DE102012205347A1 - Method and system for authentication and identification of objects - Google Patents

Abstract

A method for authenticating and identifying an object comprises the steps of: acquiring a first detection image of a selected inspection area (PB) of an object surface of the object (OBJ) for generating initial capture image data (EBD) representing characteristic properties of the object surface in the inspection area; Generating third image data (DBD) representing a third image different from the first detection image; Determining first comparison data (VD1) by a first comparison operation (V1) in which the first detection image data is compared with the third image data using a predetermined comparison process; Acquiring a test object image of a test area (PB) of an object surface of a test object (PRO) to be inspected for generating test image data (PBD), which represent characteristic properties of the object surface of the test object to be tested in the test area; Determining second comparison data (VD2) by a second comparison operation (V2) in which the test image data (PBD) is compared with the third image data (DBD) using the predetermined comparison method; and authenticating the test object by comparing the first comparison data (VD1) with the second comparison data (VD2) to determine authenticity characteristics, as well as by evaluating the authenticity characteristics.

Description

BACKGROUND

Technical area

The invention relates to a method for authentication and identification of objects, a system for authentication and identification of objects and system components for such a system.

State of the art

To secure products against counterfeiting or piracy, a large number of different methods are known. In order to identify and authenticate objects for the detection of forgery or to authenticate, individual and distinguishable characteristics of the object are usually determined and displayed mechanically or visually with the corresponding earlier (eg at the manufacturer or supplier or on the occasion of an expertise) on the original object Characteristics compared. As a rule, physical methods for the machine-aided collection of physical properties of the objects to be protected for identification and authentication at the object are used in order to be able to classify and identify them. Particularly common are optical codes which are applied to the objects by mechanical means (laser engraving, mechanical engraving, ink jet printing, screen printing, mechanical engraving, etc.). The codes are usually the identification in the foreground, while the codes are normally only secondarily a low reliable authenticity feature, since they work with technical means, the counterfeiters are potentially accessible and therefore - albeit with some effort and expense and sometimes only in the presence or spying on itself only the authorized manufacturer / user information available - can be reproduced and counterfeited.

For the authentication of objects are often features of aids that are connected in a suitable, not or only with destruction detachable form with the object to be protected or applied by suitable technologies on this. Omnipresent examples of this feature class are e.g. Hologram labels or embossed / laser-written holographic structures. Disadvantages are the at least fundamental reproducibility by unauthorized bodies (eg a counterfeiting company) and the necessity to spend additional manufacturing steps for the feature production, which must necessarily be technologically complicated, expensive and therefore expensive to develop the desired protective effect (difficult reproducibility). Many of these measures are very object-specific, because not every protective measure can work in the context of different materials and structures.

The patent US 5,067,162 describes, for example, an image correlation method for verifying the identity of objects. Using a reference image of a reference object, which is subdivided into a plurality of reference sections with characteristic properties, the identity of an object is calculated by means of autocorrelation and thus verified.

The WO 2009/097974 A1 describes a method for identifying or authenticating items due to manufacturing or processing random features. By "identification" is meant a process that serves to uniquely recognize an object. If an object is uniquely recognized, it can be uniquely assigned or it can be uniquely assigned to the detected object. For example, an identified product (object) can be assigned a price or its destination. Identification takes place on the basis of characteristics characterizing the object and distinguishing from other objects. By "authentication" is meant the process of verifying (verifying) an alleged identity. The authentication of objects, documents or data is the statement that they are authentic - that is, they are unchanged, not copied originals. Authentication also takes place on the basis of features characterizing the object and distinguishing from other objects. The features that are used for the authentication are preferably non-transferable, not copyable and not forfeitable.

In the present application, the terms "identification" and "authentication" are used in the manner defined above.

An Originals product security and traceability concept known by the abbreviation "O-PUR" proposes a labeling system for packaging paper and cardboard, but also products to identify clearly from metal and plastic. The measures focus on printing, embossing or engraving a product or its packaging with a standardized matrix code. The 5 × 5 mm "fingerprint" is designed to recognize a counterfeit reliably and with the simplest means.The idea is to exploit the individuality of the manufacturing process and to extract a unique feature code from its "fingerprint" to access it Way to identify and authenticate an original. The "fingerprint" is recorded in real time using special high-speed cameras and unmistakably stored in a database for each individual original product. "Fingerprint") is compared with the unmistakably stored in the database information and thus its originality or forgery clearly recognized.

In the conference article: "How to detect Edgar Allan Poe's 'purloined letter' - Or: Cross correlation algorithms in digitised video images for object identification, movement evaluation and deformation analysis", ISBN: 0-8194-4853-2, by Michael Dost, Dietmar Vogel, Thomas Winkler, Jürgen Vogel, Rolf Erb, Eva Kieselstein, Bernd Michel (March 2003), (In the following [Dost, Vogel 2003] the authors show that essential elements also in the patent DE 196 14 896 B4 also be used for the identification and authentication to prevent forgeries of objects worth protecting, such as products, art objects or forensic evidence. The method is based on acquisition and digitization of image recordings of the examination object and a reference object and uses an image comparison based on an analysis of the image recordings based on local two-dimensional cross-correlation calculation.

Previously known methods and systems for authentication and identification of objects have a number of disadvantages. For example, for the identification of millions to billions of objects, such as mass products, an extremely high file number and size of signature descriptions or image files for a later authenticity comparison of the object, which must be stored and handled in a logistical system, transferred or queried , Especially when this storage is done in a central database, appropriate access to the database must be ensured for identification, which causes communication costs and a time delay.

In particular, an identification in the context of a 1: n comparison is hardly practicable in the case of high numbers of disorderly stored signature or image files, since comparison algorithms yield even in the most favorable case calculation times that lead to unreasonable delays in millions of execution and not useful for mass balance can be used. Even in the case of authentication, the "correct" reference file must always be selected, which requires additional means not explicitly specified in the invention descriptions.

In addition to appropriate technical means for accessing a database, e.g. Internet connection, which possibly also allows the manipulation of data communication, are often also read-out devices with complex technology, e.g. costly special optics or sensors and radiation sources that use special electromagnetic radiation necessary.

Procedures and systems that require mandatory access to a database held by the manufacturer or authorized certification authorities are not self-authenticating.

If additional safety-specific markings are applied to the object, the protection against copying the mark is inadequate, especially in simple processes such as paper printing and marking of objects with little or poorly detectable intrinsic structure, as is often the case with metal products because the copying techniques used for counterfeiting purposes of unauthorized side, by advancing pressure-reproduction and engraving techniques continue to evolve until a protection is no longer guaranteed sufficiently secure.

Another disadvantage of previously known methods is that the reliability of the identification process is related to the degree of change of the object surface e.g. Due to contamination, wear, deformation and aging often decreases significantly, so that under certain circumstances, a real object is misinterpreted as a forgery.

Likewise, if necessary, technical special equipment required, such as sources for the generation of special electromagnetic radiation or special measurement setups by means of special sensors and sensors, can be used Radiation sources, not a universal use for the identification of a variety of different objects, since such methods are very object and material specific. Often they are also expensive and complicated in use.

TASK AND SOLUTION

It is an object of the invention to provide a method and a system for the identification and authentication of objects, which allows an evaluation of both natural object properties and artificially created object properties in such a way that the authenticity of an object, e.g. of products, documents, objects of value or objects of nature can be determined. A special objective here is an economical, robust method which permits object authentication and identification with comparatively low costs and very high reliability.

This object is achieved by a method having the features of claim 1 and a system having the features of claim 13. Further, a laser processing apparatus having the features of claim 19 is provided.

Advantageous developments are specified in the dependent claims. The wording of all claims is incorporated herein by reference.

Various aspects and features of the claimed invention will be explained primarily with reference to the method steps of the method. It is understood that the corresponding system for identifying and authenticating objects comprises the corresponding devices for carrying out the method steps.

The inventors have realized that it is surprisingly not necessary for an identity check or authentification to make a direct comparison of a current image of the surface structure (test pattern) recorded on the object to be examined with an image of the object. stored in a database to execute earlier initial capture image of the original. Rather, it has been found that it may be advantageous to identify and authenticate one to be performed on the one hand in the original capture, on the other hand in the probe capture (i.e., authentication), e.g. Correlation-based comparison with an arbitrary, in both cases same third image and a calculated description of detectable structure-related similarity relationship characteristics due.

In principle, any third-party images or third-party image data can be used. However, as a rule, images which have a sufficient gray-tone contrast, stochastic structural character and sufficient structuring on several levels of resolution are advantageous. In addition, the same size, image resolution and the same format in Ersterfassungsbild, Prüflingsbild and third image allow a simplification of the comparison process and its precise reproducible process. Experience has shown that these are well suited, e.g. certain known stochastic texture fill patterns from commercially available graphics programs.

Preferably, to identify an object, such features that can be ascertained on the object are used, which can be detected and reproduced as a two-dimensional matrix, that is to say as an image (bitmap) of values of a physical property varying over the surface. Preferably, the gray values of the surface in the region of interest are detected by means of optical imaging devices (objective) and video camera (image sensor matrix chip and read-out electronics). That is, the surface is positioned in an image capture device, illuminated if necessary with preferably adapted to the sensorial spectral sensitivity electromagnetic radiation. The reflected radiation is detected by means of the optical sensor matrix, which converts into preferably sequential electrical signals. These electrical signals are digitized in a known manner and finally transferred as an image file (preferably an uncompressed format such as *. Bmp) in a data processing device.

If necessary, with known image processing means, a conversion into a simple processable gray value format can take place, i. a picture format that now has a matrix of location-related intensities. However, similarities and correlation coefficients can also be determined directly from color images, whereby the color image data can be converted into suitable data values or data structures for the determination of the similarity.

An identity check or authentication can be performed in the form of a known comparison between image files by digital image correlation. In principle, in addition to the cross Alternatively, other known methods for calculating image changes and image structure similarities such as cluster, regression, or factor analysis may be used for this identity verification step

A preferred method variant is characterized in that the predetermined comparison method in the first and the second comparison operation comprises a calculation of a vector field which describes certain local similarity relations existing between the images to be compared in the vicinity of a number of predetermined measurement points. These vectors are referred to below as "similarity vectors". Each similarity vector is calculated such that it shows from the coordinates of a measurement point in the reference image to that point in the comparison image in which the calculation yields a singular value of the correlation coefficient, preferably its maximum. If the maximum is used, then the algorithm has found at the relevant image coordinates the comparison image pattern that is the most similar to the measurement point reference environment.

The method of comparison can essentially correspond to the method described in US Pat DE 196 14 896 B4 is used for another purpose, namely to determine deformation states in microscopically dimensioned test specimen areas, for which purpose the images of load-deformed test objects are obtained there and the similarity vector field is interpreted as the field of projection of the surface displacements into the image plane. The disclosure of this document is to this extent made by reference to the content of the present description.

If a cross-correlation-maximum-based similarity vector field calculation is performed, then the method yields a "vector field in which the direction and length of the vectors are randomly distributed, since the position of the" most similar pattern is the comparison of a surface structure image with a suitable third image, ie both the first and the second comparison "In a search environment around a measurement point in this approach is basically random and the analyzed fact underlying any real shifts that a shift-like, ie create a smooth, coordinated vector field.

If a similarity vector field is computed based on a first-captured original image with a correlation-maximum-based third image, this would be exactly identical to the pseudo-shift field that would exist between an ideally reproduced test image of the same original and the same third image. In the practical case, however, a completely exact reproduction is not feasible. In reality, even with very good repositioning, recording distortions, noise, or age-related, manipulative, or even (for example, in a manipulative intent) intended structural change of the examinee must occur. Therefore, in practice, only a minority K << J of a larger number J of vectors usually agree except for a mostly small offset-dependent offset in both components.

However, it turns out that even an unexpectedly small number of matching similarity vectors is sufficient to allow a very high degree of authenticity proof.

Already the analysis of a single measuring point can prove this: A comparison matrix of the side length 50 can be taken from a search environment of a measuring point of the side length 100 in (50 + 1) · (50 + 1) different ways. The probability that one of these extracted gray value structures most closely resembles an equal reference structure from the reference image (correlation coefficient = max) is then 1: 2601, i. if the pseudo-shift vector matches between the two compare operations, the probability is exactly 1: 2601 that this match occurs purely at random and the texture similarity indicated by the correlation maximum is not due to an actual object match, i. even an originality statement based on a single matching vector would already be extremely secure. According to the laws of stochastics, this certainty is dramatically increased by every other "right" vector from a whole. Already with agreement of only 5 from 25 vectors the risk of an incorrect originality assignment is only approx. 1: 100.000, even if for the component-wise conformity of the vectors an admissible tolerance range of 3 pixels is given due to remaining residual mispositioning.

The comparison with a third image via a similarity vector analysis carried out in the manner described thus offers very high security and is very robust.

The third image data, like the initial capture image data and the test image data, may basically be e.g. be generated by optical detection of a suitable third image. The third image data may then be e.g. stored as a bitmap or in any other processable image format.

However, it may be advantageous if, in the generation of the third image, the third image data is calculated according to a predetermined image generation algorithm. In this case, the third image data is uniquely determined by the generating data or parameters and by the image generation algorithm applied thereto. This can significantly reduce the amount of data to be stored or transferred for processing the third-party image data.

In a variant of the method, a third image is joined by tiling from a number of similar small square submatrices. This reduces the file size of the data needed for third-party image deployment. The tiling is particularly favorable in combination with the similarity analysis in the first and second comparison operation, since in this case the same submatrix of the (tiled) third image is always used when comparing the third image with the first capture image and comparing the third image with the inspection image at each measurement point can be.

In a further variant of the method, the entire third image or also a smaller submatrix used for tiling is generated by an image generation algorithm that uses as few, preferably two to five, generating parameters as possible. Possible and useful realization variants use e.g. a pseudo-random generator for calculating a gray value sequence on the basis of a seed parameter, whereby the same image structure always follows from one and the same parameter. Fractal algorithms (such as plasma fractals, perlin noise, apple males, cooking curves) are known and advantageous because of their graphical properties, in particular the structuring in several resolution levels.

The method and system, in a first level of security, enable robust self-authentication with high but conditional reliability, i. a verification of the identity or authenticity of a candidate on the basis of information available on the candidate or an accompanying document without (for example by remote data transmission) external information to be transmitted to the location of the examination, e.g. would have to be accessed from a database held by the manufacturer.

For this purpose, in a variant of the method, the first comparison data are stored in a marking associated with the object, which is to be referred to here as a "similarity signature". Once generated, they can be used at any time to determine authenticity statements, linked directly or in an accompanying certificate. The authentication thus does not require access to an external database.

The storage of the comparison signature on the object can take place in the form of an alphanumeric text or in a known graphically coded form, such as bar code or Data Matrix Code (DMC). Alternatively, the signature may also be included as part of a certificate accompanying the object, e.g. Similar to photo certificates for stamps. The generating parameters for the third image generation algorithm may also be added as part of this similarity signature.

Then, the originality property can be reliably detected at any desired time and place, based solely on the directly available properties associated with the DUT without access to external information sources.

Thus, the method is potentially self-authenticating, i. It is not necessary to resort to externally stored information when checking the identity via complicated data logistics. It also avoids the difficulty of making a one-to-many comparison when choosing the right one from a large number of large initial capture files, which would only be possible with lengthy datamining algorithms, leading to unacceptable delays.

In a second level of security, relying on an external data base of comparatively small data size, it is possible to achieve almost absolute certainty of proof of originality beyond the security of the first level. Thus, in the context of a tamper-evidence check according to the first level of security, self-authentication can be carried out with little effort, but if necessary (for example, if there is reasonable doubt about possible third-party manipulations) the absolute authentication can be resorted to with limited increased logistical and labor costs.

In the case of method variants which enable the second security level, the first comparison data are stored directly in a database separate from the object in the form of the numerical similarity signature or in its coded form (eg also as a checksum / hash code). Preferably, the first comparison data or its check code in the determination of the authenticity data then from the database read or retrieved. For example, the database may be maintained by the manufacturer of the object or by a certification authority.

In the second security level, information available in a reserved database should thus be able to be used in order to arrive at an almost absolute security of the authentication.

However, due to the favorable use of third-party image data, the information to be stored in the database can be much more compact (less memory space) than in the prior art and can be available in a readily accessible order, so that even 1: n comparisons can be carried out in a realistic time.

The second level of security allows proof of originality with absolute certainty, but not equally proof of forgery, since improper image capture or destruction of the original surface in rare special cases can intentionally or unintentionally fake a counterfeit by failing to achieve the predetermined similarity threshold in the comparison , In order to be able to prove the forgery property in case of need (eg in the case of legal disputes and damages caused by an alleged original), in a process variant with a third security level it is expedient to store the original image in a database held by the manufacturer, without However, this is included in a complicated data logistics for the transmission of data to the test site.

As a result, if necessary, a forgery-proof proof by direct correlation-based comparison of original and test surface image (possibly manually by an expert by means of direct correlation-based comparison of the original extracted original database with the test specimen image according to the in [Dost, Vogel 2003] described method) a largely reliable proof any falsification property which may be present if the surface in question has not been accidentally or deliberately damaged over the entire surface (which should normally be immediately visually discernible).

A large part of technical and natural objects has structures which are individually and relatively robust (ie, hardly modifiable by aging, handling or manipulation), can be visualized and captured and stored as an image file and therefore are suitable as an invariable feature of an individual object , The method according to the invention can use and analyze such natural structures for the purpose of identification and authentication. However, it is not limited thereto, but also provides methods and methods for solving the authentication task when such structures are not naturally available (e.g., glass, polished metal, polished silicon) and therefore need to be artificially generated.

A variant of the method thus provides for the generation of a marking on the object in the test area, the marking preferably being a non-reproducible marking, ie an individual marking which can only be generated once in the given form.

To equip objects with additional features, there are a variety of possibilities. The object may e.g. printed, pasted, engraved or combined with other objects in order to dispose of the properties suitable for the identification process to the necessary extent. In addition to the laser surface marking or in the case of (partially) transparent objects laser internal marking, special inks or even sand paper and magnetic particles or the targeted or random magnetization of magnetic object areas can be used for additional feature extension of the object. Likewise, a mark can be generated by means of electromagnetic radiation, e.g. by electron beam. A marker may be substantially two-dimensional and may be e.g. located on the object surface. Optionally, three-dimensional markings may also be used, e.g. can be generated by laser inside transparent materials near the surface

Preference is given to non-reproducible markings or markings, such as holograms, which have properties such as e.g. Shape or color can be varied and are well suited for noble luxury goods. Optionally, such a hologram can be combined with a non-reproducible marking. As a result, the non-reproducible marking can be covered by the nobly acting hologram, so that the design value of a product is increased and, at the same time, the position of the area to be compared on the object is characterized by the hologram.

For the authentication method described here, a particularly expedient variant arises when the artificially applied pattern has similar structural properties to a suitable natural surface. Important are sufficient contrast properties and a structure density possible on several levels of resolution as well as a stochastic character of the surface structure. Periodic structural components should be avoided, since in general there are several structurally undesirable structural regions.

While many known methods have the stated properties, they are reproducible in nature by the manufacturer, i. it can produce several similar structures (for example, in the case of one and the same structural shape or in the case of writing using laser or Focused Ion Beam (FIB) of a mathematically generated or prestored random bitmap). Thus, such patterns should not be used if e.g. The object of this is to determine whether a certain object has been unauthorized in larger quantities by an authorized manufacturer than authorized by the contracting entity (e.g., a marketer), since several identically-authenticated objects of the same genre may exist.

In a variant of the method, a laser processing apparatus is used to produce on an object to be marked a non-reproducible marking that can only be generated once due to the configuration of the laser processing apparatus in a given form. The marking preferably has a stochastic character of the surface structure substantially without periodic structural components.

The laser processing system is preferably configured so that the generated structures can not be reproduced even by the user of the system when properly and appropriately configured and used. This increases the security against counterfeiting.

In one variant, this uniqueness of the marking produced is achieved in that the laser processing device has a control unit and a scanner device connected to the control unit for the controlled deflection of a laser beam, wherein the control unit has a random generator for generating random signals and the scanner unit can be controlled on the basis of the random signals ,

Alternatively or additionally, the laser processing device may be equipped with a random mask device having a plurality of relatively freely movable mask elements and preferably an automatable means for exciting movements or for generating a rearrangement of the mask elements, wherein the mask elements are arranged in the beam path of the laser beam. The mask elements may be e.g. transparent microspheres or small opaque, i. act for the laser radiation impermeable particles. Due to the construction, the respective adjusting arrangements of the mask elements are unique or not individually reproducible.

Various embodiments will be explained in detail below. The foregoing and other features will be apparent from the description and from the drawings, the individual features each of which alone or in the form of subcombinations in one embodiment of the invention and on others Be realized areas and can represent advantageous and protectable embodiments. Embodiments of the invention are illustrated in the drawings and explained in more detail below.

BRIEF DESCRIPTION OF THE DRAWINGS

1 shows a schematic flow diagram of a method for authentication and identification of an object according to an embodiment of the invention;

2 shows an illustration of a preferred comparison method that works with a similarity analysis to comparative images;

3 shows a computationally generated third image, which was created by tiling from similar, equally sized submatrices based on less generating parameters;

4 shows an embodiment of a laser processing system with a random mask device for generating non-reproducible stochastic markings;

5 shows examples of random mask devices;

6 shows further examples of random mask devices;

7 shows a random arrangement of glass microspheres in a random mask device;

8th shows marks with random structures;

9 shows a random pattern generated by means of a scanner device driven by random signals;

10 shows a data matrix code;

11 shows the data matrix code 10 with indicated by arrows possibilities of displacement of submatrices, which can be coded by the relocation additional information.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

1 shows a schematic flow diagram of method steps of a method for authentication and identification of an object according to an embodiment of the invention. The reference symbol OBJ designates an object whose authenticity is to be checked at any time in the future with the aid of the method. During the later test, a test object PRO is checked, which may be the original OBJ or a counterfeit (possibly under the imitation of the original).

In a first detection, a first detection image of a selected inspection area PB of a suitable object surface of the object is detected.

In this process step, first detection image data EBD representing characteristic properties of the object surface in the inspection area PB are generated. The detection can take place, for example, optically with the aid of a first camera KAM1 or another suitable image capture device. For this purpose, the image capture device and the object are brought into the correct relative positioning to each other, so that the test area can be detected by means of the detection device. Possibly. For this purpose, a separate object handling device can be used. If the initial detection takes place in a sufficiently bright environment, a separate lighting can be dispensed with. It is also possible to provide a separate illumination for this step, ie an external irradiation with electromagnetic radiation of a suitable wavelength range. The captured image or its image data can be converted, for example, into a gray-scale image format, transmitted to a data processing system DVA and stored as first-capture image data EBD.

Furthermore, third-party image data DBD, which represent a third image which differs from the reference image, is generated (temporally before, at the same time or later in time). This process step can be carried out before or after the reference acquisition or at the same time as this.

The third image or the third image data can be generated in different ways. For example, a real third image can be recorded with a suitable detection device, for example a camera, and the recorded signals can be converted into third image data. It is also possible to generate third-party image data purely on the basis of generating parameters. The different operations for generating a third image are represented by a symbol with reference numeral ODB.

Once there is initial detection image data EBD and third image data DBD, a first comparison operation V1 can be performed. With the aid of the first comparison operation V1, first comparison data VD1 are obtained by comparing the first detection image data with the third image data using a predetermined comparison method. A preferred comparison method, in which a correlation-based similarity analysis of specific image areas is essentially carried out, is explained in more detail below.

In order to be able to evaluate the authenticity of a test object PRO to be tested, a test object image of a test area PB of an object surface of the test object to be tested is acquired in a test version for object authentication. In this case, similar or different detection devices can be used as in the reference detection described above, for example a second camera KAM2. Based on this Detection operation is generated by digitizing test image data PBD representing characteristic properties of the object surface of the test object under test in the test area PB. The test image data are kept ready for data processing, if necessary they are stored in a data memory.

Once the test image data PBD is present, second comparison data VD2 can be generated by a second comparison operation V2 by comparing the test image data PBD with the third image data DPD using the comparison method already used in the calculation of the first comparison image data.

As soon as the first comparison data VD1 and the second comparison data VD2 are present, the authentication of the test object can be carried out in a further method step. For this purpose, an authentication operation AUT is carried out in which the first comparison data VD1 and the second comparison data VD2 are compared for determining authenticity data. In this case, a mathematically evaluable similarity measure of the first and second comparison data is determined. The authenticity data is then evaluated according to a predetermined rating scale. If the similarity measure is above a predefined threshold value, then it is assumed that the test object PRO is identical to the original ORIG, so that authenticity is present (OK). If, on the other hand, the predetermined threshold value is not reached by the similarity measure, the test object is very unlikely to be identical to the original, so that a counterfeiting is assumed (NOT OK).

It can be seen that the first frame image data and the check image data are not compared directly, but indirectly via the respective comparison with the identical third image data DBD.

A preferred cross-correlation-based comparison method for comparing images or associated image data is explained in more detail below. In the method, images or image data are compared with one another in a data processing system. The comparison provides data and inferences on pattern-related similarity relations and differences of the compared images, which allow different conclusions to the observed objects depending on the fact embodied in the image.

This comparison method is used in object detection in the first comparison operation V1, i. when comparing the reference image data with the third image data, used to obtain the first comparison data VD1. In an analogous manner, in the context of the object authenticity check, the same comparison method is used in the second comparison operation V2, i. when comparing the test image data with the third image data, used to determine the second comparison data VD2.

For the computer-aided execution of a comparison, virtual measurement points, preferably at the nodal coordinates of an orthogonal equidistant grid with a grid spacing P pixels, are assumed on the digital gray value images. In the vicinity of the virtual measuring points, a cross-correlation evaluation between reference and comparison image is performed such that for a reference square at the measuring point preferably square reference matrix of the side length n pixels i and each of the comparison image in the preferably square Meßpunktumgebung, the so-called search environment of the side length m, removable submatrix of the same size a correlation coefficient is calculated in the manner described below.

For a typical page length of the image of 1000 pixels, for example, reasonable values for N, n, and m may be 49 <N <300, 15 <n <100, 50 <m <300 with n <m. Even deviations from these values are possible. In particular, the reference matrices and search environments associated with the adjacent measurement points may also overlap.

berechnet. Die Korrelationskoeffizienten ergeben dadurch ein zweidimensionales diskretes, d.h. nur an den ganzzahligen Pixelpositionen definiertes Wertefeld mit (N – n + 1)·(N – n + 1) Werten, das Aussagen zu gewissen bildstrukturellen Ähnlichkeitsrelationen des grafischen Inhalts der jeweiligen Messpunktumgebungen enthält. Aus diesen komplexen Feldumgebungen können kompaktere Teilaussagen zu Ähnlichkeitseigenschaften extrahiert werden, indem singuläre Punkte des ähnlichkeitsrelationenbeschreibenden Korrelationskoeffizientrenfelds aufgesucht werden. Als besonders stabiles und reproduzierbares Kriterium kann das absolute Maximum des Korrelationsfelds als derjenige Bereich interpretiert werden, dessen grafische Umgebungsmatrix der Seitenlänge n im Vergleichsbild die größte Ähnlichkeit zur lokalen Messpunktumgebungsmatrix der gleichen Seitenlänge des Referenzbilds enthält. Anwendugsfallabhängig zweckmäßig können auch Verfahrensweisen sein, die gewisse komplexere mathematische Feldeigenschaften näher analysieren, wie z.B. die Lage oder Lagen- und Größenrelationen von Feldminima, lokaler relativer Maxima oder Minima, Sattelpunkte etc. Between the gray values of the reference matrix indexed in ordered order according to their positions after the positions i and j in the matrix and every possible comparison matrix which can be taken from the search environment of the side length N, the correlation coefficient according to the formula now becomes

calculated. The correlation coefficients thus result in a two-dimensional discrete value field, ie, only at the integer pixel positions, with (N-n + 1) * (N-n + 1) values containing statements about certain image-structural similarity relations of the graphical content of the respective measurement point environments. From these complex field environments, more compact partial statements can be extracted into similarity properties by finding singular points of the similarity relation describing correlation coefficient field. As a particularly stable and reproducible criterion, the absolute maximum of the correlation field can be interpreted as the region whose graphic environment matrix of the side length n in the comparison image contains the greatest similarity to the local measurement point environment matrix of the same side length of the reference image. Depending on the application, it may also be advantageous to analyze procedures which more closely analyze certain more complex mathematical field properties, such as the position or position and size relations of field minima, local relative maxima or minima, saddle points, etc.

For an illustrative, better processable representation of the analysis result, a field of vectors can be used which, starting from the measurement point jump, point to the coordinates of the respective singular point, preferably the coefficient maximum. In the following, this vector field should be referred to as a similarity vector field. If there are real structure similarities between the compared images, associated with e.g. load-related displacements, the maximum coefficient coefficient vector field can be expediently understood as a projection of the displacement field on the object surface into the image plane.

Both comparison operations V1 and V2, when using the cross-correlation-based comparison, respectively provide comparison data VD1 and VD2 in the form of such a discrete similarity vector field. For the process principle, it is irrelevant how the order of comparison of the images is, i. E. it can alternatively the object image or the third image as a reference or as. the comparative image of the cross-correlation analysis can be used.

Since the original image and the specimen image do not have any structurally corresponding graphic patterns with respect to the third image, the similarity analysis does not provide a "displacement-like", somewhat smooth and continuous, similarity vector field. Rather, for each measurement point, the algorithm finds a purely random similarity vector independent of its neighbors, which points to the region of the local measurement point environment of the comparison image which is randomly similar to the reference sub-matrix. Consequently, in both comparisons, a chaotic jumble of similarity vectors of different stochastically distributed lengths and directions, such as those in FIG 2 third row, illustrated examples of similarity vector fields.

If an ideal reproduction of all image acquisition conditions were technically feasible, an exactly identical reproduction of the first capture image would result in the test piece image PBD of an actual original. It is readily apparent that then both comparison processes V1 and V2 would provide the exactly identical comparison vector fields VD1 and VD2 and thus a unique identity indication.

However, in the real technical realization, errors in reproducing the image pickup conditions are unavoidable. It comes to mispositioning, image distortion, image noise, lighting differences, aging or damage caused structural changes, etc., so that even in the actual existence of the original object whose sample image PBD can no longer be completely congruent with the first detection image. In this case, due to the error, some / several or even many similarity vectors between the first-capture image data EBD and the test image data PBD are calculated differently and only a few vectors remain identical except for a reposition-related tolerance of a few, preferably 3 to 5 pixels.

This is the case in the similarity vector fields in 2 , third row, first column (OBJ) and second column (PRO) shown. The two images of the fourth line only show the matching vectors, the error-related different similarity vectors have been hidden for better clarity of the graphical representation.

Based on 2 a possibility of practical realization is explained. The first column (left) shows an initial capture image on the original OBJ in the first row, and the third image used for the comparison operations in the second row (cf. 3 ), in the third line, the similarity vector field (corresponding to the first comparison data VD1) resulting from the first comparison V1. The second column shows the corresponding images or data which were recorded on the test object PRO during the authenticity check and by the second comparison operation was determined. In the second column, the test object corresponds to the original, so that the test should verify its authenticity. The third column shows the corresponding data for a test object PRO 'that does not correspond to the original. The two images of the fourth line only show the matching vectors, the error-related different similarity vectors have been hidden for better clarity of the graphical representation. It can be seen that there are no matches in the forgery (PRO '), while for the original 15 of the 25 similarity vectors match sufficiently.

In this case, the first capture image, the specimen image and the third image each have a page length of 500 pixels. The images are subdivided into 5 × 5 square submatrices of 100 pixels each side. In the middle of each sub-matrix is the measuring point, so that an equidistant measuring point grid with the grid spacing of 100 pixels is present. From the center of the sub-matrix, a gray value reference matrix of the side length 30 pixels is taken from the third image. In the search environment = sub-matrix of the page length 100 pixels, the gray-level structure of the page length 30 pixels that is most similar to the extracted reference matrix is then searched in the original or test object image. Here the correlation coefficient has a maximum.

The vector that can be given between the measurement point and the location of the most similar gray value structure is the local "similarity vector" for the relevant measurement-point-surrounding submatrix. In total there are 25 similarity vectors. Their fields are shown in the third and fourth lines, respectively.

When checking an object for authenticity, a statistically significant match of at least some of the stored displacement vectors must be achieved to identify an object as "real" or OK. Some erroneously determined "bad vectors" must not affect the analysis result. The second column in 2 illustrates this situation in which the original was taken in the test, ie in the examination registration. As has been shown above, a small number of valid similarity vectors is already sufficient for a high statistical certainty, so that in this realistic example it may be assumed that there is an absolute probability bordering on absolute certainty.

On the other hand, if it is not an original, then usually none of the similarity vectors will match the original recordings from column 1. At best, there are very rare random matches, which are taken into account by a minimum threshold of valid vectors of eg 5 for the proof of originality. This situation is in 2 shown in column 3.

However, since the same situation of undetectable vector matching can not be ruled out unfavorably, if gross errors are made in the reproduction of the image acquisition conditions (such as an incorrect image capture) or if the object structures were damaged unintentionally or for manipulative reasons, the proof of the In theory, the counterfeiting property in the first two levels of security can not be managed with the same high level of security as the original proof of originality, which means that the third level of security described above can be connected if required.

However, experience in process testing has shown that the method is extremely robust in this respect if the underlying patterns are sufficiently structured, contrasted and robust, and if the objects and the method are handled properly.

The self-authentication described in the first level of security, due to its complex nature and difficult traceability, as well as the unique, highly individual character of the imaged object structures, already provides a high level of security against manipulation, which can be further enhanced by encryption algorithms known per se for determining the similarity information to be written. However, manipulation by criminals can not be ruled out insofar as they spy out and successfully understand the methods used and the method-based secret data.

In order nevertheless to enable absolute security of the authentication, it may be provided in a second security level to store the identity similarity signature determined on the original or an encryption of the same or a security check code (hash code) determined therefrom in a queryable database held by the manufacturer or a trusted authentication authority. On the basis of an access to this database, it can be determined during the check whether the code found on the object is authorized, i. whether an object corresponding to this code was actually present at the manufacturer's or authorized agency's reference capture.

A similar similarity signature record (The similarity signature contains, for example, as 64-bit check code encoded eg 25 vectors as two-digit (ie pixel-accurate) vector components and preferably one to five third image generating parameters) on the one hand far less extensive than that of the prior art underlying capturing images or structure descriptions (such as edge profiles on printing margins). It is also numeric or alphanumeric in nature and can therefore be stored in a corresponding numerical order and thus easily recovered without complex comparison algorithms. A manipulation by criminals is not possible without intrusion into the data logistics (eg manipulation of the database itself).

The method described so far permits the proof of the originality property with almost absolute certainty, however, as already explained, not the proof of forgery property, since a counterfeiting can always be intentionally or unintentionally falsified (insufficient or no identical vectors) due to improper image acquisition or destruction of the original surface ,

The correlation algorithm used by the method can be varied in many ways. So it is e.g. possible to perform a correlation-based similarity calculation based on the fast Fourier transformation. There are also Wavelett-based computation variants in which additional pattern relationships are taken into account and therefore also provide acceptable comparison results for patterns that are not optimally structured.

Non-correlation-based similarity measures can also be used, such as e.g. Similarity analysis by image difference, least squares method, cluster analysis, etc. Present experience and the mathematical principles, however, show the cross-correlation analysis, possibly in one of its FFT or wavelet variants, as a preferable similarity analysis method.

Expedient method variants may result if the comparison algorithm uses not only the absolute correlation maxima (i.e., the pseudo-shift vectors) for the proof of originality, but also further characteristic information describing the calculated local correlation coefficient fields. Noise and distortion can lead to erroneously deviating maximum-based similarity vectors even when an original is present at certain measurement points. Nevertheless, it can be assumed that even at the location of the original agreement there is still a relatively high - albeit not the highest - similarity, so an analysis at this point will result in relatively high correlation coefficients, at least in comparison to the environment. A total erroneously insufficient overall agreement of the vectors could thus be replaced by the presence of a sufficient number of local relative maxima and thus the erroneous pretending of a forgery prevented.

Different procedures can be used for image acquisition. As far as technically possible in the context of the application, it is possible to reposition the test objects as exactly as possible in order to ensure the detection of one and the same surface area (test area) during the initial recording (on the original) and during the test recording (on the object to be tested). To obtain the relevant original and DUT images, a certain defined area can also be cut out of larger, higher-resolution images.

Finally, in an expedient variant of the method, the relevant surface area can be identified by optically recognizable orientation elements (eg, a frame, points or crosses delimiting the target image field, installation angles, or the like) in such a way that it is possible to use known means of digital image processing Cut out the image area from a larger environmental image and, if necessary, equalize it so that images of reproducible areas with the desired size and resolution are made available to the greatest extent possible. This variant of the method makes it possible under further conditions (camera resolution, optical quality, sufficiently concise surface pattern, low motion blur, appropriate handling, eg image area as possible filling relevant areas), the capture of the images for the originality test with widely available image capture cameras (eg digital or web cameras, notebook cameras, Pen or eye-glasses cameras, cell phone cameras, flatbed scanners) if necessary, make "out of hand" and make the originality proof in the appropriate device (eg via the mobile phone processor).

In a further expedient variant of the method, which is based on 3 is explained, basically any third image is constructed by tiling of similar, equal submatrixes. At each measuring point, the (indirect) comparison of the first recorded original image or of the original image is then carried out DUT image with the third image with always one and the same - in itself any - reference pattern. An advantage here is that the file size of the third image is reduced and a comparison can always be made with one and the same reference submatrix recognized as being particularly suitable.

In particular, this variant can be made more expedient if one does not work with fixed and stored image sub-matrices, but instead the relevant image structures are provided by a calculation according to a defined, preferably stochastic or fractal algorithm. In this case, no bitmap must be stored, but in the data processing systems, only a corresponding image generation algorithm must be implemented. In many cases, the use of such image generation algorithms, which emanate from very few, preferably one to five parameters, but which provide complex graphics, depending on the default parameters, is particularly advantageous. For this purpose, e.g. Pseudo-random number generators that provide a sequence of numbers (so-called seeds) that are the same but random in number for one and the same seed, or fractal algorithms, such as those that are randomized. Plasma fractals, perlin noise or so-called cooking curves.

The use of calculated fractal or stochastic patterns eliminates the need to store a third-party comparison graph. The third image data DBD then contained only the information about the generating parameters, possibly even information on the image generation algorithm. It can thereby provide a large number of different comparison graphics in parallel for the method and make them available by transferring the generating parameters (e.g., with a known image generation algorithm as part of the optical code associated with the object).

3 shows by way of example a 5 × 5-tiled third image produced from a cooking curve fractal, which has proven to be particularly suitable for the process because of its stochastic structure, the structuring on several levels of resolution and the good light-dark contrast.

The use of virtual, calculated third-party image structures creates an additional advantageous security feature, since it is no longer sufficient for a counterfeiter to know a particular third graphic as a key since a large number of different, ever new graphics can be provided and used as a basis for the correlation-based comparison. Instead, a counterfeiter must be able to know the algorithm, implement it, and spy on the meaning and current actual values of the generating parameters, which can be protected by suitable coding methods and transmitted in advantageous method variants via secure connections.

Technical objects and natural objects, such as Natural stones / minerals usually have a wide range of property types, e.g. Surface structures, cracks, color gradients or patterns which are very well suited for a large number of measuring methods. A homogeneous glass body such as e.g. A high-quality glass pane has a much smaller spectrum of image-varying properties because it usually consists of a comparatively small number of chemical substances, which are comparatively homogeneously distributed and the object is transparent to a large wavelength range of light. Accordingly, a mapping according to the method with conventional optical measuring methods, which is e.g. use a camera system and daylight sources as lighting for such glass only insufficiently possible. Thus, either better suitable, usually more complex measuring methods must be used or additional features in such an object or be applied.

In a preferred variant of the method, a special laser processing device LBV is used to produce on an object to be marked a non-reproducible marking that can only be generated once due to the configuration of the laser processing device in a given form. Such a marking should have a stochastic character of the surface structure substantially without periodic structural parts. With proper handling of the laser processing system, the generated structures can not be reproduced even by users. 4 schematically shows an embodiment with a random mask device ZM.

The laser processing device LBV has an object holder OH for receiving the object OBJ to be marked as well as a laser system LAS with a laser radiation source LQ and a beam guidance system SF for guiding a laser beam which impinges in focused form on a surface OB of the object. The laser beam coming from the laser radiation source and widened by means of a beam expander (possibly after passing through a controllable shutter and / or a controllable attenuator) from the mirror arrangement of a galvanometer scanner SCN in the direction of a focusing optics OPT (eg an F- Theta optics), which in the case of direct structuring focuses the laser beam onto a location in a marking region of the surface of the object to be marked or, in the case of a mask projection method, projects an image of a mask into the marking region. In the mask projection method, a mask is arranged in the object plane of the focusing optics OPT.

The random mask device ZM, which is also referred to below as a "mask" in a simplified manner, has a multiplicity of mask elements that are freely movable relative to each other and a device for stimulating movements or for generating a rearrangement of the mask elements. The mask elements are arranged in the beam path of the focused laser beam in the vicinity of the object surface OO to be marked.

In the embodiment of 5 For example, the random mask device has a plurality of miniature glass spheres GK having a diameter of less than one micrometer up to about 1 millimeter, preferably 1 to 10 microns, in a dispersion of water, isopropanol or other liquids of suitable viscosity on the underside of a transparent plate or disc GS can be adhered by adhesion and cohesive forces such as by glass, by dripping or brushing so that they form a random arrangement not to be reproduced. They can preferably be arranged with an automatic device such as a mixer arm, stirrer or vibrator to a constantly new pattern.

The distance A between mask (i.e., random mask device) and workpiece is between 0.5 to 6 mm, preferably 1 mm to 3 mm, depending on the embodiment.

Supplementary devices for rearrangement of the miniature glass spheres GK are also possible, e.g. a turntable on which the mask is affixed and which after a marking operation is rotated so as to provide a new mask and on the last used mask in a different position, by means of an automatic device such as a mixing or robotic arm, with a corresponding tool for mixing, the miniature glass balls are rearranged.

In a simple case, such a device has a glass pane GS and miniature glass spheres GK. A scanning laser beam LS can impinge on the mask at different locations and at least partially irradiate them. If a miniature glass ball GK is hit, the laser beam LS at the miniature glass ball GK is refracted such that after the exit a focusing and subsequently a beam widening SAW takes place whose energy intensity for an ablation on the object OB is too low.

If the laser beam LS hits the mask at a position without miniature glass spheres GK, the laser beam with the desired focus diameter FOC is imaged on the object and, depending on the selected process parameters, allows the ablation, i. a material removal.

New random, non-reproducible arrangement of the miniature glass spheres GK can be generated by the action of kinetic energy with random amount and direction on the mask. For this purpose, randomly controlled stirring systems are particularly suitable, which, in addition to a mechanically designed stirrer, can also be designed with nozzles for gas jets for stirring. The addition of miniature glass spheres GK in dispersion by means of nozzles or robotic arm is also conceivable, in particular if the transparent pane GS is cleaned at intervals in order to avoid the dripping off of excessive amounts of dispersion.

One advantage of the dispersion variant lies in the fact that due to liquid evaporation, such a dispersion steadily reduces in volume and can thus automatically ensure continuous mixing and rearrangement of the miniature glass spheres GK. As a result, it is not possible for a user to produce two or more structure markings having the same structures even after switching off an existing stirring or vibrating device.

In order to prevent the mask from being contaminated by ablation products, a transverse jet can be arranged between the mask and the object by means of a supersonic nozzle and suction or an aerodynamic window. Alternatively, a protective film FOL be provided on unwinding and winding rollers ROL, which can be moved continuously or discontinuously depending on the degree of contamination. In this case, the film may be inclined to the laser beam, e.g. 5 °, so that it can not hit the protective film FOL vertically, if the film is soiled with reflective material and thus a reflection of the laser beam could occur back to the laser beam optics.

If the mask is on a turntable for quick replacement, then the protective device must not be carried along by means of protective foil and unrolling and winding rollers.

With this mask, a non-reproducible, stochastic, constantly changing and unmistakable marking on arbitrary objects can be generated, whereby in the case of transparent or partially transparent materials in particular also an individual marking in the object interior is possible.

In the embodiment in FIG 5B the miniature glass spheres GK are enclosed in the dispersion in a transparent chamber KM of transparent material and set in motion and rearranged via at least one nozzle DS by random control of the pressure and the injection time. In this case, nozzles DS with a variable outlet opening and direction are preferably used in order to also use these variations for the random generation of miniature glass ball arrangements. The at least one nozzle DS pumps either only the liquid of the dispersion or the entire dispersion with miniature glass spheres GK in a closed circuit with the transparent chamber KM. Depending on the design, a plurality of nozzles DS may be mounted on several sides of the transparent chamber KM and driven at random. Devices for protecting the mask can as in the embodiment to 5A be applied.

Such a generated arrangement of miniature glass spheres is in 7 to see. On the mask, the miniature spheres have been arranged randomly in areas of single-layer island groups IG1 with free areas FB, in which arrangements AN of single or a few miniature spheres can be located, as well as in larger groups GG, which can also grow into multilayer groups MG. Thus, there is advantageously always a structuring over several levels of resolution.

In a further embodiment ( 6A and 6B ), a laser beam LS is directed onto an object or workpiece OB, wherein in the beam path of the random mask device ZM is a chamber TK made of transparent material such as glass, the bottom of which was partially covered with non-transparent, preferably fine-grained material MAT. This material preferably consists of materials which are resistant to bombardment with laser radiation, such as miniature ceramic or metal balls, although other geometric shapes such as chips can also be used. Optionally, these absorbent objects may be in a liquid which is actively cooled via a circuit or actively cooled when a certain temperature is reached. If the laser beam LS strikes the mask and is partially absorbed and / or reflected by non-transparent material MAT, marking on the object is only possible with a weakened energy intensity. By contrast, the marking can be carried out with full intensity if the laser beam LS strikes the mask in a region without nontransparent material MAT and thus can pass freely. If the laser beam LS strikes completely non-transparent material MAT, then this is completely absorbed and possibly reflected, without it being possible to interact with the object.

In order to ensure a random, non-reproducible arrangement for the non-transparent material MAT, an energy input via a vibrator drive unit RT in the form of random shock or vibration waves in the chamber TK can be initiated, creating a new random disorder of the non-transparent material MAT is guaranteed. Also, a change in position of the chamber TK, for example, by rotation of a symmetrically constructed chamber, the non-transparent material can randomly rearrange in a short time. To protect the mask from ablation products, the known protection devices of the embodiment can 5 be used.

In other variants, the drive unit MAGAG when using magnetic material by one or more permanent magnet or electromagnet MAG as in 6B be replaced. In this case, permanent magnets are guided via a randomly computer-controllable device such as a robot in defined areas of space along the chamber TK, in order to rearrange the magnetically transparent material MMAG randomly and not reproducibly. The embodiment is preferred with at least one electromagnet arranged laterally on the chamber. By varying the magnetic field strength and duration, as well as by reversing the polarity of the electromagnets random motion pulses can be induced on the magnetic material from one or more positions to rearrange this randomly. This works particularly well when several electromagnets are arranged laterally on the chamber TK and controlled by computer randomly in a predetermined parameter field. In the simplest case, such a computer control enables pulsed magnetic fields which whirl up or through the magnetic material MMAG.

As a result, random, non-reproducible markings can be generated. 8th shows in 8A and 8B shows two examples.

Random, non-reproducible object markers can also be generated by random triggering of a laser scanner. In doing so, e.g. generated in the control unit CON of the laser processing device by PC controlled random generator control signals for the scanner mirror. As a result, the laser beam is directed to random positions of the object surface at which then the processing of the object takes place punctually with a plurality of laser pulses. In addition, the temporal correction offset values between scanner motion and laser emission are adjusted in a manner that allows the laser to be emitted before the scanner mirrors have reached the rest position at the desired location of processing. With high traversing speed between the individual machining positions and moderate repetition rate (e.g., 100kHz - 500kHz) of the laser, this results in a "tail" of dots before the actual machining position, which goes into line as the scanner mirror motion decelerates. The direction of the tail is dependent on the relative orientation of two successive processing points. The length of the tail, in turn, depends on the degree of retardation of the scanner levels, the rate of deflection at the onset of emission, and the laser's repetition rate.

The object is marked during the alignment of the scanner elements or scanner mirrors by the laser beam. Depending on the position of the source and destination random coordinates, random markings as in 9 ,

Another possibility of random rearrangement or rearrangement is thermal transport of the glass beads as a result of convection, similar to a lava lamp. In some cases, the thermal energy introduced by the laser can be used for this transport or even controlled for this transport.

It is also possible to use object-borne deterministic surface patterns of artificial nature, which consist essentially of optical codes that can be generated with different writing methods, for marking objects. Two-dimensional matrix codes are preferred, such as a Data Matrix Code (DMC). An example is in 10 shown.

It has been shown that by applying the correlation-based displacement analysis method to such surface patterns not only intentionally introduced authenticity information but also other, preferably numerical, digital information can be concealed and re-read, using the existing technical facilities for writing or reading the optical codes so that no additional expenses and additional facilities are needed.

For this purpose, the two-dimensional codes to be written or written are manipulated by slightly displacing and / or also rotating certain optically resolvable areas of the same relative to their standard normal position. The size of the regions is preferably chosen such that they contain a plurality of code structure elements (eg individual blocks) and the displacement can thus be read out via a correlation-based displacement analysis. As tests have shown, areas of the order of five by five blocks are usually suitable for displacement analysis. 11 Figure 12 shows the scheme of partitioning a data matrix code into nine sub-sections of size 5 x 5 blocks and corresponding intervening non-relocated "quiet areas".

The size of the range shifts is preferably chosen to be such that it can be determined quantitatively on the one hand by correlation-based displacement analysis, on the other hand so small that the matrix code can still be read with the conventional means and the range shift does not yet strike the eye directly, i. remains hidden. In order to resolve these and accidental, e.g. To be able to lift write technology-related block shifts, sufficiently high-resolution image capture methods and sufficiently accurate write technology are required. Practically, e.g. Shifts in the order of up to about 1/10 block width appropriate.

If, for example, displacements in four lateral and four diagonal directions are permissible, then a shift range (including zero shift) can embody or represent ten different numerical values. If the shift amounts are included, 19 numerical values / range can be coded for two possible shift stages. A DMC code after 11 can thus be an eight-digit numeric Carry information if one of the shift ranges remains as a non-shift reference range (preferably the one in 11 marked with "5" middle range.). The encoded information may include an authentication signal or any other information without the usual user or even a potential forger being able to perceive the fact of any additional information present.

The reading of the contained information can be done by correlation-based displacement analysis between the image of the DMC / optical code present on the object and a standard code generated virtually from the content DMC information, which does not naturally contain the additional displacement-coded information. Both code images are expediently brought to the same image format by scaling and equalization beforehand.

In order to prevent the unauthorized reading out of digital information, all digital codes (one or more dimensional codes on the object) as well as all digital data in a database (eg formulas for calculating reference images, pictures in the original) as well as the entire internal and external digital data communication ( eg between readout device and an optionally external database) can additionally be digitally encrypted and have a digital and / or electronic signature for additional authentication. As encryption algorithms, in particular AES, Twofish, Serpent and their cascading can additionally prevent unauthorized reading out of digital information, optionally with a password or digital key.

By applying a correlation-based displacement analysis method or similarity analysis method to specifically modified object-borne deterministic surface patterns of artificial nature consisting essentially of optical codes (e.g., Data Matrix Code (DMC)), digital numerical information can thus be concealed and read again. The encoding of authentication information is just one of many applications. This proposal is independent of other methods of this application feasible. For example, a use in connection with a third image is not necessary.

However, an "ideal" code image calculated from the coded matrix contents, which is compared with the real code image, can be understood or used as a variant of the third image according to the method. The additional authenticity information concealed in the DMC can, as described in the method, be analyzed and authenticated by comparison with a third-party image of its own. An advantageous variant results if the third image used is not a stored third image or a third image calculated from parameters, but the ideal (i.e., displacement-free) graphic code which has been converted from the coded content. Then there is the possibility to quantify the shift and to read from it the covert coded additional information. This results in addition to the hidden authenticity information even more advanced uses.

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• US 5067162 [0004]
• WO 2009/097974 A1 [0005]
• DE 19614896 B4 [0008, 0026]

Claims (21)

Method for authenticating and identifying an object, comprising the following steps: Acquiring a first detection image of a selected inspection area of an object surface of the object to generate first detection image data representing characteristic properties of the object surface in the inspection area; Generating third image data representing a third image different from the first detection image; Determining first comparison data by a first comparison operation in which the first capture image data is compared with the third image data using a predetermined comparison process; Acquiring a test object image of a test area of an object surface of a test object to be inspected for generating test image data representing characteristic properties of the object surface of the test object to be inspected in the test area; Determining second comparison data by a second comparison operation in which the test image data is compared with the third image data using the predetermined comparison method; Authenticating the test object by comparing the first comparison data with the second comparison data to determine authenticity properties, as well as by evaluating the authenticity properties. The method of claim 1, wherein the predetermined comparison method in the first and second comparison operations comprises calculating a vector field of similarity vectors for a predetermined number of measurement points by a cross-correlation analysis calculation, the vectors being based on singular points of a local correlation coefficient field calculated in the measurement point environments, preferably on the latter Maximum, show. The method of claim 1 or 2, wherein in the generation of the third image, the third image data are calculated according to a predetermined image generation algorithm based on generating parameters. Method according to one of the preceding claims, wherein a third image is constructed by tiling from similar sub-matrices of the same size. Method according to one of the preceding claims, wherein the first comparison data are stored as a similarity signature in a mark associated with the object, wherein the first comparison data are preferably used in the determination of authenticity data. Method according to one of the preceding claims, wherein the first comparison data or their check code are stored in a separate database from the object, preferably in the determination of the authenticity data first comparison data or its check code are retrieved from the database. Method according to one of the preceding claims, characterized by the production of a marking on the object in the test area, wherein the marking is preferably a non-reproducible markings which can be produced only once in a given form, wherein the marking is preferably in the form of a surface structuring is produced. Process according to claim 7, wherein the marking is produced in the form of a surface structuring with a stochastic character of the surface structure substantially without periodic structural parts. A method according to claim 7 or 8, wherein a laser processing device is used to produce the mark. Method according to one of the preceding claims, wherein the test region of the object for generating local, near-surface material changes is irradiated with laser radiation such that a marking in the form of a surface structuring with stochastic character of the surface structure is generated substantially without periodic structural components. Method according to one of the preceding claims, wherein comparison data and / or generating data for the third image are stored in a code on or in the object, wherein the code is preferably a data matrix code. The method of claim 11, wherein the code contains hidden additional information by moving and / or rotating pixels or groups of pixels and / or wherein the code contains encrypted and / or digitally signed digital data and / or wherein the code additionally includes a digital and / or contains electronic signature for additional authentication. A system for authenticating and identifying an object comprising: means for acquiring a first detection image of a selected inspection area (PB) of an object surface of an object (OBJ) for generating first detection image data (EBD) representing characteristics of the object surface in the inspection area; third image data generating means (DBD) representing a third image different from the first detection image; means for obtaining first comparison data (VD1) by a first comparison operation (V1), in which the first detection image data (EBD) is compared with the third image data (DBD) using a predetermined comparison method; means for acquiring a device under test image of a test area (PB) of an object surface of a test object (PRO) to be inspected to generate test image data (PBD) representing characteristic properties of the object surface of the test object under test in the inspection area; means for obtaining second comparison data (VD2) by a second comparison operation (V2) in which the test image data is compared with the third image data using the predetermined comparison method; and a device for authenticating the test object by comparing the first comparison data (VD1) with the second comparison data (VD2) to determine authenticity properties, as well as by evaluating the authenticity properties. A system according to claim 13, characterized by means for producing a mark on the object in the test area, the mark preferably being a non-reproducible mark which can only be generated once in a given shape, the mark preferably being in the form of a surface structuring is produced. A system according to claim 14, characterized in that the means for generating a mark is a laser processing apparatus, wherein the laser processing apparatus is preferably configured to produce markers having a stochastic character of the surface structure substantially without periodic structure portions. A system according to claim 15, wherein the laser processing apparatus comprises a control unit (CON) and a scanner unit (SCN) connected to the control unit for controlled deflection of a laser beam (LS), the control unit comprising a random generator for generating random signals and the scanner unit based on the random signals is controllable. A system according to claim 15 or 16, wherein the laser processing device comprises a random mask device (ZM) with a plurality of freely movable mask elements (GK, MAT, MMAT) and a device for exciting movements of the mask elements, wherein the mask elements in the beam path of the laser beam ( LS) are arranged. Laser processing device for producing a permanent marking on an object, comprising: an object holder (OH) for receiving the object (OBJ); and a laser system having a laser radiation source (LS) and a beam guidance system (SF) for guiding a laser beam onto a surface (OO) of the object; characterized in that the laser processing apparatus is configured to generate non-reproducible marks which are only producible once by virtue of the configuration of the laser processing apparatus in a given form. The laser processing apparatus according to claim 18, wherein the laser processing apparatus is configured to generate markers having a stochastic character of the surface structure substantially without periodic structure portions. A laser processing apparatus according to claim 18 or 19, wherein the laser processing apparatus comprises a control unit (CON) and a scanner unit (SCN) connected to the control unit for controlled deflection of a laser beam (LS), the control unit comprising a random generator for generating random signals and the scanner unit based on the random signals is controllable. A laser processing apparatus according to claim 18, 19 or 20, wherein said laser processing apparatus comprises a random mask means (ZM) having a plurality of relatively freely movable mask members (GK, MAT, MMAT) and means for exciting movements of said mask members, said mask members being in the beam path of said mask elements Laser beam (LS) are arranged.

Images