CH716082A1 - Procedure for user identification. - Google Patents

Abstract

A computer-implemented method for identifying a user using an identification document (1) comprises the following steps: Receiving first image data from a terminal (2) on an identification server (3), the first image data including at least part of the identification document (1) with a photograph (11) and a security feature (12-15); Checking the authenticity of the identification document (1) by comparing the security feature (12-15) depicted in the first image data with generally applicable properties of the security feature (12-15); Extracting the light image (11) from the first image data; Receiving second image data from the terminal device (2) on the identification server (3), the second image data depicting a part of the face of the user; Checking a correspondence of the user with an owner of the identification document (1) by determining a similarity between the photo (11) and the part of the face of the user; and if the results of the tests are positive, the identification of the user is recognized as correct by the identification server (3).

Description

Field of invention

The invention relates to a computer-implemented method for identifying a user on the basis of an identification document and to an associated system.

background

In processes or actions to which only a certain person is authorized or in which the establishment of the identity of a person is central, the problem of identifying the person regularly arises. Examples are opening a bank account, purchasing a SIM card for mobile phones or performing certain services with public authorities. Conventionally, the identification is carried out in the presence of the person and with the aid of an identification document, e.g. an ID card, an identity card or a passport.

In the case of identification without the presence of the person, e.g. via the Internet, the challenge is to be able to determine the identity of the person safely and unequivocally without looking at the person or the identification document. In particular, a method for identification should therefore offer as few possibilities for manipulation as possible.

It is therefore an object of the present invention to provide a method with which a person can be identified safely and unequivocally, even if the person cannot be visually identified by another person.

Presentation of the invention

[0005] The object is achieved by a computer-implemented method for identifying a user using an identification document. Identification is understood to mean, in particular, the establishment of the identity of the user or, in other words, the verification that the user is who he claims to be. The identification document can e.g. be an ID card, identity card, biometric ID, passport or driver's license.

The method comprises the following steps, in particular in the specified order: Receiving first image data from a terminal on an identification server, the first image data depicting at least part of the identification document with a photo and a security feature: The terminal is preferably a device that is available to many users, e.g. a mobile phone, notebook or desktop PC with a camera to record image data. The identification server preferably comprises a system for data processing at a party interested in the identification or at an external service provider. In one embodiment, the first image data comprise a photograph of a page of the identification document. A security feature is understood to be a feature of the identification document which proves the authenticity of the identification document, in particular since it is difficult for unauthorized persons to reproduce or manipulate. Typical security features are an engraving, an embossing, a security thread, a kinegram, a hologram or an optically variable device (OVD). Checking the authenticity of the identification document by comparing the security feature depicted in the first image data with generally applicable properties of the security feature: The security feature is advantageously characterized by a property that is difficult, i.e. only with great effort, to be reproducible or manipulated and which is common to the general public or at least trained Staff is known so that the presence of the property can be easily checked. The property can e.g. be a nature of the material or an appearance of the security feature. In other words, the generally applicable properties are given with the definition of the security feature by an editor of the security feature or follow directly from the definition. The generally applicable properties are therefore implemented in particular in each security feature that follows the definition. Extracting the photo from the first image data: The photo is usually a portrait or passport photo of an owner of the identification document. In one embodiment the light image is processed by an image processing method, e.g. Edge detection or classification, extracted from the first image data, in particular the photo. The extraction of the photo can be carried out on the identification server or on the terminal device. The step of extracting can also take place at a different point in time, e.g. before receiving the first image data or after the following step. Receiving second image data from the terminal on the identification server, the second image data depicting a part of the face of the user: the second image data thus in particular include an image that shows at least part of the face of the user. The second image data are preferably generated while the method is running for identification. In particular, the second image data thus comprise a photographic recording with a part of the face that the user creates of himself while the method is running, advantageously with the camera of the mobile radio device. To prevent manipulation, it is furthermore advantageous that a liveness detection is carried out when the second image data is created, that is to say, in particular, that it is established that the second image data actually depicts the part of the user's face at the current point in time. Thus, e.g. it is avoided that a photo of the owner of the identification document is transmitted as the second image data, but this photo does not match the user. Checking a correspondence between the user and the owner of the identification document by determining a similarity between the photograph and the part of the face of the user: The similarity is determined in particular by image processing methods, e.g. by a correlation, by feature detection or by a distance metric, determined with statistical methods, e.g. neural networks. If the test results are positive, the identification server recognizes the user's identification as correct: The tests relate in particular to checking the authenticity of the identification document and checking that the user agrees with the owner of the identification document, as described above. Further tests, which are advantageously carried out before the identification is recognized as correct, are described below as preferred embodiments. If the identification of the user is recognized as correct, the identification server preferably emits a signal which confirms the correct identification and is in particular a prerequisite for further steps such as opening a bank account.

[0007] Advantageous embodiments include the following features: <tb> <SEP> In one embodiment, the method further comprises the steps <tb><SEP> <SEP> - Determination of identity data from the first image data: Identity data can e.g. a name, an address, a date of birth or an identification number of the holder of the identification document. Identity data can be determined from the first image data in particular via image processing methods, e.g. with text recognition. This step can also include the assignment of further identity data to the identity data determined directly from the image data, in particular in that the further identity data are retrieved from a database in which they are linked to the specific identity data. <tb><SEP> <SEP> - in the case of positive results of the tests, assigning the identity data to the user: If the identity data is determined from the image data, the user does not have to enter them in a time-consuming manner. This enables the identity data to be determined in a time-saving manner, in particular also in standardized form as they are present on the identification document and / or in the database. <tb> <SEP> In a further embodiment, the method additionally comprises the steps <tb><SEP> <SEP> - Determining a type of identification document: The type can be, for example, an ID card, an identity card, a passport or a driver's license. Furthermore, the type can be provided by an institution issuing the identification document, e.g. a national authority. The determination of the type can again be carried out via image processing methods, e.g. Feature detection and classification can be determined with statistical models, or banally via user input. <tb><SEP> <SEP> - loading of the general properties of the security feature belonging to the type of identification document: the type of identification document is usually associated with the presence of at least one specific security feature in a precisely defined form. Thus, if the type is determined, the security features and their properties present in this type can e.g. be loaded from a database.

The next two embodiments relate to the nature of the security feature and how this can be used in the identification method: <tb> <SEP> The security feature is preferably designed in such a way that it looks different from different viewing angles, in particular that a color or a shape represented in the security feature changes. In order to be able to check this feature, the first image data include several images of the security feature from different viewing angles. Examples of such a security feature are a kinegram, a hologram or an OVD. The multiple images from different viewing angles are achieved in particular by tilting the terminal device relative to the identification document or by moving the terminal device relative to the identification document between the images. The user is preferably guided by issuing instructions on the terminal when taking the plurality of images.

Alternatively or additionally, the security feature of the identification document can be designed so that it looks different when a lighting changes, in particular that a color or a shape shown in the security feature changes. In order to be able to check this feature, the first image data include several images of the security feature with different lighting, in particular with different lighting brightness. Examples of such a security feature are in turn a kinegram, a hologram or an OVD.Thus, the lighting of the security element is preferably changed between the recording of the images, be it by dimming or switching a light source on / off, by changing the angle to the incident light, by a Change of distance to a light source etc.

In one embodiment, the terminal is a mobile device with a light source with at least two brightness levels. At least two images of the first image data are correspondingly recorded at different brightness levels of the light source. Since it is available to many users, the terminal is preferably a mobile phone with a flash.

[0011] Further advantageous embodiments are distinguished by at least one of the following properties: the security feature comprises at least one of the following features: a frame around the photograph, a pattern or an engraving, the security feature is at least partially attached to the photo and includes in particular one of the features mentioned above, the security feature comprises an OVD or a kinegram.

Another aspect of the invention relates to a system for data processing, in particular an identification server, comprising a processor which is adapted to carry out the method described. Still further aspects relate to a computer program, comprising instructions which, when the program is executed by a computer, cause the computer to carry out the described method, and a storage medium with the computer program.

Brief description of the drawings

[0013] Further refinements, advantages and applications of the invention emerge from the dependent claims and from the description that follows based on the figures. Show: <tb> <SEP> FIG. 1 shows an identification document, a terminal and an identification server in one embodiment of the invention; <tb> <SEP> FIG. 2 shows a flow chart with a method for identifying a user according to an embodiment of the invention; <tb> <SEP> Figures 3 to 5 each an identification document and a terminal in further embodiments of the invention;

Ways of Carrying Out the Invention

Figure 1 shows elements that are used in a method according to an embodiment of the invention: an identity card (ID card, identification document) 1, a mobile phone (terminal) 2 and a server (identification server) 3. The ID card 1 is for example the Swiss ID and comprises a passport photo 11, a frame 12 around the passport photo 11, an engraving 13, e.g. the number “13”, and a pattern 14 of lines that partially run across the passport photo 11. Certain properties of the frame 12, the engraving 13 and the pattern 14 are precisely defined by the issuer of the ID card 1, e.g. the type of transition from passport photo 11 to frame 12, the nature and arrangement of the engraving 13 and the course of the lines of the pattern 14.This means that these elements can be used as security features, on the basis of which anyone or at least one person who knows the specific properties, the authenticity of the ID card 1 can check.

Furthermore, the ID card 1 in Figure 1 comprises a kinegram 15, the shape of which is reminiscent of a rock crystal in the case of the Swiss ID. The kinegram 15 also serves as a security feature, since it has defined properties and is difficult to reproduce. In addition, the ID card 1 includes identity data 16, in the case shown, the name “Hans Meier”. In general, other identity data 16 can also be noted on the ID card 1 in addition or as an alternative, e.g. an ID number or address. The identity data advantageously include 16 machine-readable characters as they are present on a machine-readable passport according to ICAO standard document 9303 or ISO / IEC 7501-1 in a machine-readable zone (MRZ) provided for this purpose. The ID card 1 can also have a different shape and comprise other elements, in particular in a different arrangement. In contrast, the defined and standardized properties of the ID card 1 of a certain type, e.g. a Swiss ID that enables a verification of authenticity in the first place.

The mobile phone 2 in Figure 1 includes a camera 21 for taking pictures and / or videos in a field of view 22 of the camera 21 and a flash (light source) 23. The server 3 includes a processor for executing the method and the necessary permanent and temporary storage. The server 3 is normally located at the institution that is interested in identifying the user, i.e. in the case of opening an account, e.g. at the bank or an authority. Alternatively, the server 3 can also be outsourced to a service provider who carries out the identification on behalf of the institution and at the end of the method forwards information to the institution as to whether the user was correctly identified or not. To carry out the method, the server 3 has a data connection 31 to the mobile phone 2, which is particularly suitable for the transmission of images and / or videos and comprises at least one of the following types of connection: cellular radio, WLAN, Bluetooth, LAN, Ethernet cable.

A method according to an embodiment of the invention, which can be carried out with the arrangement of Figure 1, is shown in Figure 2 as a flow diagram. The user who would like to identify himself to the institution takes a first picture of his ID card 1 with the camera 21 of the mobile phone 2 in step S1, on which the passport picture 11 and at least one of the security features 12 to 15 are shown. The first image is sent from the mobile phone 2 via the data connection 31 to the server 3 and received by the server in step S2. In step S3, the authenticity of the ID card 1 is checked on the server 3 on the basis of security features shown in the first image.

In one embodiment, the security features 12 to 14, which are related to the photo 11, are used for checking in step S3. In particular, it is checked whether, alternatively or cumulatively, the frame 12, the engraving 13 and / or the pattern 14 has generally valid properties that follow from the definition of these elements. If this is not the case, the ID card is classified as false, the identification process is aborted and negative information is output for identification in step S0. Alternatively or cumulatively, the kinegram 15 can also be checked by comparing its image in the first image with general properties.

The generally applicable properties are implemented as at least one criterion that can be tested with image processing methods on the basis of the first image. In this way, the general properties can be applied to the definition of the security feature itself, e.g. a precise geometric arrangement of the pattern 14, or to an optical effect which is caused by the security feature in a defined form and is visible in the first image, e.g. a color effect or a color change in the case of several images from the kinegram 15. The generally applicable properties and criteria can be established on the one hand on the basis of the definition of the security feature. On the other hand, they can use artificial intelligence with machine learning methods, e.g. by neural networks. In particular, the generally applicable properties and the criteria are established through monitored learning by showing the artificial intelligence various real and fake ID cards, each together with the information as to whether the ID card is real or fake, i.e. fake. Resilient and reliable criteria for authentication can only be defined on the basis of the first image.

If the ID card 1 is classified as genuine in step S3, the user makes a second picture of himself in step S4 of FIG. 2, on which his face or at least part of his face can be seen. The second image is in turn sent from the mobile phone via the data connection 31 to the server 3 and received by the latter in step S5. In step S6, a check takes place as to whether the person depicted in the second image matches the holder of the ID card 1, that is, in particular, is the same person as the person in the photo 11. For this purpose, the photo 11 is extracted from the first image. The extraction can already be carried out beforehand and in particular already on the mobile phone 2. In this case, the server 3 receives the extracted photo 11 from the mobile phone 2, e.g. in step S2 together with the first image or in step S5 together with the second image.

The verification of the correspondence of the user with the holder of the ID card in step S6 is done by determining a similarity between the photo 11 and the face or part of the face on the second image. The second image is advantageously replaced by a video or two second images, so that it can be determined that the person depicted in the second image is actually sitting live in front of the mobile phone 2 and is not only depicted on a photo that another person is in Camera 23 holds. This is e.g. brought about by the fact that the user moves during the video or between the two second images and then it is checked whether the movement process is realistic (motion detection). Such a liveness detection can alternatively also be achieved via further sensor data of the mobile telephone 2, e.g. by including a proximity sensor.

If the similarity between the face on the photograph 11 and the face on the second image does not exceed a certain level, the identification of the user is classified as incorrect, the identification process is aborted and negative information for identification is output in step S0. However, if the similarity exceeds a certain level, the identification of the user is classified as correct and, in step S7, positive information for identification is output accordingly.

The described method is preferably implemented in a computer program that runs on the server 3. This communicates via the data connection 31 with an application (app) or a web application in the browser (web app) on the mobile phone 2. The app is advantageously designed in such a way that it guides the user through the identification process, i.e. in particular outputs instructions, when and how you should take the first and second picture or video. Furthermore, it is advantageous that the app is implemented in a secure manner, that is to say, in particular, prevents manipulation of the first and / or second image or video and possibly the extracted light image.

Basically the user gives his personal data, e.g. Name, address, date of birth, e.g. at the beginning of the process in step S1 in the app. The data are also transferred to the server 3 and there e.g. by comparison with a recognized personal database, e.g. the Swiss Post, verified.

Instead of entering the personal data manually, an optional extension of the method in Figure 2 makes use of the identification data 16 on the ID card 1: The identification method starts in step S1 with the recording of the first image and without entering personal data. These data are then extracted from the first image, which also depicts the identification data 16. This can e.g. done in step S2 or S3. In addition, the server 3 can compare or enrich the extracted data with further data from a connected database. This is particularly helpful if not all identification data 16 can be extracted correctly from the first image.

Figures 3 to 5 show further embodiments of the method which make use of the generally applicable properties of the kinegram 15 for checking the authenticity of the ID card 1. The corresponding tests can therefore be implemented additionally or alternatively in step S3 in FIG. But they also have an impact on how the user takes the first picture or several first pictures or a first video in step S1 with his mobile phone 2, which are then received by the server 3 in step S2.

FIG. 3 relates to the property of the kinegram 15 of showing different colors or different shapes from different angles. Due to the standardized definition of the kinegram 15, this always takes place in a similar way, so that it is possible to derive from this in turn general properties and criteria of how a real kinegram looks on an image or video from a certain viewing angle. Correspondingly, according to FIG. 3, the user is asked to take two first pictures of the ID card 1 from different angles or a first video during which the user tilts the ID card 1 or moves the mobile phone 2 relative to the ID card 1. The tilting 24 of the ID card 1 about a defined axis, e.g. horizontally or vertically, since the user generally has to carry out the movement process as precisely as possible according to the instructions of the app, so that the generally valid properties of the kinegram 15 can actually be found in the first images or the first video.

Figure 4 shows a variant of the tilt 24 in Figure 3. Here, the user takes two first pictures at different distances between ID card 1 and mobile phone 2 or a first video, during which he performs such a distance change 25, so either the ID card 1 or the mobile phone 2 moves relative to the other object. In this case, too, characteristic color or shape changes of the kinegram 15 can be determined in the first images or the first video, which can be used in the test in step S3. This is in particular due to the change in the lighting and / or the field of view 22 of the camera 21 relative to the incident light.

Figure 5 shows a further variant, wherein no relative movements 24 or 25 as in Figures 3 and 4 are necessary, but the lighting of the ID card 1 is changed between the at least two first images or during the first video. For this purpose, the flash 23 of the mobile phone 2 is particularly suitable, which is controlled in such a way that it causes different brightnesses in the first images or during the first video. This embodiment is particularly easy to implement, e.g. the app takes two first images at short intervals, with the flash 23 being switched on in one image and off in the other. In addition, the brightness of the flash 23 can be controlled by the app in more than two stages, so that further optical effects of the kinegram 15 are visible on the first images or the first video.

While preferred embodiments of the invention are described in the present application, it should be clearly pointed out that the invention is not limited to these and can also be carried out in other ways within the scope of the following claims.

Claims (9)

1. Computer-implemented method for identifying a user on the basis of an identification document (1), comprising the steps - Receiving first image data from a terminal (2) on an identification server (3), the first image data depicting at least part of the identification document (1) with a photograph (11) and a security feature (12-15), - Checking the authenticity of the identification document (1) by comparing the security feature (12-15) depicted in the first image data with general properties of the security feature (12-15), - Extracting the light image (11) from the first image data, - Receiving second image data from the terminal (2) on the identification server (3), the second image data depicting a part of the user's face, - Checking that the user agrees with an owner of the identification document (1) by determining a similarity between the photograph (11) and the part of the user's face, - If the test results are positive, the identification of the user is recognized as correct by the identification server (3). 2. The method according to claim 1, further comprising the steps - Determination of identity data from the first image data, - If the results of the checks are positive, assign the identity data to the user. 3. The method according to any one of the preceding claims, further comprising the steps - determining a type of identification document (1), - Loading of the general properties of the security feature (12-15) belonging to the type of identification document (1). 4. The method according to any one of the preceding claims, wherein the security feature (12-15) is such that it looks different from different angles, in particular that a color or a shape shown in the security feature (12-15) changes, wherein the first image data comprise several images of the security feature (12-15) from different viewing angles. 5. The method according to any one of the preceding claims, wherein the security feature (12-15) of the identification document (1) is designed so that it looks different when the lighting changes, in particular that a color or a shape represented in the security feature (12-15) changes, wherein the first image data comprise several images of the security feature (12-15) with different lighting, in particular with different brightness of the lighting. 6. The method according to claim 5, wherein the terminal (2) is a mobile device with a light source (23) with at least two brightness levels, wherein at least two images of the first image data are recorded at different brightness levels of the light source (23), in particular wherein the terminal (2) is a mobile phone with a flash. 7. The method according to any one of the preceding claims, having at least one of the following properties: - The security feature (12-15) is at least partially attached to the photograph (11), - The security feature (12-15) comprises at least one of the following features: a frame (12) around the photo, a pattern (14) or an engraving (13), - The security feature comprises an OVD or a kinegram (15). 8. A system for data processing, comprising a processor which is adapted to carry out the method according to any one of the preceding claims. 9. A computer program comprising instructions which, when the program is executed by a computer, cause the latter to carry out the method according to any one of claims 1 to 7.