The present disclosure relates generally to individual glyphs, and more particularly to an individual glyph generation method, and to an individual glyph inspection method and system.
Security printing, publishing, and imaging are important components of product differentiation, tracking and tracing, inspection, authenticating, forensics, as well as other anti-counterfeiting initiatives. Security printing involves providing each package with a unique ID, in the form of, for example, a smart label, deterrent or mark. Such unique identifiers may be overt and/or covert, and may contain authenticable data. Thus, such marks are particularly suitable for product track and trace, inspection, and authentication. Image based forensic services have been used to detect and aggregate counterfeits in a supply chain. Such services are particularly useful when products do not contain specific security deterrents. In particular, such services analyze printing that has occurred on the product in order to investigate the authenticity.
Features and advantages of embodiments of the present disclosure will become apparent by reference to the following detailed description and drawings, in which like reference numerals correspond to similar, though perhaps not identical, components. For the sake of brevity, reference numerals or features having a previously described function may or may not be described in connection with other drawings in which they appear.
Embodiments of the system and methods disclosed herein advantageously enable simultaneous object authentication and forensic-level identification. The system and methods utilize a true resolution imaging device, which, when coupled with a custom image analysis system, enables a single printed character to simultaneously provide fiducial marking, inspection information, authentication and forensics. Such characters may be reliably read in the same manner at different times and with different devices.
Referring now to
Once the glyph is generated, it is printed using a desirable printer, as shown at reference numeral 102. Generally, the printer selected will be one that will be used for deployment of the glyph on an object. Non-limiting examples of suitable printing techniques include inkjet printing (e.g., thermal, piezoelectric, continuous, etc.), laserjet printing (e.g., thermal laserjet), electrophotographic printing, gravure printing, flexographic printing, offset printing, screen printing, dot matrix printing, or any other suitable printing technique that can print the characters/graphics selected or generated for the glyph(s).
One or more of the characters and/or graphics of the printed glyph is/are then captured with a true resolution imaging device, as shown at reference numeral 104. In one embodiment, the entire glyph is captured, and in another embodiment, a character and/or graphic (or portion of a graphic) of interest is captured. It is to be understood that so long as the desired character and/or graphic is in the captured image, and the character and/or graphic is large enough to conceivably vary as a function of angle (e.g., the captured image of a graphic is not a single pixel in size), a portion of the character and/or graphic may be captured and analyzed. In one non-limiting example, a single tile of a 2D barcode may be a sufficient captured image.
The true resolution imaging device described herein includes hardware that is able to capture an image that is overwhelmingly similar to the original image (e.g., the printed glyph). Some suitable example(s) of such an imaging device (and method(s) of using the same) are described in a related patent application PCT/US09/44777 filed concurrently herewith on May 21, 2009, entitled “Imaging a Print Aberration,” which application is incorporated by reference herein in its entirety. More particularly, the size of the pixels on the image sensor in the device corresponds to the size of the pixels imaged on the surface of a substrate. In some instances, the true resolution is sufficiently high to resolve substrate/ink interaction features of printed images (described further hereinbelow). The images captured via this device provide forensic evidence (associated with some probability) that is generally not achievable using other imaging devices, such as desktop scanners and mobile cameras. Non-limiting examples of the true resolution imaging device include a 1:1 magnification, 1 to 5 micron true resolution lens-based multi-mega-pixel USB CMOS imaging device (e.g., 1:1 magnification, 3.8 micron true resolution Dyson relay lens-based 3 mega-pixel USB CMOS imaging device), USB microscopes, and iDetector™ (from GSSC), with varying degrees of true resolution. In general, resolving capability is defined by the width of the smallest line that can be successfully read, or by other suitable modulation transfer function methods.
The true resolution imaging device 16 generally includes a self-contained illumination source that affords the capture of individual printed characters and/or graphics with printed parasitics (i.e., anything not intentionally printed, such as satellites (ink droplet tails), or porosity (absorbance of the ink into fibers of the substrate). Furthermore, such devices 16 generally capture a relatively small area (e.g., 5×5 mm) at high resolution to achieve a suitable image. However, it is to be understood that multiple frames or devices 16 may be used simultaneously to crease a much larger image (i.e., in pixels of height or width).
It is to be understood that the glyphs printed and the glyph images captured are not limited to monochrome output. For example, microscopic spatial aberrations (or parasitics) in color may exist in the same way as aberrations exist in a monochrome printing process. Furthermore, in a cyan magenta yellow (CMY) printing process, there may be microscopic variations in the registration or alignment of the color planes.
As shown at reference numeral 106, the captured printed character(s) and/or graphic(s) of the glyph are analyzed to determine at least one parameter associated therewith. It is to be understood that a single character and/or graphic may be analyzed, select characters and/or graphics may be analyzed, or each character and/or graphic making up the glyph may be analyzed. The system used for the analysis includes custom software that performs character/graphic boundary analysis. Boundary analysis may include analysis of the shape, boundary texture, and boundary parasitics of the individual characters and/or graphics.
The analysis may be accomplished via any image analysis technique that is able to account for pixel-to-pixel edges, boundaries, gradients, etc. at full resolution of the image. In one embodiment, shape analysis software or statistically comparable metrics are used for the analysis. Non-limiting examples of such software and metrics include a contrast-insensitive thresholding algorithm to binarize the image; a perimeter-pixel sequence generating algorithm to produce a high-resolution version of a modified Freeman shape code; a small angle-sensitive, shape coding feature set (SCFS) that is sensitive to relative changes in printed glyph radius, parasitics and parasitic complexity; two types of moving average representations of features of the small angle-sensitive shape coding feature set (MA-SCFS); multiple recursive overall shape comparison (ROSC) metrics; and optimal scaling and registration of the glyph image with another image and comparison of such other image.
One or more of the previous metrics may be run on the character(s)/graphic(s) to identify parameters associated with the particular character(s)/graphic(s). The analysis technique(s) selected for computation will depend on the type of glyph/printed mark being analyzed. Furthermore, the parameters identified as a result of the analysis may vary depending, at least in part, upon the glyph itself, the printer used, and the combination of ink and substrate used. The identified parameters may include the shape of the character/graphic, the boundary texture (which includes satellites), porosity, a modified shape descriptor, or combinations thereof. Satellites are unintentional printed marks that appear outside or around the boundary of a character or graphic (see, e.g.,
Referring briefly to
Since the printer, ink, and substrate used to print the glyphs contribute to the distinctive characteristics/parameters of the character(s) and/or graphic(s) making up the glyphs, the stored characteristics/parameters may be used for comparisons with glyphs that are deployed on various objects (described further hereinbelow). It is to be understood that the generated character(s)/graphic(s) are initially analyzed and stored in order to address various comparison workflows often associated with printed and deployed glyphs. Table 1 below illustrates such comparison workflows. More specifically, during generation and initial analysis of the glyphs, a large number of characters and/or graphics are imaged and analyzed based on the workflows in Table 1 so that the variances can be compared within the same workflow and between different workflows. This enables a multitude of probabilities to be generated for the glyph(s) during the generation stage, so that when a deployed glyph is analyzed, the probability of its authenticity may be determined by comparing it with known, previously analyzed glyphs. Furthermore, the large number of characters and/or glyphs is analyzed during generation so that group variances are known. As shown in Table 1, group variances for glyphs printed using the same print technology and substrate may be different from the group variances for glyphs printed where the print technology, substrate or other measurable parameter (such as those described in Table 1) is changed.
Using the shape analysis software programs or metrics described above, the methods disclosed herein enable such workflows to be utilized on the deployment end (see
When generating the glyph for storing associated information in a registry, the workflow used is known, and thus a suitable analysis technique may be selected. Furthermore, as discussed immediately above, when the deployed glyph is analyzed, workflows may be identified, and thus suitable analysis technique(s) may be selected. As such, some of the analysis techniques may be pre-defined by the workflow used. However, it is to be understood that other suitable analysis techniques may be selected based upon what type(s) of glyph marks (e.g., satellites, porosities, edge properties, etc.) are available.
The small angle-sensitive, shape coding feature set (SCFS) may be used to determine the parameters when the character and/or graphic is printed using different printer models or is printed twice with the same printer model. In these examples, SCFS is sufficient for distinguishing the same character printed twice, either with the same printer model or a different printer model. It is to be understood that it may be desirable to analyze a large set of the same character and/or graphic in order to determine the population statistics for satellite location and type and/or porosity location for a given printer model. A large set may include any desirable number of the character and/or graphic that results in population statistics that enable probability values to be generated for that character and/or graphic. It may also be desirable to define the variability in a set of different characters printed using the same printer and compare this variability with the variability in measuring the exact same character twice, possibly with a different true resolution imaging device.
The third workflow in Table 1 is illustrated in
The fourth workflow shown in Table 1 compares an original character to its copy. In general, σ2Copy>σ2Char, so the SCFS features allow for distinguishing between real and copied characters. A threshold for the ratio is determined based upon the SCFS analysis. When comparing deployed glyphs, the resulting ratio will be compared to the threshold stored in the registry to determine the probability of authenticity.
What is believed to be the most challenging workflow, workflow 5 in Table 1, requires the use of the ROSC metrics. The ROSC metrics allow different imaging devices 16 to be used on the same characters (thereby enabling distributed supply chain monitoring applications). For ROSC, σ2Camera<<σ2Char∥σ2Copy∥σ2PSM∥σ2PDM.
Referring back to
The embodiments of the glyphs disclosed herein advantageously provide forensic-level security without the need for additional security deterrents. However, it is to be understood that additional security deterrents or other non-forensic identifying marks may be incorporated onto the object with the glyph, and in some instances, may be linked to the glyph in the secure registry. Linking the glyph to a non-forensic identifying mark enables one to use the content encoded in or visible on the mark as a search query when searching the secure registry database for authentication purposes. In some instances, this type of query simplifies the search of the database because the content is specifically linked to particular glyph(s), as opposed to searching image parameters that may be associated with multiple glyphs. Examples of such non-forensic identifying marks include any security feature with a unique number, such as, for example, watermarks, graphical alphanumerics, scrambled indicia, bar codes, serial numbers, or other unique identifying alphanumeric and/or graphic indicia that allow one to perform an indexed image-to-image comparison using the database.
After its generation and when desirable or appropriate, the generated glyph is deployed (i.e., printed) on an object. It is to be understood that the term “object” as used herein is to be interpreted broadly and may include, but is not limited to any type of object, product, document or package. Likewise, the term “package” is to be interpreted broadly herein to include any unit for containing a product, displaying a product, or otherwise identifying a branded good. Non-limitative examples of such packages include labels, anti-tamper strips (which tear when removal is attempted, thereby damaging both visual and physical aspects of any deterrents thereon), tickets, coupons, and other single-used items, boxes, bags, containers, clamshells, bands, tape, wraps, ties, bottles, vials, dispensers, inserts, other documents, or the like, or combinations thereof.
Once an object is deployed, it may be accessed, transmitted, and/or processed through a variety of channels, including inspection channels (e.g., at distribution nodes in a supply chain network), stocking and/or point-of-sale channels (e.g., at a retailer), and/or end consumer authentication channels (e.g., at or before point-of-sale, for recall testing, etc.).
The system 20, shown in
The image is received via computer software 28 that is capable of receiving image. The software 28 is generally part of the analysis system 30, but may be located at the registry 24 (as shown in
Comparison between the parameters of the character(s)/graphic(s) of the image of the deployed glyph 22 and those of images of other glyphs previously stored in the registry 24 is performed (by an image comparison system 38 of the analysis system 30, which includes programs in operative communication with the registry database) to determine whether the deployed glyph is authentic or counterfeit. Such a comparison may be accomplished via two modes, both of which are shown in
The first of the two modes is shown at reference numerals 606, 608, and 610 of
It is to be understood that if an image ID (e.g., ID 26) is stored in the registry 24 and both the image and the corresponding parameters are also stored in the registry 24, a query using similar information from the object and glyph may be run to identify the best match. The probability that the glyph identified from the query is a match will be assessed, and from the assessment, a match or non-match reported.
Once the best match is identified, the parameters of the best match are compared with the parameters of the glyph 22 to generate the probability of authenticity of the deployed glyph 22. In a non-limiting example, the population of all images in a class is used to determine the expected value (typically in a Hamming distance) of the difference between any two images (i.e., the glyph 22 image and the image/parameters associated with an image stored in the registry 24). Then, the actual distance between any two images is compared to this expected value using Chi-square analysis.
In another embodiment, the best match may be determined by using the glyph image 22 itself to query the database of the registry 24 for other like previously stored images. This is an image-to-image comparison (as opposed to an initial parameter-to-parameter comparison) to determine the best match. When determining the best match in this manner, lower resolution refinement is used to hone in on the best matching images. As such, in this embodiment, rather than looking at all the images in the registry 24 at, for example, 3 micron true resolution, down-sampled versions of the images may be searched for faster refinement to the best possible matches. Such down-sampling may be particularly useful when other mark(s) 26 is/are not associated with the glyph 22 image, and thus are not available for use in searching the registry 24. It is to be understood, however, that when the other identifying mark(s) 26 are included on the object 18 with the glyph 22, such other marks(s) 26 may be used to narrow the query for this type of search of the registry 24. From the comparison, the image most closely resembling the glyph 22 image (i.e., the best match) is analyzed (by the system 30) to determine its parameters. The generated parameters of the stored image are then compared with the parameters of the glyph image 22 to generate the probability of authenticity of the deployed glyph 22 as described above.
The second of the two modes is shown at reference numerals 612, 614, 606 and 616 of
The boundary analysis software programs and metrics, in combination with the true resolution device 16, enable characters/graphics of images of deployed glyphs 22 to be analyzed, and such parameters are compared with stored parameters of authentic glyph images previously analyzed via similar methods. While the results of the analysis will likely vary, a statistical probability that the deployed glyph 22 is authentic is generated, an example of which is described hereinabove.
In some instances, details of the printer, cartridge, and/or substrate used to generate the original glyphs will be stored in the registry 24 with other parameter information. When the probability of authenticity is very high (which is based upon the Hamming distances described hereinabove), one may conclude that the printer used to generate the original stored glyph has also been used to generate the deployed glyph 22.
In any of the embodiments of the method disclosed in
Any information resulting from the comparison (including the conclusion as to whether the glyph 22 and the associated object 18 are authentic or counterfeit) may be transmitted (via a secure connection) from the registry 24 to the user who initially transmitted the image of the glyph 22 (or some other authorized user).
While several embodiments have been described in detail, it will be apparent to those skilled in the art that the disclosed embodiments may be modified. Therefore, the foregoing description is to be considered exemplary rather than limiting.
Filing Document | Filing Date | Country | Kind | 371c Date |
---|---|---|---|---|
PCT/US2009/044853 | 5/21/2009 | WO | 00 | 9/22/2011 |
Publishing Document | Publishing Date | Country | Kind |
---|---|---|---|
WO2010/134919 | 11/25/2010 | WO | A |
Number | Name | Date | Kind |
---|---|---|---|
5883986 | Kopec et al. | Mar 1999 | A |
5901224 | Hecht | May 1999 | A |
6341730 | Petrie | Jan 2002 | B1 |
6419162 | Petrie et al. | Jul 2002 | B1 |
6641051 | Illowsky et al. | Nov 2003 | B1 |
6671397 | Mahon et al. | Dec 2003 | B1 |
6819776 | Chang | Nov 2004 | B2 |
6871789 | Hilton et al. | Mar 2005 | B2 |
7028188 | Moore | Apr 2006 | B1 |
7046829 | Udupa et al. | May 2006 | B2 |
7106905 | Simske | Sep 2006 | B2 |
7583267 | Stamm et al. | Sep 2009 | B2 |
7630552 | Houle et al. | Dec 2009 | B2 |
7657091 | Postnikov et al. | Feb 2010 | B2 |
7676038 | Simske et al. | Mar 2010 | B2 |
7830557 | Simske et al. | Nov 2010 | B2 |
7878549 | Simske et al. | Feb 2011 | B2 |
7916863 | Simske et al. | Mar 2011 | B2 |
8184867 | Otto et al. | May 2012 | B2 |
20020018593 | Oohmura et al. | Feb 2002 | A1 |
20030052171 | Gyllenskog | Mar 2003 | A1 |
20030156733 | Zeller et al. | Aug 2003 | A1 |
20040078333 | Hilton et al. | Apr 2004 | A1 |
20050008216 | Smith et al. | Jan 2005 | A1 |
20060036566 | Simske et al. | Feb 2006 | A1 |
20060036614 | Simske et al. | Feb 2006 | A1 |
20060036649 | Simske et al. | Feb 2006 | A1 |
20060061088 | Harrington et al. | Mar 2006 | A1 |
20060244988 | Oishi | Nov 2006 | A1 |
20060293891 | Pathuel | Dec 2006 | A1 |
20080006615 | Rosario et al. | Jan 2008 | A1 |
20080304110 | Simske et al. | Dec 2008 | A1 |
20090144799 | Simske | Jun 2009 | A1 |
20110068181 | Simske et al. | Mar 2011 | A1 |
20110116681 | Simske et al. | May 2011 | A1 |
20110170145 | Govyadinov et al. | Jul 2011 | A1 |
20110310404 | Simske et al. | Dec 2011 | A1 |
20110310441 | Simske et al. | Dec 2011 | A1 |
20120051601 | Simske et al. | Mar 2012 | A1 |
20120212324 | Pollard et al. | Aug 2012 | A1 |
20120269427 | Simske et al. | Oct 2012 | A1 |
20120280029 | Simske et al. | Nov 2012 | A1 |
20120286028 | Simske et al. | Nov 2012 | A1 |
20130114113 | Simske et al. | May 2013 | A1 |
Number | Date | Country |
---|---|---|
05242294 | Sep 1993 | JP |
09062778 | Mar 1997 | JP |
09130612 | May 1997 | JP |
1020080022854 | Mar 2008 | KR |
WO 2009142635 | Nov 2009 | WO |
WO 2010134919 | Nov 2010 | WO |
Entry |
---|
Wei Deng, Printer Identification Based on Distance Transform, IEEE 2008. |
International Search Report and Written Opinion for PCT/US2009/044853 dated Dec. 30, 2009 (10 pages). |
Simske, S., et al., “The Image Based Forensic Service for Product Authentication”, HP Tech Con 2008, 10 pages. |
Adams, G., et al., “Adding Robust Anti-Counterfeiting to 2D Barcodes”, HP Tech Con 2008, 3 pages. |
Ingenia Technology, http://www.ingeniatechnology.com/technology.php, last accessed Oct. 28, 2008 and May 29, 2009. |
Cortegra, “Biometric Authentication Technology,” http://www.cortegra.com/ba-biometric—authentication.html, last accessed Oct. 28, 2008 and May 29, 2009. |
Khanna et al: “A survey of forensic characterization methods for physical devices”, Digital Investigation, Elsevier, Amsterdam, NL, vol. 3, Sep. 1, 2006, pp. 17-28. |
Mikkilineni, Aravind et al, “Printer forensics using SVM techniques: Proceedings of the IS&T's NIP21” Int'l Conf on Digital Pronting Technologies, Sep. 30, 2005, pp. 1-4. |
Schulze, Christian et al, “Evaluation of Graylevel-Features for Printing Technique Classification in High-Throughput Document Management Systems”, Aug. 7, 2008,pp. 35-46. |
Wei Deng et al: “Printer Identification Based on Distance Transform”, Intelligent Networks and Intelligent Systems, 2008. ICINIS '08. First Int'l Workshop on, IEEE, Nov. 2008 pp. 565-568. |
Number | Date | Country | |
---|---|---|---|
20120051601 A1 | Mar 2012 | US |