method. This method of managing the camera is the same as described above. Additionally, receiving or obtaining the first one or more images of the person's face may involve determining when the camera is pre-capturing a full view of the person's face by using an image processing method, and automatically firing the camera when it is determined that the camera is pre-capturing such a full view of the person's face. - Receiving or obtaining one or more second images of the person's face included in the DI 402 document by processing one or more images of the DI document using an image processing method, said one or more images of the DI document being received or obtained by the authentication method carried out to authenticate the DI document; - Determine a similarity indicator 403 by comparing one or more first images of the face with one or more second images of the face using an image processing method. For example, the similarity indicator can be determined based on a percentage of similarity between two images. - Authenticate the identity of the person (404) depending on the result of the authentication method used to authenticate the DI document and whether the determined similarity indicator is within a similarity threshold. This threshold can be appropriately modified to a default value. The main idea behind validating a person is to compare, for example, a validated photo of their face with an unvalidated photo of their face. A similarity indicator is then calculated. If the similarity indicator is greater than a threshold, the unvalidated photo is validated. Consequently, the person validation method is considered correct. Person validation, as part of the authentication process, requires a validated photo of their face. If the ID document validation process is correct, then the photo of the face printed on the document should be correct, and, for example, a selfie taken by the device could be used to compare the bearer's / owner's face with the validated photo of their face on the ID document. When the ID document comprises a memory chip containing image data accessible via wireless communication and the device comprises functionalities for establishing data communication with the memory chip (e.g. NFC), the authentication method, particularly for validating the person, may further comprise obtaining or receiving (from an external source) one or more third-party images of the person's face from the ID document chip via NFC communication between the device and the chip. As mentioned earlier, NFC technology is just one possible example for storing pre-recorded data in the DI document. Any other technology could be used, such as an RFID-based protocol. In this example, determining the similarity indicator also involves comparing one or more third images of the person's face with one or more first images of the person's face and / or one or more second images of the person's face using the image processing method. Although only a number of examples are described herein, other alternatives, modifications, uses, and / or equivalents are also possible. Furthermore, all possible combinations of the described examples are covered. Therefore, the scope of this disclosure should not be limited by particular examples but should be determined solely through a proper reading of the claims that follow. If a claim contains numerical references in parentheses to drawings, these are for the purpose of enhancing clarity and should not be construed as limiting the scope of the claim. Furthermore, although the examples described with reference to the drawings comprise computer apparatus / systems / devices and processes carried out on computer apparatus / systems / devices, the invention also extends to computer programs, specifically computer programs on or in a carrier, adapted to put the system into practice. The invention is defined by the appended claims.

Claims

Claims 1. An authentication method for authenticating a document comprising: receiving (101) one or more images obtained from the document; detecting (102) one or more elements of the document; obtaining (103) at least one parameter of the detected elements by processing the one or more images obtained from the document using an image processing method; determining (104) at least one meta-feature by comparing at least one parameter of the detected elements with at least one reference parameter of one or more pre-obtained elements; determining (105) an indicator of document authenticity by verifying whether the at least one meta-feature satisfies at least one predefined threshold associated with the at least one meta-feature, wherein a meta-feature is at least a measurement of a difference of at least one parameter of a detected element relative to the reference parameter of one or more pre-obtained elements.

1. An authentication method according to claim 1, wherein the comparison of at least one parameter of the detected elements with at least one reference parameter of pre-obtained elements comprises: obtaining at least one reference parameter by calculating at least one statistical element of at least one parameter of the pre-obtained elements from a plurality of documents; wherein the predefined threshold comprises the at least one statistical element.

2. An authentication method according to claim 1, wherein the comparison of at least one parameter of the detected elements with at least one reference parameter of pre-obtained elements comprises: decoding at least part of the detected elements to obtain a unique document identifier; accessing byat least to a database of reference parameters of pre-obtained elements by using the unique identifier obtained from the document.

4. An authentication method according to claim 3, wherein the at least one reference parameter of at least one pre-obtained element is obtained at or after the time of manufacture or issuance of the document by: detecting one or more elements of the document; determining at least one parameter of at least one element of the document; decoding at least some of the detected elements to obtain a unique identifier of the document; storing the unique identifier and the at least one parameter of at least one element of the document in the database as at least one reference parameter.

5. An authentication method according to any of claims 1 to 4, wherein the at least one meta-feature comprises at least one of: at least one measurement of a difference in aalignment deviation parameter between the alignment deviation of at least two different elements of the detected elements and the alignment deviation of the same elements with respect to at least two different elements of the reference model; at least one measurement of a difference in a spacing parameter between the spacing difference of at least two different elements of the detected elements and the spacing difference of the same elements with respect to at least two different elements of the reference model; at least one measurement of a difference in a height parameter between at least one detected element and the same element of the reference model; at least one measurement of a difference in a width parameter between at least one detected element and the same element of the reference model; at least one measurement of a difference in a color parameter between at least one detected element and the same element of the reference model.at least one element of the reference model; or at least one measurement of a difference in a texture parameter between at least one detected element and the same at least one element of the reference model.

6. An authentication method according to any of claims 1 to 5, wherein the at least one meta-feature comprises at least one measurement of a difference in a parameter of excess or deficiency of printing material between at least one detected element and the same at least one element of the reference model.

7. An authentication method according to any of claims 1 to 6, wherein the at least one element is at least one of: a pattern and drawing element, a text element, a light-sensitive element, an optically sensitive element, an infrared-sensitive element, an ultraviolet-sensitive element, a magnetic-sensitive element, or a thermally sensitive element.

8. An authentication method according to any of theClaims 1 to 7, wherein the one or more images of the document are obtained by means of a camera, and the authentication method further comprises: determining when the camera is capturing a full view of the document by using the AI-based image processing method, trained to detect said DI document, and automatically triggering the camera when the image processing method determines that said full view of the document is captured.

9. An authentication method according to claim 8, wherein the authentication method further comprises adjusting one or more focus and lighting conditions by regulating the camera's illumination.

10. An authentication method according to claims 7 and 9, wherein the one or more images obtained from the document are at least two photographs of the document, wherein at least one is taken with a flash and at least one is taken without a flash.A method for detecting light-sensitive elements, the at least two photographs being taken in random order, separated by a random time interval.

11. An authentication method according to any of claims 1 to 10, wherein the authentication method further comprises: determining at least one additional authenticity indicator of the document by verifying whether at least one authenticity validation satisfies at least one predefined requirement associated with the at least one authenticity validation.

12. An authentication method according to claim 11, wherein the document comprises a chip containing data, the authentication method comprising: receiving stored data from the chip; comparing the data received from the document with at least one piece of data stored on the chip; and determining an additional authenticity indicator of the document.

13. An authentication method for authenticating the identity of a personcomprising: carrying out an authentication method for authenticating a person's document according to any one of claims 1 to 12, the document being an ID document; receiving one or more first images of the person's face; receiving one or more second images of the person's face contained in the ID document by processing one or more images of the ID document using an image processing method, said one or more images of the ID document being received by the authentication method carried out to authenticate the ID document; determining a similarity indicator by comparing the one or more first images of the face and the one or more second images of the face using an image processing method; and authenticating the identity of the person depending on a result of the authentication method carried out to authenticate the ID document and whether the similarity indicator is within a similarity threshold.

14. AAn authentication method according to claim 13, wherein the DI document comprises a memory chip containing image data, the authentication method comprising: receiving one or more third images of the person's face from the DI document chip; determining the similarity indicator further comprising comparing the one or more third images of the person's face with the one or more first images of the person's face and / or the one or more second images of the person's face using the image processing method.

15. A computer device comprising a memory and a processor, implementation instructions stored in the memory and executable by the processor, the instructions comprising functionality for executing an authentication method according to any one of claims 1 to 14.

Images