WO2013052924A1 - Procédé et appareil pour l'identification par analyse fractale - Google Patents

Procédé et appareil pour l'identification par analyse fractale Download PDF

Info

Publication number
WO2013052924A1
WO2013052924A1 PCT/US2012/059139 US2012059139W WO2013052924A1 WO 2013052924 A1 WO2013052924 A1 WO 2013052924A1 US 2012059139 W US2012059139 W US 2012059139W WO 2013052924 A1 WO2013052924 A1 WO 2013052924A1
Authority
WO
WIPO (PCT)
Prior art keywords
fractal
image
printing
objects
pill
Prior art date
Application number
PCT/US2012/059139
Other languages
English (en)
Inventor
Adam Hanina
Maurice LEPOUTTRE
Lei GUAN
Original Assignee
AI Cure Technologies, Inc.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from US13/338,602 external-priority patent/US8720790B2/en
Application filed by AI Cure Technologies, Inc. filed Critical AI Cure Technologies, Inc.
Priority to EP12838300.7A priority Critical patent/EP2761536A4/fr
Publication of WO2013052924A1 publication Critical patent/WO2013052924A1/fr

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/52—Scale-space analysis, e.g. wavelet analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/60—Type of objects
    • G06V20/66—Trinkets, e.g. shirt buttons or jewellery items
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06K—GRAPHICAL DATA READING; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
    • G06K19/00—Record carriers for use with machines and with at least a part designed to carry digital markings
    • G06K19/06—Record carriers for use with machines and with at least a part designed to carry digital markings characterised by the kind of the digital marking, e.g. shape, nature, code
    • G06K19/06009—Record carriers for use with machines and with at least a part designed to carry digital markings characterised by the kind of the digital marking, e.g. shape, nature, code with optically detectable marking
    • G06K19/06046—Constructional details

Definitions

  • This application relates generally to the identification of objects using printed identifiers, and more particularly to the identification of small, uneven, differently shaped, or other objects, such as medication pills, using one or more identifiers embedded in a fractal-based printed identifier (or an identifier based upon another mathematical system) to create a visually complex fingerprint.
  • a counterfeit medication or a counterfeit drug is a medication or pharmaceutical product which is produced and sold with the intent to deceptively represent its origin, authenticity or effectiveness.
  • a counterfeit drug may contain inappropriate quantities of active ingredients or none at all, may be improperly processed within the body (e.g., absorption by the body), may contain ingredients that are not on the label (which may or may not be harmful), or may be supplied with inaccurate or fake packaging and labeling.
  • Counterfeit medicinal drugs include those with less or none of the stated active ingredients, with added, sometimes hazardous, adulterants, substituted ingredients, completely misrepresented, or sold with a false brand name. Otherwise legitimate drugs that have passed their date of expiry are sometimes remarked with false dates.
  • Low-quality counterfeit medication may cause any of several dangerous health consequences including side effects, allergic reactions, in addition to their obvious lack of efficacy due to having less or none of their active ingredients.
  • Radio frequency identification which uses electronic devices to track and identify items, such as pharmaceutical products, by assigning individual serial numbers to the containers holding each product.
  • the U.S. Food and Drug Administration (FDA) is working towards an Electronic pedigree (ePedigree) system to track drugs from factory to pharmacy.
  • This technology may prevent the diversion or counterfeiting of drugs by allowing wholesalers and pharmacists to determine the identity and dosage of individual products.
  • Some techniques, such as Raman spectroscopy and Energy Dispersive X-Ray Diffraction (EDXR) can be used to discover counterfeit drugs while still inside their packaging.
  • Marking individual pills with one or more identifiers is considered a useful method for identification, but has been traditionally thought of being cost prohibitive while offering only minimal improvement over packaging marking.
  • One or more barcodes may be employed (either one or two dimensional) and may be printed to individual medication pills, instead of, or in addition to being printed to the medication packaging.
  • Such a printing process may be implemented by employing one or more appropriate printing apparatuses, such as a pad printing apparatus provided by Printing International® N.V./S.A., for example.
  • a pad printing apparatus provided by Printing International® N.V./S.A.
  • Laser marking has also been used to print high-resolution images or barcodes directly onto pills. In consumable products, Mars®, Inc.
  • US Patent No. 7,311,045 describes a system for printing multi color images on a candy by maintaining a directional registration of the candy between printing steps. In each instance, holding each individual medication pill or candy is performed by vacuuming the pieces in place, and holding the piece firmly in place between steps so that orientation of the piece during printing does not change.
  • Other patents and applications assigned to Mars®, Inc. describe a number of systems and methods for printing food grade inks onto shaped candy elements.
  • the described printed elements may fail to provide robust images sufficient to act as a unique identifier for a particular batch of processed elements.
  • any damage to the barcode, difficulty in printing the barcode (such as deformation based upon printing surface), or obscuring a portion of the barcode may render the barcode inoperative.
  • barcodes may be easily copied and applied to counterfeit objects. None of these systems are sufficient for imparting robust identification information to a pill or other candy object. [010] Similar problems of identification of other products or objects, such as consumer products and the like, may also arise.
  • a standardized process for labeling and identifying medication and other objects may comprise a system and method for printing an identification pattern to a plurality of regularly or irregularly shaped and arranged objects.
  • a fractal pattern is preferably printed as the identification pattern onto a plurality of medication pills or other objects.
  • computer vision may be applied to read these applied fractal labels to automatically confirm identification and authenticity of the pills or other objects, irrespective of orientation, partial occlusion, or partial damage of the printed fractal image.
  • fractal dimensions the number of times the fractal pattern is recursively printed within itself
  • different levels of fractal dimensions may be printed and therefore read from these fractal images employing different resolution imaging devices, providing varying levels of security and precision in the identification process while allowing for ease of identification and a reduced usability burden.
  • imaging devices with lower resolution imaging capabilities may be able to resolve one or two fractal dimensions and provide a lower, less expensive, consumer oriented level of authentication, while higher resolution imaging devices may be able to resolve seven or more fractal dimensions, thus providing additional security where desired.
  • Overt on-product marking including holograms, packaging graphics, etc. Overt features allow the public to see whether a medication is false or not, which may be useful during the drug taking process to flag counterfeits, and to allow individuals to confirm that they are taking the correct medication.
  • Covert marking including invisible ink, embedded images and watermarks, etc.. Covert features are unidentifiable to the public and are usually places on the medication packaging. Such covert markings provide a higher level of security than overt marking as they are more difficult to copy.
  • Forensic marking including chemical, biological and DNA taggants.
  • Forensic features which are integrated into the physical properties of the medication, are available for even higher security authentication or where scientifically tested authentication of the item may be required. Of course, destruction of the product may be required for authentication, and a change to the manufacturing process of those items is necessary for implementation.
  • Track and trace solutions including bar codes, serialization, Radio Frequency Identification/RFID tags, etc.
  • Track and trace technologies allow for near real-time medication tracking throughout the supply chain from the manufacturer to the pharmacy , and to the patient, and have been typically employed by entities along the supply chain. While end users have not been traditionally included in this solution, as noted above, such systems are becoming more readily available for end users.
  • Fractal geometry is a branch of mathematics that deals with geometric objects that are too detailed to be described by standard Euclidean geometry [4].
  • fractal dimension the "fractal dimension" of the fractal is greater than its topological dimension (i.e., 1 for a line, 2 for a square, 3 for a cube).
  • a fractal is a rough or fragmented geometric shape that can be split into parts, each of which is (at least approximately) a reduced-size copy of the whole, a property called self-similarity.
  • a fractal often has the following features:
  • More complicated fractals having a greater number of dimensions may be employed for use with objects needing higher levels of security, as these more complex fractals are more difficult (or impossible) to copy, requiring at least higher quality printers and readers that may not be easily available.
  • a low-cost tool such as a webcam combined with computer vision software may be provided to a patient to properly image and identify a medication at a lower level of security.
  • the fractal image provides a robust identification system that is resilient against identification when partially occluded, or when printing is imperfect because of object shape, position, surface texture or the like.
  • one or more predetermined characteristics of a pill or other object may be employed to be used as part of an identifier for the object.
  • the fractal image used to prevent against counterfeiting as it is difficult or reproduce, but further various characteristics of the object, such as color, shape, texture, markings and the like, may combine with such a fractal image to produce a unique fractal/object characteristic combination.
  • the fractal image may act as a calibration tool to determine any influence the color, for example, of the pill may have on the actual fractal color, thus allowing for an accurate determination of pill color.
  • these noted characteristics of the object may cause one or more distortions in the shape, color, or other attribute of the printed fractal image, these distortions being potentially resolvable and recognizable at differing desired imaging resolutions, providing differing levels of security.
  • barcodes include thick lines and may hide any details of geometry changes.
  • the resulting identification images are robust even if portions of the printed images are not properly printed, or are damaged, obscured or otherwise occluded.
  • these pills or other objects need only be maintained in approximately a single layer during the printing process. Strict orientation and arrangements of the medication pills during printing or subsequent imaging is not required (in that fractal patterns are recursive patterns that may be viewed at any level of detail and give the same information). Additionally, the entire portion of the fractal need not be properly printed on the pill, providing additional robustness in the printing process. Thus vacuuming of the pills in a particular orientation for printing is not necessary, allowing a relatively disorganized set of pills to be printed at one time.
  • multiple fractal patterns may be overlaid, thus producing a more complex identification image.
  • These overlaid fractals may be electronically combined before printing, thus requiring a single printing pass, or may use multiple printing passes, thus, printing multiple fractals at different times on the pills.
  • These overlaid fractal patterns may also be selected to provide different imaging results, such as a first fractal image allowing for a more accurate measurement of shape and color of the medication pill, and another fractal image providing various medication information.
  • the geometry of the pill and any unique or identifiable geometric characteristics may be more accurately determined through a measurement of distortion of the fractal images by the shape of the pill.
  • a conveyer mechanism may be provided for forwarding, in either a continuous or batch processing manner, a plurality of medication pills to a printing area, in addition to the use of the individual pill printing schemes of the prior art. If forwarded in a batch manner, a fractal image is preferably printed on the area containing the medication pills in a manner employing, by way of example only, ink jet printing technology of the type described above with reference to the '045 patent, the contents thereof being incorporated herein by reference.
  • one or more medication pills may be coated with a laser or light sensitive material that allows for marking of the pill, while being consumable by humans.
  • a single image may be provided to the plurality of medicine pills at one time. In such a manner, each pill will receive only a portion of such an image. Because of the self referential nature of fractals, this part of the image will be sufficient to provide an identifiable amount of information that may be recovered with computer visual recognition technology. Of course, batches of such pills may be manually or otherwise placed in a printing area, and removed after printing.
  • fractal pattern that may have bounds in the direction across the direction of travel of the plurality of pills, but is continuously repeatable in the direction of travel, may be applied.
  • a very large number of such pills may be processed continuously and inexpensively, and therefore in accordance with other continuous manufacturing and processing of such pills.
  • the invention may be applied to any objects in which such an identification application may be beneficial, including any type of small, uneven, or irregularly-shaped object part, including for pricing, and anywhere a traditional bar code or two dimensional bar code may be employed, and where the integrity of the item makes it important to track and identify the object.
  • L-systems consist of an alphabet (a set of characters) and production rales, which specify the rales for character replacement. Similar to Chomsky grammars, the main difference is that L-systems apply production rales simultaneously rather than sequentially. If each line segment of an initiator is represented by a single character and the production rules substitute this character with a string representing a generator, then there is a direct mapping between L-system strings and fractal curves. This mapping provides the ability to represent a fractal curve in a computer as a sequence of characters.
  • the production rule for this generator is F ⁇ F + F - F - FF + F + F - F.
  • F means moves forward by one unit and draw a line segment
  • + means turn left
  • - means turn right, and is set at 90 degrees.
  • FIG. 24 illustrates the construction of a Peano space filling curve, named after Guiseppe Peano who first discovered them.
  • the alphabet is the set ⁇ F; + ; -; R; L ⁇ where ⁇ F; +; -) have the same interpretation as the quadratic Koch curve discussed above, and ⁇ R.
  • L ⁇ are place-holder characters used to generate the L-system strings.
  • the initiator is L, and there are two productions rules: pi: L ⁇ LFRFL - F - RFLFR+ F + LFRFL and p 2 : R ⁇ RFLFR + F + LFRFL - F - RFLFR.
  • Fractals can be represented as an Iterated function systems (IFS). IFSs were first introduced by Hutchinson (J. E. Hutchinson, "Fractals and self-similarity," Indiana Univ. Math. J. 30 (1981), pp. 713-747), and popularized by Barnsley. M. F. Barnsley, Fractals Everywhere (Academic Press, San Diego, 1988). IFS provide a simple and convenient definition for many fractals. An IFS consists of a finite
  • IFS's have unique attractors, some of which may be fractals.
  • the typical requirement for an attractor to be a fractal is that it have non-integer dimension (this is discussed below). If all the contractions in the IFS are similarities, then the fractal is self-similar, i.e., composed of exact copies of itself. Otherwise, the fractal is self-affme, i.e., approximately self-similar.
  • the attractor of the IFS may then be computed using either the deterministic or random iteration algorithm, and tested for its fractality with user-defined parameters such as fractal dimension. If accepted, the fractal can then be saved in a database by the numerical values comprising its IFS representation, along with its fractal parameters.
  • Figure 10 illustrates the deterministic algorithm for the Sierpinski triangle based upon 10 iterations of the deterministic algorithm for the Sierpinski triangle, starting with the unit square.
  • an initial set therefore comprises a unit square.
  • a first iteration shows the replacement of each square in the first iteration with a set of three smaller squares.
  • a similar substitution is made, replacing each square in the current iteration with the three square configuration.
  • the complexity of the image is greatly increased.
  • a shown 10 th iteration substantial complexity is achieved. This fractal is generated by three contractions:
  • Attractors of an IFS with non-integer dimension are considered fractals (ex: the Sierpinski triangle has dimension D « 1.585).
  • Dimension is a frequently referenced parameter when discussing fractals, and its estimation is an area of much research ⁇ cite ⁇ Theiler90 ⁇ .
  • Dimension can be expressed as the exponent that relates the scaling of a geometric object's bulk (area, volume, mass, etc.) with it's size (length, perimeter, diameter, surface area, etc.):
  • the second method for generating such a fractal may comprise a random iteration algorithm.
  • a random iteration algorithm preferably works in the following manner according to the following algorithm:
  • the algorithm preferably produces one point at a time.
  • the algorithm preferably outputs a sequence of points which may then be plotted. Detailed images can be created from many (> 100, 000) iterations.
  • Figure 11 presents an example of the random iteration algorithm for a self-affine fractal. As the number of points computed increases (moving from portions a to b to c in Figure 11 the algorithm fills in more of the attractor and provides a more accurate approximation.
  • Figure 11 illustrates the random iteration algorithm for the fractal defined by the foll ove:
  • the inventors of the present invention have recognized that one of the principal features of fractals is their complexity. Complexity here is defined as fine detail at arbitrary resolution levels. In the case of fractal curves generated by L-systems, a few simple rules may yield extremely complex curves after just a few iterations.
  • One measure of complexity is the number of line segments created at each iteration. In figure 21, the generator is an equilateral triangle, and the initiator consists of 4 line segments. Hence, at the th iteration, the curve contains line segments. At iterations, the Koch curve already has line segments. In figures 2 and 3, the growth is even greater; at iterations, the quadratic Koch curve has line segments, while the Peano space-filling curve has line segments per subsquare subsquares connecting line segments line segments.
  • each line segment is drawn with one of two styles (different colors, different line types, etc.), then it can be treated as a single bit of information. Therefore, in accordance with one or more embodiments of the present invention, every eight line segments may then be treated as a byte representing an ASCII character. Since the line segments are drawn sequentially from the string generated by an L-system, order is preserved and can be used to encode strings of data. Using these features, one can therefore encode information directly onto the fractal curve and treat it as a barcode. To provide robustness to occlusion, data strings can be encoded multiple times on a fractal curve.
  • Error correction can also be built-in if necessary.
  • fractal curve barcodes will be robust to geometric distortion (i.e., printing on curved surfaces) since it is the sequencing of line segments that matters, not their absolute position on a grid or in a matrix. Alignment will not be an issue, as sequences can be encoded into the fractal curve to signal the beginning and end of the data string.
  • Kiani et al. described above, have discussed the use of fractal curves as identifiers.
  • the invention set forth in accordance with the present application differs in several ways from their work: type of curve;
  • Kiani et al. only describe the use of Hilbert curves, a type of space-filling curve.
  • the Koch curve in Figures 22 and 23 are examples of non-space-filling curves.
  • the present invention encompasses any fractal curve that is non-intersecting and can be generated by an L-system. Thus, the reach of the present invention may be applied to a far greater group of fractal images.
  • Kiani et al propose encoding binary information by printing or not printing line segments.
  • the inventors of the present invention had determined that at least one difficulty with this approach is that in the decoding process, it is impossible to tell whether a line segment is missing because it represents a particular encoding or if it is missing because of occlusion, damage, noise, etc. Therefore, this most important aspect means that upon acquisition of a sequence of missing line segments, it is not possible to determine whether these missing segments are intended to be missing encoding segments, or whether a portion of the image is occluded. Once this inability to differentiate is present, the identification system employing the system cannot be trusted.
  • Kiani et al use a Hopfield neural network to recover the line segments from an image of a fractal curve. This requires training a computer to recognize various patterns. Various embodiments of the present invention employ line segment detection methods that do not require any training, only standard computer imaging techniques.
  • the invention accordingly comprises the several steps and the relation of one or more of such steps with respect to each of the others, and the. apparatus embodying features of construction, combinations of elements and arrangement of parts that are adapted to affect such steps, all as exemplified in the following detailed disclosure, and the scope of the invention will be indicated in the claims.
  • Figure 1 depicts an exemplary medication pill with an exemplary fractal image printed thereon in accordance with an embodiment of the invention
  • Figure 2 depicts an increased resolution portion of the exemplary fractal image of Figure i ;
  • Figure 3 depicts an exemplary medication capsule with an exemplary fractal image printed thereon in accordance with an alternative embodiment of the invention
  • Figure 4 depicts a distorted fractal image printed on a portion of a medication pill in accordance with yet another an embodiment of the invention
  • Figure 5 depicts a conveyer system for conveying batch processing groups of objects for processing in accordance with an embodiment of the invention
  • Figure 6 depicts a group of pills for batch processing in accordance with another embodiment of the invention.
  • Figure 7 depicts application of different fractal patterns to a group of medication pills in accordance with an embodiment of the invention
  • Figure 8 is a flowchart diagram depicting a printing and imaging process in accordance with an embodiment of the invention.
  • Figure 9 is a flowchart diagram depicting image processing in accordance with an embodiment of the invention.
  • Figure 10 illustrates output from a deterministic algorithm for a Sierpinski triangle in accordance with an embodiment of the invention
  • Figure 11 depicts an example output from a random iteration algorithm for a self-affine fractal in accordance with an embodiment of the invention
  • Figure 12 illustrates a computation of box counting method for the Sierpinski triangle in accordance with an embodiment of the invention
  • Figure 13 depicts application of an IFS to different fractals in accordance with an embodiment of the invention
  • Figure 14 depicts a deterministic algorithm for the Sierpinski triangle, starting with an initial image including numbers in accordance with an embodiment of the invention
  • Figure 15 depicts a zoomed in portion of Figure 14
  • Figure 16 depicts encoding of the word "fractal" on a fractal using contraction mapping that generated the fractal in accordance with an embodiment of the invention
  • Figure 17 depicts encoding a barcode in different levels of a self-similar fractal-like image in accordance with an embodiment of the invention
  • Figure 18 depicts a fractal image similar to that of Figure 17, but with additional complexity added by removing one or more portions of the image in accordance with an embodiment of the invention
  • Figure 19 depicts a standard fractal progression with modifications thereto to provide additional complexity to the image in accordance with an embodiment of the invention
  • Figure 20 depicts a plurality of image cells in accordance with an embodiment of the invention.
  • Figure 21 depicts a further image including iterations of a complex image in accordance with an embodiment of the invention.
  • Figure 22 depicts initiator, generator and a number of iterations of a Koch curve
  • Figure 23 depicts construction of a quadratic Koch curve using L-systems
  • Figure 24 depicts construction of a Peano space-filling curve using L-systems
  • Figure 25 depicts encoding of a fractal curve in accordance with an embodiment of the invention.
  • Figure 25 depicts encoding of a fractal curve in accordance with an embodiment of the invention.
  • an ink jet, laser marking system, or other printing or etching system may be employed in order to print a plurality of medication pills with a predetermined fractal image.
  • Other acceptable printing systems may also be employed, and may include, by way of example only, and without limitation, laser printing, laser etching, photographic imaging, photolithography techniques, solid ink printing, or the like.
  • One or more known edible ink products may also be employed in the printing process. Invisible, UV sensitive, heat sensitive, and other appropriate inks may be employed.
  • Each medication pill may further be coated with a laser or other light sensitive or otherwise sensitive coatings that, when caused to react, may be employed to mark the medication pills, while remaining edible by humans.
  • single or multiple color printing may be employed.
  • a fractal image 110 is shown printed to a medication pill 100.
  • the fractal image covers a substantial portion of the pill surface while possible still allowing for viewing of the pill surface based upon printing techniques, pill positioning, fractal selection, etc.
  • fractal image 110 may cover substantially less than all of the surface of medication pill 100.
  • fractal image 110 is once again printed to medication pill 100. Also shown is a magnified portion 210 of fractal image 110. Because of the self-similarity nature of fractal images, as can be seen, magnified portion 210 of fractal image 110 looks substantially similar to the whole of fractal 110, and is theoretically identical. This self similarity continues, theoretically, to an infinitely small printed image. Of course in practice, the levels (or dimensions) of self similarity available are limited by printing resolution, and the ability to "see" these multiple dimensions may be dependent upon an imaging resolution of an imaging device. The present invention exploits these features of reality in order to provide a varied solution applicable in different security situations.
  • fractal patterns is not limited to pill-shaped medication.
  • a fractal image 310 may be printed to a capsule medication 300 comprising first and second capsule portions 301 and 302.
  • the fractal image may take the shape of the capsule, and may be distorted thereon in a predictable manner.
  • any discrepancy between positioning of fractal images on the capsule portions 301 and 302 may be employed to determine the possibility of the capsule having been opened or otherwise tampered with.
  • solid capsule shaped pills may also be printed, but will not include capsule portions.
  • the use of high resolution printing and imaging techniques may be employed when more robust security measures are required, increasing the difficulty of copying such an image, and leading to greater accuracy in identification of each item.
  • Determinations of the maximum resolution of such printed fractals may be identified by building such fractals from a smallest possible printed pattern, thus bounding the lowest fractal resolution.
  • such fractal images may be generated in a traditional form, by starting with a largest fractal image, and then dividing these larger images into smaller and smaller objects, until a predetermined limit or printing resolution is reached.
  • Such increased resolution may also allow for more precise measurements of the details of the fractal image, such as one or more distances between various portions thereof, ratios of one or more various measured lengths or angles of various portions thereof, etc., texture of the medication pill surface, color of the medication pill surface, etc. which may then be compared to expected values to confirm authenticity and identification.
  • fractals may be employed based upon a desired level of security.
  • more complex versions of fractal images may be employed where a more secure identification system is desired.
  • Variations in such fractal images may include changes in angles, length, number of pixels employed, distribution of one or more characteristics or pixel density, purposeful omission of particular pixels, use of particular color combinations on a planned or randomized basis.
  • a higher resolution printing process may also be employed, thus allowing for more precise printing of multiple fractal dimensions, and eventual recall and analysis of the fractal images by higher resolution imaging devices at deeper acquisition resolution.
  • combination fractals may be employed to provide additional robustness against counterfeiting, and for determining identity.
  • fractals may be particularly chosen to allow for determination of different types of information.
  • a first or more fractals may be employed to measure for detection of distortion that may be a result of the shape of the pill, while a second or more fractals may be employed for encoding information and for prevention of replication of the fractal images.
  • various color gradient application may allow for the calibration of the fractal image, the pill, or other object.
  • a desired level of security may first be defined at step 810. Then, at step 820, a corresponding necessary printing resolution may be determined, and at step 830, one or more appropriate printing technologies may be identified that will provide sufficient printing resolution to allow for the desired level of security. Finally, at step 804, a predefined fractal image is printed to the medication pill using the determined printing technology.
  • steps 810, 820 or 830 are predetermined, they can be skipped, or at a minimum, responses to these steps can be predefined.
  • imaging steps may be employed.
  • a desired level of identification security may be defined at step 850, and thereafter, at step 860, a corresponding imaging apparatus may be selected.
  • a webcam associated with a mobile device or the like may be employed to image a high level and one or more additional dimensions of the fractal image, even if substantially more fractal dimensions have been printed.
  • an imaging device able to image to a much higher resolution, thus allowing for the confirmation of existence of any desired number of fractal dimensions, may be employed.
  • the selected imaging apparatus may be used to acquire an image of the printed fractal and pill.
  • the identity of the medication pill may be confirmed to the desired level of security. As with printing, if predetermined, any of steps 840 and 850 may be skipped or predetermined (as if the user only has a single imaging device).
  • user may present a medication pill with a fractal identification image printed thereon to a webcam or other more sophisticated imaging device at step 910.
  • This device may provide local identification and confirmation of the medication, or may forward such information to a remote location for further processing, and a processing step 915 to make this determination may be employed, or such a determination may be made in advance.
  • a processing step 915 to make this determination may be employed, or such a determination may be made in advance.
  • an acquired image or video sequence of images may be transmitted to a remote location for processing.
  • remote processing may be performed, and at step 930, results of such processing may be returned to the imaging device.
  • step 915 If at step 915 it is determined that local processing is to take place, then processing passes to step 940 and the pill is analyzed locally. After such analysis, the user is notified of the authenticity of the pill at step 950.
  • the remote server or local device may analyze the imaged pill, identify the pill, and may indicate a determination of authentication or counterfeit. If counterfeit (as determined locally or remotely), the user may be instructed to not take the pill, or alternatively that the pill is authentic in conjunction with step 950.
  • a fractal image 410 may be printed to a medication pill 400, and where a portion 411 of fractal image 410 may be printed on a vertical or other portion 401 of pill 400 other than a front face thereof.
  • portion 411 of fractal image 410 is printed over a pill feature 412, the edge.
  • Printing over edge 412, and along vertical portion 401 will once again cause the fractal image to be distorted in a predictable manner. This predictable distortion may be used to further confirm that the pill is authentic, placing yet another barrier to a counterfeit medication.
  • recognition of predictable distortion of the fractal image and additional measures to avoid such copying may also be employed in accordance with various embodiments of the invention.
  • a fractal pattern may be calibrated to include one or more of object pattern, shape, texture, markings, line thickness (such as through the use of thicker inkjet lines, or by altering a wavelength of a marking laser, for example), or the like.
  • object pattern shape, texture, markings, line thickness (such as through the use of thicker inkjet lines, or by altering a wavelength of a marking laser, for example), or the like.
  • line thickness such as through the use of thicker inkjet lines, or by altering a wavelength of a marking laser, for example
  • various calibration lines or the use of symmetrical fractal patterns may be employed for distortion detection.
  • incorrect object color may change an overall color of the fractal image applied thereto, thus indicating a non-authentic object.
  • ink may be distributed in a known manner, creating a unique signature and allowing for any recognition system to better determine the shape of the object more accurately.
  • a plurality of fractal images may be overlaid on an object, thus making copying even more difficult as various interactions between the various images may be more difficult to determine.
  • these attributes may influence perception of the fractal image by an imaging system.
  • a fractal image may be deformed or otherwise influenced in predictable ways, thus allowing for the user of such shape to be employed to further differentiate authentic objects.
  • deformation of such a fractal pattern may be determined in a number of likely orientations of an object, and then may be so classified and found on printed objects. Further, if orientation of the object can be controlled for printing, then precise fractal deformation may be determined. As such imaging systems may learn such expected fractal deformation, the deformation may be employed as part of the object identification system.
  • information may be coded into the fractal, by placing such coding information into one or more parameters that may be stored in a parameter file used to generate the fractal.
  • a batch number or the like may be used in the place of particular parameters to be chosen by the user.
  • Recognition of the fractal, and reverse engineering thereof to recreate the parameter file may then provide access to the batch number or the like by the user, for example. It is anticipated that only a predetermined number of parameters may be employed for particular coding, others of the parameters being adopted to vary in a pseudo-random or other predetermined manner in order to make it more difficult to predict future likely parameter combinations.
  • fractal images may be printed at manufacture, at distribution, or in combination.
  • a manufacturer printed fractal may be provided for counterfeit prevention, while a second fractal, printed on top of the manufacturer fractal, or alongside thereof, may be provided and printed by, for example, a pharmacist or the like.
  • These fractal images may include particular prescription, patient, prescribing doctor, date, and other patient and administration specific information.
  • the manufacturer information may be provided in accordance with a fractal information, while the particular patient information may be provided by a one or two dimensional barcode, or other information providing process.
  • the local or manufacturer fractal images may be used.
  • randomly generated fractals may be employed and recognized from a lookup table to be associated with a particular batch processing unit.
  • Other methods may provide a library of fractal images, and indications of which fractals are to be applied to different type, shaped, or colored objects. Thus, it may be determined that a particular type of fractal image is best applied to a particularly shaped object. Next, from this subset, a further subset may be determined as best for the particular color of the object. A fractal image from this subset may then be used or encoded further, and then applied. As will be further described below, such a hierarchical selection process may also speed the eventual fractal acquisition and recognition process.
  • Colors may be omitted from the fractal printing process deliberately in order to increase the number of variations of the code. For example, omission of specific areas may indicate batch number or date. Furthermore, a range of colors may be included with absolute colors such as black, white, etc. acting as reference points. Use of such a range of colors allows for more patterns to be created and utilized, thus increasing a range of possible unique fractal images. Selection and/or omission of particular pixels in such a fractal image may be further used for variation to allow for randomization of predetermined fractal images.
  • a conveyer mechanism 510 forwarding one or more batch processing units 520.
  • Conveyer mechanism 510 is shown as a gravity-fed mechanism including a plurality of rollers, but any such conveyer system may be employed, including gravity-fed, belt driven or otherwise powered conveyer systems, and may further be provided with or without a belt system for conveying the batch processing units 520.
  • any method for forwarding the batch processing elements, including hand delivery of the units may be employed.
  • conveyer mechanism 520 may comprise any desired method, apparatus or system for placing one or more objects in a location to be imaged in a manner as will be described below.
  • each batch processing unit 520 contain a plurality of individual objects, and in accordance with a preferred embodiment of the invention, a plurality of medication pills or the like.
  • a plurality of medication pills 610 are shown in Figure 6.
  • medication pills 610 are preferably arranged in batch processing unit 520 in an unstructured manner, but generally in a single layer. While slight overlap may be tolerable in accordance with the invention, a single layer presentation of the medication pills will allow for maximum exposure of the pills to an imaging apparatus, shown at 530 in Figure 1.
  • Such a batch processing unit may comprise from one to any number of properly physically locatable pills, and may further comprise a physical structure for holding the pills, or may simply comprise a conveyer or other forwarding or holding mechanism for presenting the one or more pills to a printing mechanism.
  • imaging apparatus is employed to administer a predetermined fractal image to the plurality of medication pills 610 at one time.
  • Such printing may comprise a raster printing system, or may print or otherwise transfer a complete image to the plurality of medication pills substantially simultaneously.
  • each medication pill 610 will be printed with a portion of the predetermined fractal image.
  • these portions will include sufficient information to allow for proper identification of various fractal dimensions at various desired imaging resolutions, thus providing unique flexibility in imaging based upon desired security levels.
  • each medication pill 610 is preferably forwarded for processing to have a fractal pattern imparted thereon.
  • a fractal pattern may comprise a repetitive pattern 710, a continuous pattern 720, or other desired fractal image. Each may be used in either a batch processing or continuous processing situation.
  • Such fractal patterns may further comprise one or more combination fractal patterns, in which two or more fractal patterns are combined to provide a resulting complex fractal pattern.
  • These complex patterns may be combined before printing, thus imparting the complex fractal pattern in a single printing pass, or alternatively, each pattern may be printed in a separate pass, thus layering the two or more individual fractals to provide a resulting complex fractal image.
  • a continuous processing system may also be employed.
  • a similar conveyer belt may be employed, in which medication pills or other objects are continuously passed beneath imaging apparatus 530, and may preferably be employed in conjunction with a continuous manufacturing process of such objects.
  • a fractal pattern may preferably be chosen that may be continuously replicated in a direction of travel of the medication pills, while being bounded in the direction across travel, or may be easily repeated in the direction of travel so all medication pills or other objects are printed with at least a portion of the fractal pattern.
  • a particular image processing apparatus 530 may affect how the printing ink is distributed on the surface of the medication pill or other object, thus, in combination with the printed fractal pattern, producing a printer signature, i.e. a printer specific rendering of the particular fractal pattern.
  • a printer signature i.e. a printer specific rendering of the particular fractal pattern.
  • the particular characteristics of the printer including nozzle tolerance, humidity and a host of other factors may influence an output fractal image.
  • Such a printer signature may be further used as an identifying feature of the printed fractal image.
  • Such a signature may be similarly determined when other imaging techniques, such as those described above, are employed.
  • This idea of printer signature may also be extended to product signature and camera (or more broadly, imaging apparatus) signature.
  • a product signature may be based at least in part on how the product absorbs the ink, how the pattern at higher resolution distributes the ink, texture and reflectivity of the object, shape of the object, etc.
  • Unique texture, shape, color specific to the object (pill) will "code” or distort the fractal into a unique ID.
  • a camera or other imaging apparatus used to eventually image the printed fractal image may also have a unique signature in distorting the fractal image that may also act as an increase in security as it may be difficult to anticipate a camera that is to be used for imaging, if not an authentic system.
  • These signatures as opposed to being deficiencies of the system, may be embraced to strengthen the robustness of the system. Such image influences resulting from unrelated characteristics of systems used to implement the system will be difficult/impossible to replicate. Through a decision fusion process combining the results of analysis of any one or more of these attributes, an overall picture and confidence of authenticity or counterfeit may be determined.
  • the different hardware and pill interaction signatures including shape, texture, color of the object, or the like will all help to further "code” or distort the fractal in a unique way.
  • the unique attributes of the product / item will help to create a unique fingerprint for the fractal.
  • the inventive system is therefore able to learn the unique characteristics of the product through computer vision training or the like, and not simply apply an out-of-context code to the item.
  • a threshold may determine confidence over suspicion of counterfeit (i.e. how confident the system is that the item is authentic).
  • Many instances of low confidence levels (even if above threshold levels), as received and accumulated over time from any number of different users at a centralized location, may indicate a potential counterfeit issue and raise a flag remotely to anti-counterfeit authorities to double check a medication source, or alert a user to report the possibility of a counterfeit medication source.
  • the relative positioning of the ink jet dots may be changed, in order to be arranged, for instance, in a form of a proximity to a center of some printed object or other marker or attribute.
  • these same dots may be use to allow for the encoding information even in cases when only low printing resolution is available, but high magnification image acquisition may be available when the pill or object is to be identified.
  • any encoded ID is repeated within the shape.
  • the ID may still be determined.
  • Many varieties of data can be encoded in the fractal (date, time, batch number, location, manufacturer, dose, item etc).
  • writing or reading of the fractal image does not require alignment of the object perfectly as only part of the fractal may be good enough (may need a percentage of the fractal to be printed on a surface for a specific webcam resolution and inkjet resolution).
  • various reference points may be used to aid in determining authenticity.
  • distances and ratios to other reference points may be determined and used to confirm authenticity, or a level of confidence in that authenticity, in a manner as described above.
  • an object may be scanned, and existence of at least a portion of such a fractal may be determined. If there is enough information in the portion of the fractal image to confirm identity, then identity is confirmed.
  • various pieces of fractal images may be pieced together to determine enough information.
  • an occlusion or the like may be effectively disregarded through such piecing together of the fractal image.
  • Such information to be pieced together may be taken from one or multiple fractal dimensions.
  • a user may be instructed to bring such an object closer to an imaging apparatus (or otherwise zoom in on the object), or be asked to switch to another medication pill for re-identification, in order to improve capture resolution. This may be particularly important in difficult imaging environments, such as in the existence of bad lighting conditions or the like, which may reduce a confidence of precision of imaging.
  • Such improved resolution may be employed alone, or in conjunction with anticipated effects from one or more object attributes, as noted above, in order to identify an authentic printed fractal image.
  • any applicable imaging system such as a high resolution system, or a webcam system, will benefit greatly over the use of barcoding.
  • imaging of the fractal image can be performed at any angle of the pill.
  • various augmented reality solutions may be employed in order to properly image the medication pill and fractal image, thus truly freeing up the user to image the fractal image without any real issues regarding orientation or placement of the pill.
  • Such augmented reality solutions may also provide additional information regarding the medication pill, including patient name, medication administration schedule information, prescribing doctor's name, contact information, or any other information that may be useful for the user to view.
  • an authentication, or counterfeit such information may be provided to a remote location to accumulate such information.
  • Each pill identification instance will result in an authenticity confidence score.
  • instances of low authenticity confidence may be reported to a centralized location, along with a medication image, GPS data, as well as time and date stamps.
  • Higher-level authentication tests may be carried out at local pharmacies using higher resolution imaging devices.
  • a geographic nexus of particular counterfeits will be determined alongside likely illegal distribution channels, thereby aiding anti-counterfeit officials.
  • investigations may be employed in that area. Further, proper identifications can be confirmed.
  • Variation in batch coded information may be employed in order to further allow for confirmation of particular medication generation time and location stamps.
  • Such information may be forwarded over the Internet or other transmission system, such as transmission over a cellular telephone connection or the like, to a centralized location for analysis and accumulation, for example.
  • Selection of the actual patterns to be employed may be performed in accordance with consideration of one or more parameters to be encoded into the fractal image, and further based upon a surface or medication pill upon which the fractal image is to be printed. As different information may be encoded into each fractal pattern, the selection and encoding of this information will make changes to the fractal pattern in subtle manners. Based upon a printing surface, expected distortion, amount and type of information to be encoded, printing technology to be employed, or level of resolution in printing and imaging desired, different fractal patterns may be preferred and employed. In fact, each such printed fractal provides a multi-dimensional pattern that comprises the above noted fractal signature. These dimensions may include one or more of fractal image, texture of the surface, color of the medication pill and shape and contours of the medication pill. These features may be employed to aid in object recognition.
  • selection of particular types of base patterns may be performed in accordance with one or more particular tasks, pills or desired results.
  • one or more simple fractal patterns such as a Cantor fractal patterns may be employed or lower security situations where identification is most important.
  • More complex types of fractal base images may be employed for other, security intense applications.
  • Image analysis may incorporate such techniques as global and local feature detection and description, histogram analysis, image thresholding, image segmentation, etc.
  • Global features may include shape, contours, image moments, object/pattern orientation, etc.
  • Local features may include corners, blobs, line segments, gradients, keypoints, anchor points, etc.
  • the information extracted at this stage may then be used for comparison between test images and stored images. The information may be compared in various ways, including the use of feature vectors.
  • fractal dimension is usually estimated J. Theiler, "Estimating fractal dimension," J. Opt. Soc. Am. A 7, 6 (1990), pp. 1055-1073.
  • estimators of dimension are the box-counting dimension and the correlation dimension. Both of these are instances of generalized dimension:
  • the number of boxes that contain a part of the attractor is counted, and the (box-size ⁇ number of boxes) pairs are plotted on a log-log plot, shown in portion d of Figure 12.
  • the slope of the regression line in this particular instance is calculated as 1.58496.
  • fractal dimension estimation of an image to be tested may be implemented in the following manner:
  • Test image may be provided from any standard camera, such as on cellphones or tablet pc's. Any preprocessing of the image (thresholding, rotation, perspective transformations, Gaussian filters, Laplace filters, etc.) may be implemented here.
  • the image may be applied to any desirable surface, in accordance with the invention, and may in particular be applied to a medication pill or other object for which authentication or identification is desired.
  • the sizes r will typically range from L/2 to 1 , where L is the side length of the image.
  • the results calculated in step 2 are preferably compared to results in a list of possible IFS fractals, and the IFS code with the closest fractal dimension is selected.
  • Such a list of possible fractals are preferably printed, and this same calculation is performed for each, thus generating a table of dimension values for different fractal images that may be applied to an object or otherwise printed in accordance with one or more embodiments of the invention.
  • a stored dimension value that is closest to the measured and determined fractal value is preferably indicated as the matching fractal.
  • a user-defined threshold may preferably be implemented so that if the difference between the computed dimension and expected dimension exceeds the threshold, then no IFS code may be selected from the database and a null value is preferably returned.
  • This threshold makes sure that only close images may be considered, and provides a level of confidence that there is an actual match between the computed dimension of the images fractal image, and the expected dimension of one of the stored fractal images.
  • an inventive lacunarity analysis may be implemented to determine fractal dimension when the image being analyzed is a fractal.
  • Lacunarity may be estimated by a gliding-box algorithm.
  • Such a gliding box algorithm provides that for each pixel p in an image, a box of radius r is preferably centered at p, and the number of pixels s in the box which are part of the fractal pattern is counted.
  • n(s, r) is the number of boxes of size r with s pixels.
  • the frequency distributions may be converted into probability distributions Q(s, r) by dividing by the total number of boxes N(r) of size r.
  • the first and second moments may then be computed as
  • lacunarity estimation of a test image may be implemented in the following manner:
  • Test image can be provided from any standard camera, such as on cellphones or tablet pc's. Any preprocessing of the image (thresholding, rotation, perspective transformations, etc.) will be implemented here.
  • the image may be applied to any desirable surface, in accordance with the invention, and may in particular be applied to a medication pill or other object for which authentication or identification is desired.
  • the scales r will typically range from 1 XoL/2, where L is the side length of the image.
  • step 2 Compare the results of the algorithm to results from stored images:
  • the results calculated in step 2 are compared to the results in a list of possible IFS fractals, and the IFS code with the closest lacunarity is selected.
  • Such a list of possible fractals are preferably printed, and this same calculation is performed for each, thus generating a table of lacunarity values for different fractal images that may be applied to an object or otherwise printed in accordance with one or more embodiments of the invention.
  • a stored lacunarity value that is closest to the measured and determined lacunarity value is preferably indicated as the matching fractal.
  • a user-defined threshold may preferably be implemented so that if the difference between the computed lacunarity and expected lacunarity exceeds the threshold, then no IFS code may be selected from the database and a null value is preferably returned.
  • This threshold makes sure that only close images may be considered, and provides a level of confidence that there is an actual match between the computed lacunarity of the images fractal image, and the expected lacunarity of one of the stored fractal images.
  • lacunarity may be alignment dependent, depending on the fractal image employed, a method for aligning such a fractal image may be employed.
  • alignment points may be provided to allow for such alignment. These alignment points may be provided as part of the fractal image (comprising a modification thereof), or may be printed on top of the fractal image to indicate alignment.
  • this lacunarity step may be used alone or in conjunction with the dimension calculation described above to determine a matching fractal.
  • feature vectors generated from image analysis and fractal features are preferably combined to form a single feature vector to be used in comparison.
  • image analysis feature vectors from test images may be compared to image analysis feature vectors of stored images, fractal feature vectors from test images may be compared to fractal feature vectors of stored images, and the scores from the two comparisons are preferably combined.
  • feature vectors may be classified as accepted or rejected based on user-defined criteria, with a final decision being made on the various decisions. Any of the extracted fractal images may employ a lookup table, locally or remotely online, to determine or otherwise classify the extracted features or otherwise stored images
  • information selection - using some but not all of the gathered information - may be implemented.
  • Information may be selected using bootstrapping, boosting, machine learning, or other processes.
  • this self-similarity may be leveraged by implementing any of the previously described algorithms or techniques at multiple resolutions, i.e., capturing an image at multiple resolutions, filtering an image, resizing an image, etc.
  • the information gathered at each resolution level may be compared to detect the level of self-similarity This information may be used for both identification and verification purposes. It is these multiple levels of resolution capture that may provide additional robustness in the image capture process.
  • the algorithm may preferably be designed in a cascaded manner, including the ability to first check low resolution. If there is a recognition failure (i.e.
  • processing may stop. If a wrong identification cannot be confirmed, or if more information is necessary, a higher resolution level may then be considered, and so on. Further, such multi-resolution is robust to occlusion. Any information missing at a lower resolution may be available at a higher resolution because of self-similarity.
  • a major issue in computer vision is the environment in which image capture is taking place, and how any environmental factors affect image quality.
  • the quality of an image can be influenced by any number of environmental factors. These environmental factors may include lighting, camera resolution, printer quality, capturing methods, camera angle, distance to camera, etc. Any previously described algorithm or technique or the information gathered from those algorithms or techniques may be adjusted to account for environmental factors. This adjustment may include filtering images before analysis, adjusting the algorithms or techniques, weighting of information, etc.
  • Image quality may be estimated both at a global level and a local level.
  • [0123] 2 isolate regions with low quality and analyze regions with good quality. Because of the self similarity and repeatability of fractal images, it is possible to determine portions of an image that provide a higher resolution of information, and process these portions of the image. Thus, in accordance with various embodiments of the invention, it is possible to identify one or more desirable portions of an image related to image quality, and process these portions of an image. Furthermore, selective image portion processing may be employed not only for image resolution, but also certain portions may be provided with different colors, allowing for color filters or other color selection tools to allow for processing of the different portions of the image by different devices, and have these portions isolated and separately processed.
  • portion (a) and (e) display two different fractals, -lj and A When the 5 s of A ⁇ are applied to both fractals, A ⁇ remains unchanged, but A ⁇ is clearly different.
  • portion (b) comprises a single transformation Si applied to A
  • portion (c) comprises a second transformation S 2 applied to A
  • portion (d) comprises the union of portions (b) and (c).
  • portions (f), (g), and (h) comprise the application of the same transformations, but to a different fractal.
  • this process can be implemented to form an identification system for fractal images based on transforming them by the contraction mappings of an IFS.
  • a query image q may be tested against a set R of reference images of fractals by transforming q by the ⁇ 3 ⁇ 4's of each r 6 R.
  • the change from q to S (q) may be measured for each r (any measurement may be user-defined, though the Euclidean norm is the standard method), and the reference image that resulted in the minimum change may be selected as the closest match.
  • a user-defined threshold t for the distance measurement may be implemented, so that if min(c3 ⁇ 4) > t, q is rejected as not being in R.
  • the deterministic algorithm for displaying fractals described above combined with the ability of laser marking technology to print highly detailed images at extremely small sizes, provides an inventive method for printing information at various scales within a fractal in accordance with various embodiments of the invention.
  • Figure 14 depicts the deterministic algorithm for the Sierpinski triangle, starting with an initial image including numbers. The algorithm is clearly drawing the same attractor from Figure 10, while employing the numbers rather than the unfilled box as in Figure 10.
  • Figure 14 shows five iterations after (portions (b) - (f)) after the initial image (a).
  • Figure 15 comprises a zoomed in portion of the imaged shown in portion (f) of Figure 14.
  • the attractor retains the details of the starting image, even at a zoomed in, higher resolution portion thereof.
  • the printed fractal will appear to be a normal fractal, but upon magnification with a microscope, the initial image can be recovered.
  • a combination of ink jet printing and laser etching may also be employed.
  • the self-similarity of fractals that satisfy the open-set condition ensures that multiple copies of the initial image will be clearly visible, providing robustness against occlusion or partial removal of the fractal.
  • the method may be implemented in the following manner in accordance with an embodiment of the invention:
  • the image can be arbitrary - it can contain an alphanumeric sequence, an image of a person's face, corporate logos, etc.
  • the deterministic algorithm ensures that regardless of the initial image, the fractal will have the same shape. This initial image will be repeated a multiple of times in the final fractal.
  • the open-set condition ensures that the contraction maps do not overlap.
  • the fractal may be chosen arbitrarily or selected by the user.
  • Print the fractal onto an object can be performed via laser marking or any other method capable of meeting the user's resolution requirements.
  • the complexity of fractals can be utilized in serialization - printing the same fractal but in different manners, allowing the fractal to carry information.
  • This inventive method takes advantage of the fact that fractals defined by an IFS are the attractors of a discrete time stochastic dynamical system defined by the contraction mappings of the IFS. This can be seen in the random iteration algorithm, where an initial point is chosen and a sequence of points is generated. In the language of dynamical systems, these sequences are referred to as trajectories, or orbits. Since the fractal is an attractor of the dynamical system, any orbit that starts on the fractal will stay on the fractal. Hence, individual sequences of points can be marked on the fractal, and used for serialization.
  • the orbits themselves can carry information. This is most easily seen with IFS's consisting of 2 contraction mappings, but can be applied to IFS's with any number of mappings. With 2 map IFS's, the mappings can be labeled So and Si.
  • An orbit may be generated by function composition, i.e. a k - S k (a 0 ) - S j S j ⁇ - ⁇ ⁇ (3 ⁇ 4 (a 0 ) ) ) , where jk £ ⁇ 0, 1 ⁇ , k £ 1, 2, . . ..
  • an orbit can be denoted by the sequence jkjk-i ⁇ ji- Since the jk's are either 0 or 1, there is a 1-1 correspondence between binary numbers and orbits of an IFS.
  • sequences of ASCII characters can be marked on a fractal by the orbit generated by their binary representation. This is illustrated in Figure 16, depicting encoding of the word "fractal" onto a fractal using the contraction mappings of the IFS that generated the fractal.
  • step 4 Mark the points of the orbit generated in step 4 on the fractal itself.
  • the points can be marked via special characters, color, or other conspicuous markings.
  • step 5 Compare the results of step 4. Using a user-defined metric, the orbit point closest to a transformed point is selected. The number corresponding to the contraction mapping which produced the closest transformed point is recorded.
  • additional 1710c information is superimposed on the previously provided information without replacing this information.
  • this particular embodiment of the invention provides component parts of the image at various levels of resolution.
  • the length of each of the individual sides of the image, and the various angles therebetween may be varied.
  • a similar triangular fractal image can be provided with additional complexity by removing a portion of the initial image 1810a (see the missing portion thereof), and by rotating or otherwise changing the position of the initial image when reproduced.
  • second level portions of the image 1810b are similar to portions 1810a, but changed in orientation.
  • a similar change in orientation is yet again shown by portions of the image 1810c.
  • image progression 1010 may be presented, and comprises modification of absolute position and orientation (1030) of one or more of the line segments in each iteration to provide still further complexity in the image. This complexity also comprises unpredictable information that may aid in providing additional robustness to the system.
  • a plurality of image cells 1110 may be provided. These image cells 1110 may comprise any desirable information therein. These plurality of image cells may be combined to comprise an image cluster 1120. Thus, the individual image cells may be provided structure relative to other image cells. Any desirable relationship between these cells may be provided. Additionally, each of these image clusters 1120 may be further combined to comprise a larger image cluster 1130, and may be placed within a single image cell 1110 of a larger cell, as desired. This process may be continued to as many levels as desirable, and thus provides a method of building up an image from a smallest desirable building block, rather than the methods describe above in which a largest image portion is provided and subdivided to provide additional complexity to the image.
  • Figure 21 depicts a still further embodiment of the invention in which a complex image 1210a is provided at a predetermined position and size. Additional predetermined sized versions of this image 1210b may be provided at a predetermine positions relative to the initial image 1210a. This information therefore may be provided across an extended area, and provided at predetermined sizes and positions to encode various information. Other additional configurations and sizes may be further provided.
  • fractal or fractal like images may be employed to allow for improved identification and authentication systems and methods.
  • Such systems and methods may be applied to any desirable object, and may preferably be applied to a medication pill to allow for identification and authentication of medication down to the individual pill level, thus improving robustness of the identification and authentication system against counterfeit or other identification errors.
  • an inventive encoding process may be implemented in the following manner, although additional encoding processes may also be employed:
  • Step 1 involves ASCII to binary conversion.
  • the binary representation of the string "L-system”: [0142]
  • the string "L-system” contains characters, so the binary representation contains digits. Any fractal curve (or other mathematical system) with at least line segments can be used to encode the string.
  • fractal curve or other mathematical system
  • other non-binary encoding systems may be employed, so that each character may comprise one or two or more states.
  • Step 2 involves preprocessing of the binary string generated from step 1.
  • Error correction methods such as Reed-Solomon can be implemented, along with redundancy features such as string duplication or interleaving. This step may also be skipped if desired, in accordance with various implementations of the invention.
  • Step 3 involves determining the parameters based on the specific application.
  • the size of the preprocessed binary string from step 2 is compared to parameters including printing area, size of the fractal curve, and number of line segments needed to encode the binary string.
  • An appropriate fractal curve with the needed number of iterations is selected based on the parameters. Any additional factors may also be employed in order to determine a desired fractal curve (or other mathematical system), as long as the basic elements set forth herein are met.
  • Step 4 involves the printing of the encoded fractal curve based on the information gathered from step 3.
  • Figures 25 and 26 illustrate this process.
  • Figure 25 depicts encoding of a fractal curve in which (a) is the 2nd iteration of a fractal curve created with a single line segment as the generator and the initiator from figure 2. It contains line segments, (b) is the same curve encoded with the string "L-system”; 's are represented by solid segments, 's by dashed line segments.
  • Figure 26 also depicts encoding of a fractal curve in which (a) is the 2nd iteration of the fractal curve from Figure 23. It contains line segments, (b) is the same curve encoded with the string "fractal barcoding technology"; 's are represented by solid segments, 's by dashed line segments. Note the solid region on the curve: this represents a synchronization sequence.
  • An inventive decoding process for such a generated fractal barcode first determines a type of barcode employed. The determination preferably depends on whether the barcode is assumed to be generated in accordance with a template or is accepted as a free form curve embedded in an image. If the decoder expects a fractal barcode in a template, then the decoder will preferably first extract the fractal barcode from the background, rotate and align it, and then check line segments at expected locations, in effect comparing the portions of the fractal image to the template.
  • the decoder will preferably determine a starting location on the curve, determine where the next line segment is located, iterate through each sequential line segment, and extract information based on the characteristics of the line segments. Therefore, in accordance with an embodiment of the invention, the decoding process for a templated fractal curve can be implemented in the following manner:
  • Step 1 involves accepting an image of the fractal barcode and determining the corner points of the template.
  • the fractal barcode may be placed in a square region with the corner points clearly delineated for easy recognition.
  • other method for aligning such a fractal barcode may be employed.
  • Step2 involves placing the fractal barcode in proper alignment for line segment extraction. Since the fractal barcode is preferably extracted from real-world images, it may be assumed that the barcode region is not perfectly aligned. Based on the relative location of the corner points, the image is preferably properly rotated and aligned into the desired region using a perspective transform, which maps quadrilaterals to quadrilaterals. Other methods of such additional alignment may also be employed.
  • Step 3 involves iterating through the sequence of line segments at their expected locations.
  • Line segment detection may be performed using the Hough transform or other appropriate method, and binary digits are preferably assigned on whether a line segment has been detected or not at appropriate locations.
  • the output of iterating through all the line segment locations is a binary string.
  • Step 4 involves the decoding of the binary string extracted from step 3.
  • the decoding process for a free form fractal curve can be implemented in the following manner:
  • Step 1 involves accepting an image of the fractal barcode. No requirements are preferably made on positioning, rotation, alignment, etc.
  • the fractal curve is extracted from the background using standard computer vision techniques (thresholding, Canny edge detection, etc.).
  • Step 2 involves using the Hough transform (or other appropriate mathematical system or the like) to detect line segments. The line segments are ordered sequentially, so the decoder will preferably iterate through the fractal curve, extracting the location of each line string one after another.
  • Step 3 involves checking the status of each detected line segment.
  • status is preferably the line color or type, but any type of line segment differentiator may be employed.
  • a binary digit will be assigned to each line segment based on the line type, creating a binary string.
  • Step 4 involves the decoding of the binary string extracted from step 3. If the fractal curve is closed (as in figure 5), then the binary string will be synchronized to recover the proper starting position and the resulting binary string decoded. Thus, after traversing the entire fractal curve, the position of the fractal curve may be shifted until a starting point of the fractal curve is properly aligned, thus allowing for extraction of proper codes. This may include sequentially applying one or more rules to such a determined binary string to further determine if such a sequence follows one or more rules for decoding. Other methods may be employed for extracting the information from the fractal curve.
  • encoding of the one or more line segments may include any number of differentiable formats.
  • encoding a system employing an English alphabet may be realized using 26 different types of line segments in the image. Any other encoding system may also be embodied.
  • Step 1 comprises initialization of the line segment where information will be stored. The user can select any format they choose for encoding the line segment.
  • Step 2 comprises selection of an L-system fractal.
  • the fractal can be chosen arbitrarily or selected by the user.
  • Step 3 comprises parameter determination. Parameters include size of area where the fractal will be printed, number of iterations for generation of the fractal, limitations of printing technology that will be used in step 4, and limitations of scanning equipment for information retrieval.
  • Step 4 comprises the printing of the resulting fractal pattern from the parameters derived in step 3.
  • Such a fractal image may be applied as a security labeling system to any item that is created in batches and may be varied in shape, such as medication pills in the manner as describe above.
  • the fractal image may be applied as a game on candy as a replacement to a "scratch and win" system, thus creating a show and win application, requiring the showing of the candy with the fractal image thereon being shown to an imaging device such as a web cam on a computer, mobile device or the like.
  • Such a fractal image may be applied to handbags or other fabric/clothing on the inside of a garment or label (as difficult to replicate directly onto a 3d texture / surface).
  • Such a fractal image system may also be employed with an identification systems employing facial recognition, or other biometric identification system.
  • identification of a patient or other user may be made employing one or more known identification systems, such as those noted above or others. Pill identification may them be performed, and a confirmation that the particular identified user is to take the identified pill.
  • the system may be able to determine whether the particular pill being imaged is being taken at the right time by the correct person.
  • a link between the pill and user may be established and confirmed. Release of such personalized information may therefore be predicated on proper biometric or other identification.
  • inventive fractal identification systems may be employed to provide an audit trail for a pill or object manufacturer.
  • inventive fractal recognition system upon use of the inventive fractal recognition system by a consumer, seller, or other individual, it may be possible to log results and alert the manufacturer if imaged fractal identifiers show low confidence (based on fractal integration with object - shape, color, texture, curvature), thus perhaps indicating an intent to replicate or otherwise provide a counterfeit product.
  • imaged fractal identifiers show low confidence (based on fractal integration with object - shape, color, texture, curvature), thus perhaps indicating an intent to replicate or otherwise provide a counterfeit product.
  • Such information, along with location data may be provided to authorities or other systems for tracking such counterfeiting, and in order to determine or identify counterfeit drug distribution points, or other areas with such high counterfeit drugs. Consumers may be provided with an incentive to check the identity of such fractal images, thus increasing availability of such widely spread identification information.
  • inventive fractal imaging system benefits of employing the inventive fractal imaging system are myriad.
  • the use of an ink jet printing process is easily available, and relatively inexpensive while remaining flexible.
  • Other printing or etching processes such as one or more of those noted above, may be employed when other combinations of cost and security are to be considered, or when mass printing is to be employed.
  • Mass batch and continuous processing avoids costs associated with properly printing images on individual pills or other objects.
  • the fractal images may be quickly modified, by including changes to the parameters for generating the fractal images, thus being indicative of various coding included in the fractal image.
  • a fractal library database that changes over time may be employed, in the manner noted above, so that changes over time may be documented and later confirmed.
  • Such a fractal database may also be tailored to webcam resolutions commonly used in smartphones, thus providing fractal images with resolution acceptable and able to be imaged by standard webcams in smartphones. Since the ID can be confirmed via a consumer with a mobile device and smart phone, it means that no special scanning hardware is needed. Thus, manufacturers or others may educate consumers to image and verify authenticity of an image, such as through imaging using a webcam on a mobile device or the like.
  • Confidence levels will allow for different security options. Thus different levels of resolution may be employed based upon a level of desired security. Higher resolution images, requiring higher resolution imaging devices may be employed for higher security applications, while lower resolution imaging and printing may be employed for lower security options. If the inkjet is a low quality printer, then distribution at higher resolutions (as measured by the camera) may suggest that fraud was committed if the ink distribution (or one or more other printing attributes) is different to what is expected.
  • inkjet or other etching or printing technologies may allow for rapid changing of patterns, one can quickly update a fractal pattern to be printed, and link the pattern to specific dates of production and/or batch numbers. This may be especially useful for perishables or medications that go out of date. Also may be useful for fashion items where dates of production are important. Use of older fractal patterns on newer objects may suggest counterfeiting, for example. Inks that fade over time may be employed, thus indicating passage of substantial time, or the like. Alternatively, inks that may be wiped off or otherwise removed may be employed to allow for maintenance in the integrity of the item, while still providing desired levels of security.
  • the inventive technology allows pills and objects of different shapes, curvature, and sizes to be labeled uniquely.
  • the unique shape, color and texture of the item allows for a unique ID to be printed and can help to differentiate close but not identical items.
  • the shape of the pill or item may have an influence on the way the fractal is printed allowing for expected distortions in the pattern to be recognized. This robust system may be particularly useful in identifying counterfeit items that may have slightly different shapes.
  • one or more fractal patterns may have a number of different colors built in that act as a calibration code.
  • pills or other objects appear to be slightly different in color when compared to the fractal - pill combination (a known color gradient and range of the fractal and ratio with the pill), then a warning sign can be issued.
  • comparison of the color of the pill or other object to the fractal, or consideration of an effect the color of the pill or other object may have on the color of the fractal may aid in the determination of status of the pill or other object.
  • the fractal may be distorted in a predictable, measurable manner in accordance with this known object surface texture. This allows for texture to be identified as well of a specific surface and differences flagged.
  • Fractal identification labeling is far superior to existing imaging labels because it does not require the whole fractal necessarily to be printed on the medication or other object.
  • Use of the inventive printed fractal image will allow for occlusion by the user (due to fingers blocking image or poor environmental conditions) as only a portion of the image may be required to reach desired confidence levels.
  • Such a fractal identification label does not suffer from occlusion problems (e.g. traditionally, one number hidden or obscured may inhibit use of the ID system).
  • Computer vision may therefore be used to learn and to identify replicating blurred images in the fractal and/or expected ratios of patterns, colors or shapes.
  • fractal identification may operate better than a number, or more traditional barcode, as it is more difficult to replicate.
  • the problem with present day technology is that it does not label the pill, or if it does label the pill the information is inaccessible to the public or can be easily copied or destroyed.
  • the present invention is novel because it concentrates solely on the printing of the pill and the organic nature of the fractal patterns themselves, and the complexity embedded within them, to make the system extremely difficult to replicate.
  • the fractal patterns may blanket the entire surface of the pill. Any random segment of the pill, no matter how small or how large, will be able to be used to identify it. This is a tremendous improvement over barcodes , which are static fixed-form labels that do not lend themselves to different resolutions, and are also easily copied, damaged, and do not adapt to the physical configuration of a pill. Progressively higher levels of fractal resolution will also allow for progressively higher levels of security authentication. It is the first solution that addresses high security pill identification needs with public accessibility.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Printing Methods (AREA)

Abstract

La présente invention concerne un procédé et système pour l'application et la lecture d'une image fractale vers et depuis une pluralité d'objets pour agir comme une étiquette d'identification. Le système comporte une imprimante pour l'impression d'un motif fractal sur une pluralité d'objets et un lecteur pour la lecture du motif fractal imprimé. Une telle image fractale est robuste vis-à-vis des difficultés et incohérences d'impression et d'imagerie, et est difficile à copier, permettant ainsi une protection contre la contrefaçon.
PCT/US2012/059139 2011-10-06 2012-10-05 Procédé et appareil pour l'identification par analyse fractale WO2013052924A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
EP12838300.7A EP2761536A4 (fr) 2011-10-06 2012-10-05 Procédé et appareil pour l'identification par analyse fractale

Applications Claiming Priority (10)

Application Number Priority Date Filing Date Title
US201161544056P 2011-10-06 2011-10-06
US61/544,056 2011-10-06
US13/338,602 2011-12-28
US13/338,602 US8720790B2 (en) 2011-10-06 2011-12-28 Method and apparatus for fractal identification
US201261623193P 2012-04-12 2012-04-12
US201261623148P 2012-04-12 2012-04-12
US61/623,193 2012-04-12
US61/623,148 2012-04-12
US201261683849P 2012-08-16 2012-08-16
US61/683,849 2012-08-16

Publications (1)

Publication Number Publication Date
WO2013052924A1 true WO2013052924A1 (fr) 2013-04-11

Family

ID=51019554

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2012/059139 WO2013052924A1 (fr) 2011-10-06 2012-10-05 Procédé et appareil pour l'identification par analyse fractale

Country Status (2)

Country Link
EP (1) EP2761536A4 (fr)
WO (1) WO2013052924A1 (fr)

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2015160797A1 (fr) * 2014-04-15 2015-10-22 Apdn (B.V.I.) Inc. Chiffrement et déchiffrement de contour
US9904734B2 (en) 2013-10-07 2018-02-27 Apdn (B.V.I.) Inc. Multimode image and spectral reader
US9963740B2 (en) 2013-03-07 2018-05-08 APDN (B.V.I.), Inc. Method and device for marking articles
US10047282B2 (en) 2014-03-18 2018-08-14 Apdn (B.V.I.) Inc. Encrypted optical markers for security applications
US10519605B2 (en) 2016-04-11 2019-12-31 APDN (B.V.I.), Inc. Method of marking cellulosic products
US10741034B2 (en) 2006-05-19 2020-08-11 Apdn (B.V.I.) Inc. Security system and method of marking an inventory item and/or person in the vicinity
US10745825B2 (en) 2014-03-18 2020-08-18 Apdn (B.V.I.) Inc. Encrypted optical markers for security applications
US10920274B2 (en) 2017-02-21 2021-02-16 Apdn (B.V.I.) Inc. Nucleic acid coated submicron particles for authentication
US10995371B2 (en) 2016-10-13 2021-05-04 Apdn (B.V.I.) Inc. Composition and method of DNA marking elastomeric material
CN112927131A (zh) * 2021-01-16 2021-06-08 中建三局第一建设工程有限责任公司 图片拼接方法、装置、计算机设备和存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5065447A (en) * 1989-07-05 1991-11-12 Iterated Systems, Inc. Method and apparatus for processing digital data
US20070194034A1 (en) * 2006-02-17 2007-08-23 Vasilios Vasiadis Device for printing pills, tablets or caplets in a precise manner
US7304228B2 (en) * 2003-11-10 2007-12-04 Iowa State University Research Foundation, Inc. Creating realtime data-driven music using context sensitive grammars and fractal algorithms
US20080138604A1 (en) * 2006-05-02 2008-06-12 John Kenney Authenticating and identifying objects using markings formed with correlated random patterns
US20080290168A1 (en) * 1996-05-31 2008-11-27 Scott Lindsay Sullivan Pill printing identification

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5894801A (en) * 1997-09-30 1999-04-20 Ackleey Machine Corporation Methods and systems for sensing and rectifying pellet shaped articles for subsequent processing

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5065447A (en) * 1989-07-05 1991-11-12 Iterated Systems, Inc. Method and apparatus for processing digital data
US20080290168A1 (en) * 1996-05-31 2008-11-27 Scott Lindsay Sullivan Pill printing identification
US7304228B2 (en) * 2003-11-10 2007-12-04 Iowa State University Research Foundation, Inc. Creating realtime data-driven music using context sensitive grammars and fractal algorithms
US20070194034A1 (en) * 2006-02-17 2007-08-23 Vasilios Vasiadis Device for printing pills, tablets or caplets in a precise manner
US20080138604A1 (en) * 2006-05-02 2008-06-12 John Kenney Authenticating and identifying objects using markings formed with correlated random patterns

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP2761536A4 *

Cited By (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10741034B2 (en) 2006-05-19 2020-08-11 Apdn (B.V.I.) Inc. Security system and method of marking an inventory item and/or person in the vicinity
US9963740B2 (en) 2013-03-07 2018-05-08 APDN (B.V.I.), Inc. Method and device for marking articles
US10282480B2 (en) 2013-10-07 2019-05-07 Apdn (B.V.I) Multimode image and spectral reader
US9904734B2 (en) 2013-10-07 2018-02-27 Apdn (B.V.I.) Inc. Multimode image and spectral reader
US10047282B2 (en) 2014-03-18 2018-08-14 Apdn (B.V.I.) Inc. Encrypted optical markers for security applications
US10745825B2 (en) 2014-03-18 2020-08-18 Apdn (B.V.I.) Inc. Encrypted optical markers for security applications
EP3132422A4 (fr) * 2014-04-15 2017-12-06 APDN (B.V.I.) Inc. Chiffrement et déchiffrement de contour
WO2015160797A1 (fr) * 2014-04-15 2015-10-22 Apdn (B.V.I.) Inc. Chiffrement et déchiffrement de contour
US9412284B2 (en) 2014-04-15 2016-08-09 Apdn (B.V.I.) Inc. Contour encryption and decryption
US10519605B2 (en) 2016-04-11 2019-12-31 APDN (B.V.I.), Inc. Method of marking cellulosic products
US10995371B2 (en) 2016-10-13 2021-05-04 Apdn (B.V.I.) Inc. Composition and method of DNA marking elastomeric material
US10920274B2 (en) 2017-02-21 2021-02-16 Apdn (B.V.I.) Inc. Nucleic acid coated submicron particles for authentication
CN112927131A (zh) * 2021-01-16 2021-06-08 中建三局第一建设工程有限责任公司 图片拼接方法、装置、计算机设备和存储介质

Also Published As

Publication number Publication date
EP2761536A4 (fr) 2016-04-13
EP2761536A1 (fr) 2014-08-06

Similar Documents

Publication Publication Date Title
US9569650B2 (en) Method and apparatus for fractal identification of an object
US9400909B2 (en) Method and apparatus for identification
EP2761536A1 (fr) Procédé et appareil pour l'identification par analyse fractale
US9053364B2 (en) Product, image, or document authentication, verification, and item identification
US8712163B1 (en) Pill identification and counterfeit detection method
US7757952B2 (en) Method and apparatus for counterfeiting protection
US20170109600A1 (en) Method for object recognition and/or verification on portable devices
Van Beusekom et al. Text-line examination for document forgery detection
CN109472337B (zh) 一种基于随机特征的标签防伪方法和装置
US10929639B2 (en) Method and system for encoding fingerprint of things
CN110073368A (zh) 用于认证插图的方法
CN104537544A (zh) 带覆盖层的基于背景纹理特征提取的商品二维码防伪方法及系统
Cozzella et al. Drug packaging security by means of white-light speckle
US20200065829A1 (en) Commodity anti-counterfeit verification system based on natural biological information
JP2013534683A (ja) 消費者による製品のマーキング及び認証のための方法及び装置
US20160321531A1 (en) Composite information bearing devices
US20200250422A1 (en) Method of authentication using surface paper texture
US7766245B2 (en) Virtual code window
EP2748754B1 (fr) Système et procédé d'authentification judiciaire
US9361562B1 (en) Method and apparatus for fractal multilayered medication identification, authentication and adherence monitoring
US20070253593A1 (en) Methods for making an authenticating system
US10977534B1 (en) Compositions, systems, and methods for unique identifiers embedded in machine verifiable marks having error correction features
Pollard et al. Model based print signature profile extraction for forensic analysis of individual text glyphs
US20190163944A1 (en) Composite information bearing devices
US10380391B1 (en) Method and apparatus for fractal identification

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 12838300

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

REEP Request for entry into the european phase

Ref document number: 2012838300

Country of ref document: EP

WWE Wipo information: entry into national phase

Ref document number: 2012838300

Country of ref document: EP