Patent Application
20230398805
The present invention relates to a printing method and authentication method for a print of a digital image to be produced, comprising a method of printing authentication marks by applying an at least amplitude modulated halftone print in a detection zone to an object, the printed area of the detection zone comprising contiguous halftone cells, in each of which halftone cells a halftone dot is printed from a matrix of printable halftone elements. It also relates to verification of an original print produced by means of such a halftone printing method.
A significant part of global counterfeiting crime is the copying and impersonation of printed documents and packaging. This affects not only government ID documents such as passports, identity cards, etc., but also documents related to the proof of origin of commercial products. These include certificates, accompanying documents, proof of origin and, to a large extent, packaging of branded products. The widespread distribution of the products, i.e. the size of their markets and the expected profits for a counterfeiter are motivating factors. Accordingly, especially well-known brands with a high-quality promise and thus a high retail price or street price suffer as targets of counterfeiting crime. Practically all branches of industry in the field of consumer and industrial goods are affected; well-known examples are spare parts for passenger cars, watches and medicines. Basically, all types of packaging are affected, such as blister packs, cardboard packaging, hard packaging (cans, etc.), especially those whose designs can be reproduced by a printing process such as offset, flexo or digital printing. The quality of the counterfeited packaging is good to very good in some cases, whereby a good counterfeit is one that does not catch the eye of a consumer or service employee at first glance, but only in direct comparison with the original. A very good counterfeit can only be detected by the eye of a trained expert or even, in the case of specific investigations, only by a forensic examination. The faithful reproduction of packaging designs and other documents belonging to the original product is made easier with the easy availability of high performance scanners and the usually easily recognisable or communicated visual elements on e.g. packaging that prove the originality of the product. The exact execution of a logo in terms of its colours and its geometric dimensions, the function of a barcode or an imitated serial number are no obstacle for a counterfeiter. Considering the fact that in the end consumer markets counterfeit products are often put into circulation only a few days after the new release of a product, shows on the one hand the efficiency of organised counterfeiting crime and on the other hand the still very inadequate measures to protect branded products. There is thus a great need for copy protection features of printed original packaging and documents, the verification of which is robust and reliable as well as associated with a justifiable effort. In terms of the effort involved in verification, a forensic laboratory examination would not be justifiable as an example. Original manufacturers, industrial customers and consumers rather demand quick examinations with ubiquitous means, which usually amounts to verification with a smartphone and suitable application programs (app).
Digital watermarks can serve as copy protection to a certain extent, although they are primarily designed to protect information embedded in an image. If the message (a “content”) embedded in the object, e.g. in an image, is to be extracted, a password or similar is required. A secure and at the same time reliable extraction of the message requires measures that are partly contradictory in terms of their effectiveness. For example, correction coding for redundant extraction of embedded information is a gateway for hacking attempts. The function of pure copy protection in the sense of original recognition (copy detection) is not necessarily achievable with digital watermarks; especially not if the original print consists of a photographic image whose quality must not be reduced by integrated protection measures. In contrast, an additional embedded message is subordinate, although advantageous in some use cases. Some examples of digital watermarks consist, for example, in the exploitation of a lenticular structure on the data carrier (U.S. Pat. No. 10,065,441 B1), the alteration of the colour tone by incrementally changing the amount of colour (U.S. Pat. No. 10,127,623 B1), the replacement of a special colour (such as Pantone) by the basic colours of a colour system (such as CMYK) (U.S. Pat. No. 10,270,936 B1) or other types of modulation of the printed image which, on closer inspection, constitute a visible intervention in the design.
In principle, an original can be recognised by digital fingerprinting, as the copy of an original print always differs from the original to a small extent, unless it is a so-called total forgery produced by the manufacturer or a packaging service provider or printer certified by the manufacturer (so-called 3rd shift or night shift forgeries). The causes lie in the flow of the printing ink, the ink absorption by the paper used, etc. Ordinary “content fingerprinting” of object features is not very robust and has high error rates. In addition, original recognition via digital fingerprinting requires large IT resources and thus leads to relatively slow verification procedures. Total counterfeits can be largely ruled out with additional functions, e.g. with an imprinted time stamp in conjunction with a serial number affixed to the packaging. Such an additional function is more suitable for an investigative verification of originality in a second step.
EP 3 686 027 A1 describes a method for printing authentication marks by applying an at least amplitude modulated halftone print in a detection zone to an object. This method uses adjacent halftone cells, in each of which a halftone dot is printed from a matrix of printable halftone elements, whereby individual tone values of the halftone print each correspond to a halftone level of a halftone mountain for a halftone dot. In this process, the assigned halftone level of the halftone mountain is modified in the detection zone in a predetermined manner for a plurality of tone values of halftone dots to be printed, so that a predetermined matrix image of the halftone elements to be printed is assigned to it while the tone value of the print remains constant.
DE 10 2018 115146 A1 concerns a method for producing security elements in an image that are invisible to the human eye and cannot be copied, in particular for checking the authenticity of images, wherein the image is imaged by means of a printing grid, wherein the printing grid consists of individual pixels. At least one field is defined in the print raster, wherein by means of manipulation of pixels in the field and/or by means of manipulation of the entire field, non-copyable encrypted information is deposited for comparison with at least one database. The image thus has at least one non-copyable security element, wherein the image has evaluable information within its print raster in such a way that the image has at least one field which has a manipulation of the pixels which is not visible to the human eye and/or a manipulated field which is not visible. The alteration of the halftone/raster is achieved by, for example, exchanging the halftone angle between two or more colours, changing the halftone angle of at least one colour, changing the running width or halftone frequency of the cross-line halftone of at least one colour, changing the frequency or amplitude in the case of frequency-modulated halftones of at least one colour, changing the amplitude or frequency in the case of amplitude-modulated halftones of at least one colour.
Based on this state of the art, there is a need for a relatively simple printing and downstream copy detection method that
This task is solved with a halftone printing method described herein.
Known methods for reading information require, like the QR code, in addition to at least one detection zone in which the information to be read is contained, at least one finder zone with which the existence, position, orientation of the detection zone can be determined. This can be done, as in the case of an EAN scanner or even a QR code, partly by user guidance, in that the user holds the recording device with which the information is photographed from a support in such a way that the entire code area is recorded. Then, in the QR code, the orientation of the area printed with information is determined by predetermined markings. It is now a task of the present invention, in addition to hiding the information proving an original, to also provide the finder zone or finder zones necessary for finding this information in such a way that they are also not noticeable to the unarmed eye, while being recognisable for automatic machine detection in exactly the opposite way.
It is also essential here that these finder zones are not necessarily located at the edge of an image. Yes, it is precisely part of the invention that the edges of an image, even or especially if they only represent a transition to a white, here unprinted edge zone, e.g. of packaging, are not included in the final determination of a finder zone, since a white surface by definition has no detectable halftone dots.
The invention presented here solves the problem by representing picture elements with selected halftone dot shapes. The solution follows the fact that the printout of a digital original undergoes a change through the printing process itself, in which deviations are recognisable on a microscopic level. For example, the printing ink is not distributed exactly over the space on the image medium specified by the recorder elements (smallest printing elements, or rel for short). The size of the individually controllable exposure element is the exposure pixel. Its size results from the imagesetter resolution, it corresponds to the diameter of the laser dot; the higher the imagesetter resolution, the smaller the Rels.
The structure of the medium (paper, cardboard, coated board) and the flow behaviour of the ink favour this process, which leads to a widening and deformation of the halftone dots. A scan and further printing based on the scan adds further blurring to the print image of the copy, which, given a suitable digital template, is recognisably different from the original print in that an image capture device such as a smartphone camera with suitable software can distinguish precisely this copy from the original print. It is of particular interest that suitable microscopic elements are not added to the image as an independent graphic, but are part of the image composition. At this point, it is a good idea to replace standard round, square-round or ellipsoidal halftone dots with halftone dots of more significant shapes. For example, a round halftone dot will not change its shape significantly during printing, whereas a, for example, U-shaped halftone dot 1, as shown in
The objects composed of the halftone dots, on the other hand, are transferred in the copy in apparently similar quality to the original, as is shown by the transformation of the digital original of a character 1 or 4 to its appearance in the original print 2 or 5 or in the copy of the original print 3 or 6. It is part of the present invention to present a method which makes it possible, on the one hand, to detect the characteristic microscopic alteration of the digital original during the first printing and, on the other hand, to evaluate the alterations which the copy undergoes in relation to the original as a criterion for excluding the proof of originality. It is also part of the invention that cameras on common smartphones with dedicated software are sufficient to detect the necessary microscopic subtleties on the printed image. The method of the present invention can also be applied to colour prints. The method presented is particularly aimed at protecting original products against counterfeiting.
The method of printing and authenticating a print to be made of a digital image according to the invention comprises printing authentication marks by applying an amplitude modulated halftone print in a detection zone to an object, the printed area of the detection zone consisting of asymmetric halftone dots, at least two mutually non-parallel finder edges of at least one finder zone being printed to determine the position, boundary and orientation of the detection zone, and a method of authenticating such a print comprising providing an image capture device having a microprocessor for executing an authentication program, providing the resulting print images predetermined from the print data for a predetermined number of halftone dots of the printed object from a detection zone, and providing a computer program for comparing the print images predetermined from the halftone dot data; the method comprising capturing an image of the printed article; detecting the at least two finder edges to define the detection zone from the image with halftone dot accuracy, comparing the captured print image of the detection zone with the resulting print images, and deciding on the basis of the comparison whether or not an original print is present on the printed article.
Advantageously, each finder edge along a predetermined path of the printed image consists of adjacent rows of halftone dots, the difference between the halftone dots of the adjacent rows being selected from the group comprising symmetrical halftone dots versus asymmetrical halftone dots, predetermined different halftone angles of the halftone dots, AM modulation versus FM modulation of the halftone dots, said difference from the group being predeterminable to be independently different or the same for each finder edge. In other words, one finder zone can be determined by the difference between symmetrical halftone dots versus asymmetrical halftone dots (as shown in
The difference between the halftone dots of the adjacent rows of a finder edge can also include different AM modulation of the halftone dots on both sides of the finder edge. The difference of the AM modulation can be realised in particular in the amplitude or the frequency of the two AM modulations, possibly of at least one colour.
Thus, a finder zone defined by a finder edge may have asymmetric halftone dot shapes, with the halftone dots existing beyond said finder zone on the other side of said finder edge each forming a zone of symmetric halftone dot shapes from the remaining printed image.
Alternatively, a finder zone defined by a finder edge may have symmetrical halftone dot shapes, with the halftone dots existing beyond said finder zone on the other side of said finder edge each forming a zone with asymmetrical halftone dot shapes from the remaining printed image or from the detection zone.
Further alternatively, a finder zone defined by a finder edge may have symmetrical halftone dot shapes with a first halftone angle, each adjoining a zone with a second halftone angle from the remaining printed image or from the detection zone beyond said finder zone on the other side of said finder edge (the first and second halftone angles being different from each other).
The predetermined number of asymmetric halftone dots in the detection zone may be arranged in a matrix of at least two rows and two columns; the examples shown assume at least three rows and a length of 10 or more halftone dots, but in principle a smaller number is possible. In other words, the predetermined number of halftone dots in the detection zone may be divided into zones with asymmetric and symmetric halftone dot structure, these zones being arranged in a matrix of at least two rows and two columns.
The asymmetrical halftone dots can be provided in one of the last two colour jobs to be printed, which are most visible and evaluable, in a multi-colour print.
Also, in the case of colour printing, the finder edges can then be provided by defining halftone dot shapes and/or halftone angles of an identical or a different colour layer. Advantageously, the asymmetric halftone dots to be evaluated have a grey tone value between and 75%. The same applies to the symmetrical halftone dots, although smaller and higher values up to 100% are also possible there.
At least two finder edges may meet in a corner point of a finder zone so that a finder zone is directly identified, or finder edges of one or more finder zone(s) are provided at the edge of the printed image or in at least one pair of intersecting finder zone strips.
In the method of printing authentication indicia by applying an amplitude modulated halftone print in a detection zone, a matching template can be generated based on print data selected from the group consisting of print substrate data, print ink data and print guide data.
Advantageously, this basis for comparison is then trained by original prints and print proofs, whereby optionally the captured image of the printed object undergoes a conversion of the image by a graph algorithm into the format of the basis for comparison for a direct comparison in the procedure for authentication.
Then, preferably, capturing the image of the printed object in the authentication process may comprise capturing a plurality of images with different camera parameters selected from the group consisting of varying the focus and varying the exposure time to produce an image stack, the data of which is converted into an aligned image stack; to be subsequently converted into the format of the basis of comparison. This allows the resolution to be increased so that simpler cameras of mobile communication devices can be used more easily.
The distribution of the finder zones and the detection zone(s) is provided in a predetermined matrix containing digital information.
The detection zone may be checked against the basis of comparison based on the recorder elements composing the grid points therein, and the comparison may include a threshold of corresponding matches of detected recorder elements with the recorder elements of the basis of comparison.
Advantageously, a plurality of separate detection zones is then provided, and either a total threshold value determined over all detection zones or individual threshold values of the individual detection zones are then used as a basis for decision-making.
Starting from a digital template of the halftone dots from the pre-print stage, a blurring step is switched on in which a blurred model is generated from the digital template based on the data from the group comprising the printing substrate, the printing ink and the printing guide. This can optionally be trained with a subsequent training step using original prints or proofs of the printed model for a trained model to create a matching template for an image analysis of a selected section of the printed image to be checked, whereby a matching of matching template and the data set of the image to be authenticated provides the statement “original” or “copy” after applying a quality matrix.
A print to be checked can be translated into a dataset with the same architecture as the matching template using a graph algorithm, where optionally the mathematically formalised correspondence of the halftone pattern corresponds to a dense network of nodes and edges aligned with the halftone points of the print image.
Prior to the application of said graph algorithm, the detection of the print to be tested can be performed by generating a sequence of images with different camera parameters from the group comprising variation of the focus, in particular in non-equidistant steps, variation of the exposure time and variation of the camera position, wherein the obtained image stack is aligned in an alignment step to obtain an alignment vector field, wherein subsequently the further parameters varying between the images from the above mentioned group are determined to obtain a result which is processed with said graph algorithm.
Preferred embodiments of the invention are described below with reference to the drawings, which are for explanatory purposes only and are not to be construed restrictively. The drawings show:
Selection criteria of halftone dot shapes can be of any nature, e.g. based on special or unusual halftone dot definitions in the Raster Image Process (RIP) machine used in the course of screening in pre-print. In this case, the shape of the halftone dots is necessarily linked to a certain tonal value. However, the halftone dot shapes can also be freely definable and only follow the rule that a halftone dot with a tone value of a certain number of printing recorder elements (also halftone elements=smallest printing parts of halftone dots), the shape of the halftone dot can otherwise be arbitrary. Arbitrarily shaped halftone dots can be created during layout, whereby the RIP is set up in such a way that the predefined halftone dots are adopted unchanged for the print template. The object of the present invention is not the creation of the halftone dots per se, but suggestions as to how, with the aid of their unusual geometric shapes, an original print can be distinguished in a simple manner from a copy of that print. Crucial to the execution of the invention is the way in which the specially designed halftone dots 8 can contribute to the tamper-evidence of an image. Proposals for the design of halftone dots per se are known, e.g. U.S. Pat. No. 8,456,699 B2 (growth of halftone dots (print dots or clustered dots) based on selected halftone elements (pixels)).
In the present invention, verification of the original is to be feasible with simple means, preferably a smartphone. The invention thus describes, as it were, an originality indicator integrated in the image, which is recognised by a smartphone with a corresponding application program. Another option is to combine the tamper-evident indicator with an embedded message, which is another advantage of the invention.
The tamper-evident indicator consists essentially of an accumulation 7 of preferably apparently randomly shaped halftone dots 8. This accumulation 7 or matrix is placed on a zone of predetermined size at a predetermined location in the base image. The base image is, like later image 26 of
The indication of the originality of the print can be carried out, as explained above, in particular only with picture elements (i.e. such accumulations 7 in one or more zones) which are subject to AM raster process or screening according to the invention. Halftone dots in the base image outside said zones may have a shape customary for amplitude modulated halftone printing (AM halftone printing), for example round or elliptical, but do not have to. This describes a process in which one (or each) finder edge has two zones of picture elements with different amplitude modulated (AM) screenings/halftone printing.
Advantageously, the special sub-areas 10 differ in their halftone structure from that of the surrounding base image 210. However, it may also be envisaged that the special sub-area detection zone 10 constitutes only part of the area printed with asymmetrically designed halftone dots and that only the one or more finder zones 19 or 20 are provided with, for example, symmetrical halftone dots. With brief reference to
The condition of non-parallelism of the finder edges 211, 212 to each other can also be called intersecting finder edges. This intersecting point can exist, for example, as a corner point of a finder zone 19 in the evaluation, although the image evaluation does not need to use this intersecting point as a corner point of a finder zone. This intersection point in the case of straight lines that are not parallel to each other can lie outside the image/print, since what is important here is a distance, in particular length next to the alignment, of this finder edge and not the recording of the intersection point itself. Nevertheless, an orthogonality of the finder edges 211, 212 to each other is preferred. Since this simplifies the determination of the position, boundary and orientation of the detection zone 21. In addition to the direct pixel-precise determination of the detection zone 21 from the finder edges 211, 212, one or more finder zones can also be determined first in order to then determine the detection zone 21 based on these. In the extreme case, there is only detection zone 21, two of whose edges are used as finder edges.
Finally,
The change of the digital artwork via the original print to the print of a scan of the original print, i.e. the copy, is demonstrated in
The identification zone 21 or detection zone, which is used to check the originality of the print, which comprises the actual originality indicator, is located at a selected point in the image and is analysed with pinpoint accuracy. The position of the identification zone 21 or of the tamper-evident indicator can be fixed or can also be coded in the finder marks. The zones 19, 20 and 21 may also be adjacent to each other. The component referred to as the surrounding image 210 may or may not have the same halftone print as the detection zone 21. Essentially, there are at least two finder edges 211 and 212 which are not parallel to each other and which represent the edge of one or more finder zones.
Here, finder edge 211, 212 means not only the line drawn here as an auxiliary line but the existence of adjacent rows of different halftone dots existing along a path, the difference between the halftone dots of the adjacent rows being selected from the group comprising symmetrical halftone dots versus asymmetrical halftone dots, predetermined different halftone angles, AM modulation versus FM modulation.
If necessary, the tasks of finder, alignment marker and originality indicator can be combined. One such possibility is for example
In the embodiment according to
Essential for this is a grey value in the range of 20% to 80%, or in the case of colour printing a corresponding half-tone value of the printing colour, in particular 25% to 75%, so that the difference in the halftone dot between symmetrical and asymmetrical dots at recorder element level is recognisable for image evaluation. At a higher or lower value, such a finder edge 211, 212 becomes more and more an ordinary edge, which is also recognisable as such by the unaided eye, since then the transition from asymmetrical to symmetrical halftone dot elements is no longer recognisable, but an image component comprises an edge. However, a finder edge is also present if, for example, asymmetrical halftone dots and/or a certain halftone angle are provided on one side of the finder edge, usually in several rows next to each other, and an 80% to 100% grey value print is provided on the other side, possibly in one colour, also in several rows. This is because a symmetrical halftone dot distribution with a grey value of 100% corresponds to a printed edge.
Further embodiments for a combination of zones with halftone dots of different shapes are shown in
In other words, the section 32 of
The halftone angles of all images
The advantage of an approach to changing colour spaces is easier recognition by an image-recording system, especially at low resolutions. The principle presented above of distinguishing halftone dot shapes of a basic image from halftone dot shapes of certain other parts of the image consisting of halftone dots of other geometry is illustrated in connection with
The printout of a scan from this print will have a further deformation of the halftone dots on top (=the last printed colour layer) and thus be recognised as a copy with the help of digital image capture devices in conjunction with dedicated software. The original print itself is produced from a digital image template and develops in the course of the printing process due to the influences of the printing process, the colour and media properties in a predictable or predeterminable way to a printout that is like a fingerprint of the original.
The printing steps which lead to the results “original” and “copy” in the invention can in principle be described as a process in which, shown by way of example in
The digital master refers to the raster data for printing form production, e.g. the files for the laser imagesetter in offset printing. The corresponding files contain all data about the structure of all halftone dots of a colour separation of the image to be printed. Ideally, each halftone dot consists of groups of square pixels, which in their entirety form a halftone dot. The transfer of the ink to the printing medium, for example a coated cardboard, is a physical process in which various influencing factors based on the rheological properties of the ink used and properties of the printing medium as well as the process control, e.g. the amount of ink applied, lead among other things to a deformation of the halftone dot.
The deformation of a halftone dot under given printing conditions can be described with a point spread function (PSF, also called blur kernel). Known point spread functions are based, for example, on a two-dimensional Gaussian smoothing or mean filtering of neighbouring pixels. A dot spread function describes the printed image as a function of all the essential printing parameters, in particular the flow and drying behaviour of the ink, the ink absorption of the medium and the process control. It is advantageous to train the mathematical model 48 for the soft drawing of the halftone dots for predefined printing conditions 49. The predefined conditions are, for example, the type of cardboard used, the ink and specifications for the guidance of the printing press, for example the ink application. It is particularly advantageous if the mathematical model is trained for each subject, for example image motif on an original packaging for a specific brand product. Such a trained model 50 for the dot gain on an original packaging produced with a printing process certified for the model advantageously serves as a standard for verifying the originality of a packaging, which can be performed with a suitable image capture device (smartphone) and dedicated software at any time and any place.
For authentication, requirements are placed on the image-recording system in terms of hardware and recording methods that allow resolution down to the size of a halftone element, i.e. the smallest printing part of a halftone dot. An image printed by the offset process is considered a high-quality print if the halftone has a frequency of 80 lines per centimetre or finer. 80 lines/cm corresponds to a size of 15.6 μm for a halftone element. It can be shown that capturing a halftone element of this size with a conventional smartphone camera in one shot is not feasible.
As image capturing devices, these smartphones are used for a preferred image analysis of a rasterised image according to the invention with such typically 12MPixel smartphone cameras, on the other hand, with the support of super resolution (Super Resolution) and/or mathematical deconvolution methods (Deconvolution), which are also used for applications in astronomy and microscope imaging, among others. Super resolution has long been state of the art (see e.g. Borman et al, Super-Resolution from Image Sequences, Department of Electrical Engineering, University of Notre Dame, 1998). For image enhancement based on Super Resolution, software is available for consumer and less professional applications, such as Chasy Draw IES or Topaz Gigapixel AI.
Super-resolution and deconvolution methods, see e.g. “Pragmatic Introduction to Signalprocessing”, Tom O'Haver, Department of Chemistry and Biochemistry, The University of Maryland at College Park; available at https://terpconnect.umd.eduhtoh/spectrum/TOC.html, essentially use multiple images taken under conditions that are roughly similar but differ only slightly or moderately in one or more of these conditions. From these differences, information about the fine resolution is derived. The goal of such a procedure could be either a high-resolution image or the direct measurement of high-precision features on a low-resolution image. Scene content, focus, exposure, position and movement of the smartphone affect the outcome of these methods.
As shown in
A super-resolution method generally achieves a 2-4-fold increase in resolution, which in the case of a 12 megapixel image amounts to about 9 microns for sampling a raster element 69. Deconvolution methods follow a similar approach but assume very blurred images taken from closer range. A combined use of super-resolution and deconvolution can result in an 8-fold increase in sampling frequency compared to a normal image taken from the usual minimum near limit, achieving about 4 microns of resolution 70 to measure point features. Thus, depending on the camera model used or to be used, the comparison can be made directly, after applying a super-resolution method and/or after applying a deconvolution method.
Based on this approach, that it is necessary for a number of cameras, in particular smartphones, to use a quality improvement that results in a higher resolution, the authentication of an image is made possible here with the help of a short video sequence or a series of individual shots of this image, for example executed by a smartphone with a 12 megapixel camera, using a suitable Super Resolution Algorithm 56. A sensor of a common smartphone in combination with a Super Resolution Algorithm is sufficient for this purpose. Alternatively or additionally, one can also use a deconvolution method, which is integrated in Matlab and Octave, for example.
The starting point of each of these methods is the acquisition of several images with some fixed parameters such as the resolution and the light output, where some parameters cannot be influenced or are unknown. First, the position of the smartphone is given by a guidance with the hand, which leads to a movement in X, Y or Z direction with a speed of some mm/s, resulting in an offset of 60 μm for a movement of 1-2 mm/s or a movement of 1-3 pixels/s in the image plane. Ambient light also has an influence, especially some types of neon light. The resulting images will therefore differ slightly due to a small shift and the lighting conditions. Shutter speed can also cause camera shake and thus blur.
The matching template 52, which is a quasi normalised version of the original print, can for example be formally described by nodes and edges according to geometric graph theory, which is described by the reference signs 51, 52. However, other approaches are also possible to characterise the template. For example, a content fingerprinting method according to EP 2 717 510 B1 is also suitable.
In the case of the graph-theoretic approach, a print 55 to be inspected, which may be a copy, is translated into a data set 59 with the same architecture as the template 52, like the digital template or original, using a graph algorithm. In the extreme case, the mathematically formalised equivalent of the halftone pattern corresponds to a dense network of nodes aligned with the halftone dots of the print image.
Taking into account the auxiliary procedures to increase the camera resolution, a sequence of single shots or a video stream according to
The print 55 under test is captured with different camera parameters 60. By varying the focus in non-equidistant steps, the analysis reveals the critical differences of the halftone dots by deconvolution of the blur of the video stream (which is analysed as single frames). Likewise, the variation of the exposure time serves to reveal microscopic pressure peculiarities, compensating for the light differences coming from the 50 Hz light source. The result is an image stack 61. The method calculates the alignment from several individual images 62 to obtain an alignment vector field 63, which forms the basis for image synthesis with high resolution. Estimates are also obtained in a similar manner for parameters that vary between images, such as lighting conditions. Then processing 64 of the aligned images is performed to obtain results 65. A mathematical representation is generated for a high-resolution image that can be compared to the matching template 52.
The process of aligning the individual images begins with a reasonably register-accurate superimposition of the individual images, which is a simple step even with blurred images. In the next step, information about the exact position of the process-oriented halftone dots flows into the process. Here, the position of the process-oriented halftone dots must be known at different points in time. In the case of alternating halftone dots of regular (process-oriented) and irregular shape, an attempt can be made to align a smaller part of the image with a shift of one pixel once in x and once in y until an alignment with correct process-oriented halftone dots is found. An alternating pattern defines how many processes need to be executed. Therefore, regularly shaped halftone points favour the unfolding process.
The process of image recognition is shown in
A regularly shaped border of halftone points favours the estimation of the position in the blurred image, as only one edge from left to right (from background to foreground) is taken into account, which is easier to implement at comparison level.
Another embodiment is that edges 80 of the halftone dots in the direction of a rasterization line tend to form a channel that is as straight as possible. This effect leads to an increased geometric stability of the raster image in a preferred direction, which can be used for alignment of the raster image.
Halftone points can thus advantageously be modelled in such a way that they provide information for the orientation of the halftone image and the coding of the originality, respectively.
In principle, it is possible for the deconvolution process used in the present invention to restore the shape of the halftone dots defined in the pre-print artwork, i.e. to reverse the softness caused by printing. This is a reverse operation to the convolution of the image information, which manifests itself as the softness of the halftone dots. A comparison of halftone images with halftone dot shapes resulting from deconvolution can be made with various mathematical descriptors, e.g. on the basis of centroid distance functions, area functions, chord length function, the use of quadratic shape matrices or curvature-based scale spaces, etc.
The finder zone 190 has a finder edge row number 222 of eight and a finder edge length 223 of twelve grid points, all of which are asymmetrical and thus form the finder zone 190. In other words, the actual finder edge 212 has, on the finder zone side, a finder edge row number 222 of one to eight with a length predetermined with the finder edge length 223. It has the same finder edge length 223 on the outer image side, since this is predetermined by the limited area, while the finder edge row number 224 is shown here to be selectable between one and three. This results, for example, in a finder edge zone 225 to be evaluated by the authentication method of 12 by 3 grid points on both sides of the finder edge centre line 212. The evaluation does not have to be symmetrical, the row number 224 and 222 can be selected differently.
The identification zone or detection zone 21 has a finder edge row number 222 of eight and a finder edge length 223 of twelve halftone points, all of which are asymmetrical and thus form the detection zone 21. The numbers here are the same as finder zone 190, which they do not have to be. In other words, the actual finder edge 211 has on the detection zone side a number of one to eight finder edge rows 222 with a length predetermined with the finder edge length 223. It has the same finder edge length 223 on the outer image side, since this is predetermined by the limited area, while the finder edge rows 224 are shown here to be selectable between one and three. This results, for example, in a finder edge zone 226 to be evaluated by the authentication method of 12 by 3 halftone dots on both sides of the finder edge centre line 212. The finder edge zone 226 may also end at the edge 213 and this edge may represent a further horizontal finder edge 212 (not shown in FIG.), since the outer image area 210 in the vicinity of the detection zone 21 is symmetrical and the edge is also recognised as symmetrical as a full black edge with a grey tone of 100%. But the detection zone 21 can also be in the inner area of the image. The length or distance of the twelve asymmetrical halftone dots can be detected by the authentication procedure and can be used for the orientation and scaling of the overall image. The more finder edges 211, 212 are used, the easier, faster and more accurate the pixel-precise detection of the detection zone 21 can be achieved.
The finder edge zones 225 and 226, i.e. matrices (arrays) of a width of halftone dots predetermined by the zone and the evaluation method, are also drawn by way of example in
In summary, the invention has a large number of individual features, some of which also constitute independent technical gauges:
A method of detecting copies of black and white and colour images, wherein
| Number | Date | Country | Kind |
|---|---|---|---|
| 20207154.4 | Nov 2020 | EP | regional |
This application is the United States national phase of International Patent Application No. PCT/EP2021/081408 filed Nov. 11, 2021, and claims priority to European Patent Application No. 20207154.4 filed Nov. 12, 2020, the disclosures of which are hereby incorporated by reference in their entireties.
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2021/081408 | 11/11/2021 | WO |
