Joel Zlotnick, Jordan Brough and Troy Eberhardt

Commercial dot halftone patterns are often avoided in security design in favour of custom line artwork for two reasons. First, many counterfeiters use commercial off-the-shelf printing devices which employ similar microscopic dot halftone patterns to make the images. Secondly, the organic, non-repeating nature of security line artwork is not easily replicated using software-assisted techniques. However, as highly complex halftones generated by proprietary security algorithms can be hard for human counterfeiters to comprehend and replicate, complex security halftones are increasingly used in security designs. This article examines the evolving role of halftones in security printing, and how to obtain the greatest security value from complex halftone designs.

About halftones
A detailed discussion of halftone technologies is beyond the scope of this article, but some key points must be reviewed for context. Except for dye sublimation printers, most hardcopy printing devices are unable to vary the density of a pixel-like dot element on a sheet of paper the way a computer monitor can vary the brightness of a pixel. Instead, spacing or dot size is varied to simulate tones. This reality has important implications for how many counterfeits are created using commercial printing technologies, and how genuine security document artwork is designed.

For example, offset printing presses cannot vary the value of an ink by changing the thickness or density of the ink layer. Instead, an amplitude-modulated (AM, or raster) halftone process is typically used to translate a digital or continuous tone image into a pattern of dots with a fixed ink layer thickness and fixed spacing, but which vary in size. Darker areas consist primarily of larger dots, while lighter areas are made up of smaller dots (see Figure 1). Similarly, a typical inkjet printer cannot vary the density of the tiny droplets of ink it delivers, but can vary the spacing between the droplets. This is accomplished by using a frequency-modulated (FM, or stochastic) halftone process. FM halftones cluster more droplets in areas of higher density (dark areas), and fewer droplets in areas which are to appear lighter (see Figure 2). For purposes of counterfeit detection, the differences between AM and FM half­tones are less significant than their similarities: namely, most common commercial printing technologies can only simulate digital or continuous tone images using micro­scopic coloured dots. This is true for inkjet printers, laser printers, colour photocopiers and most process colour offset and flexographic printing workflows.

Figure 1: Amplitude-modulated dot halftone. Dot centres are consistently spaced, but dots change in size to control image density.
Figure 2: Frequency-modulated (stochastic) dot halftone. All dots are similarly sized, and image density is controlled by dot quality.

Because most counterfeits produced with commonly available office equipment contain patterns of micro­scopic cyan, magenta, yellow, and black (CMYK) coloured dots, most security designers intentionally create genuine document artwork with line art and spot colours that do not resemble dot halftones under magnification. For example, Figure 3 shows the line art background of a genuine US visa, and Figure 4 shows the simulated artwork produced by an inkjet printer. This analysis of microscopic print is a powerful technique because it facilitates the examination of unfamiliar documents. The security features of the visa in Figure 3 do not need to be known to easily determine that the visa shown in Figure 4 is digitally reproduced and is likely counterfeit.

Figure 3: Genuine US visa line art background pattern.
Figure 4: Counterfeit US visa showing inkjet print characteristics and a frequency-modulated halftone pattern.

The comparison in Figures 3 and 4 demonstrates why dot-based halftones are avoided in most genuine document artwork. While this design strategy has clear merit, it does not address detection of sophisticated counterfeits that are not printed using halftone tech­niques, or consider other potential benefits of using non-commercial security halftone technology to create highly complex security artwork. Sophisticated counterfeiters that possess expertise in graphic arts may attempt to completely replicate the line art images and print counterfeits using spot colours instead of halftone patterns. By introducing complex security halftones to documents which already contain line art­work, security designers have the ability to incorpo­rate complex patterns that can frustrate counterfeiters attempting to replicate document artwork. Below it is demonstrated that halftones can and do play a role in achieving the latter goal.

Strategies for halftones in security printing
A fundamental concept in counterfeit deterrence is to construct a secure document from materials and processes to which counterfeiters have limited access. This is true not just for dedicated security feature technologies that are frequently judged according to their lack of commercial availability, but also for the manufacturing processes and designs that strategically incorporate all of these components in a document together. Like other manufacturing processes used in security printing, halftones are neither inherently good nor bad. They can increase or decrease the security of the document depending on how they are used.

Following this basic concept, how are halftones adapted from their typical commercial applications for custom use in security work? The commercial AM and FM halftones in Figures 1 and 2 possess a key similarity: the basic halftone unit in each case is a dot. This suggests an obvious strategy that is already in wide use: adopt a basic artwork unit that is not a dot. Some examples include lines that are straight (see Figure 5) or wavy (see Figure 6), but other halftone shapes such as circles, triangles and hexagons can be used as well. The examples in Figures 5 and 6 are classified as AM halftones because the lines change in width (amplitude) instead of changing in spacing (frequency) to simulate darker and lighter areas.

Figure 5: Amplitude-modulated line halftone in a Georgian 200 lari banknote. Lines in this pattern change in width, similar to how dots change in size in a dot halftone.
Figure 6: Amplitude-modulated line halftone in a Haitian 1,000 gourdes banknote. The lines are wavy instead of straight, but still change thickness to control density.

As previously mentioned, dot-based AM halftones change size in two dimensions until they overlap in areas of complete ink coverage, but retain the same basic shape with no change in frequency. In contrast, consider a halftone element which may be one of many different shapes depending upon the tonal value required at that particular location, each corresponding to a different discrete density value (for example, square = 10%, triangle = 20%, oval = 30%, etc.). The halftone element shown in Figure 7 looks like four solid rectangles at its maximum density. Shapes of lesser density are composed of various parts of the rectangle (arranged in four columns) based on the required density. It is hard to determine how many different halftone shapes are in this image and how those shapes correspond to density; this confusion is precisely the goal of using complex patterns. Similarly, Figure 8 shows a related example where the space occupied by each halftone element is a square of fixed size, but different densities are represented by different patterns of lines. Figure 9 demonstrates how each of the shapes corresponds to a different tonal value, and how each shape was randomly rotated to generate more complexity. This example could be customised further by adding a greater quantity of line patterns for each density level.

Figure 7: Security halftone pattern in a Swiss 1,000 franc banknote. The shape of the halftone unit changes at different levels of density.
Figure 8: Security halftone pattern. Each level of density is represented by a different shape, with each shape also randomly rotated in the space allocated for greater complexity.

In the next example, consider an AM halftone that contains several different halftone elements but does not allocate fixed space to each element. The halftone shown in Figure 10 contains several shapes in the magnified area, including spirals, circles, pentagons, dots and text. Unlike the examples in Figures 7 and 8 which constrain the halftone unit to a fixed space, the shapes in Figure 10 are placed in an arbitrary pattern. The spiral shape can be found both within and separate from the circle, the pentagon may contain either one or two dots (in different positions), and some of the spirals contain a dot, while others do not. Even more importantly, the example in Figure 10 is an AM halftone because lighter and darker areas of the larger image are created by varying line width (amplitude) in the microscopic shapes rather than varying the distance (frequency) between the shapes. The line width variation occurs in a slightly different way for different shapes of the same type, depending on where they are placed in the microscopic halftone structure. As a result, each of the elements of the same type appears slightly different from one another because the changes in line weight occur in different positions within the elements. The resulting effect is that no single shape element is exactly the same as others of the same type across a document. This is a subtle but important point, because it means that it is not possible to copy one pentagon and paste it into position as a different pentagon in the document and achieve the same tonal value. Copy and paste strategies are of limited value in replicating this artwork, and the design resists regenera­tion using step-and-repeat or copy-and-paste strategies.

Figure 9: Chart of density values used in Figure 8. Each density value has its own distinct shape and each shape is randomly rotated to increase complexity.
Figure 10: Security halftone pattern in a Namibian 200 dollar banknote. This design does contain repeating shapes, but not repeating line thicknesses in those shapes.

Similar to Figure 10, the halftone elements shown in Figure 11 include images of stars, dots, text designs, diamond-shaped geometric patterns, and trees. As in Figure 10, these elements are placed next to each other arbitrarily. The tree images sometimes encompass other artwork elements, such as stars and dots, which change from tree to tree. What is different about Figure 11 is the rotation and modification of the design of some individual halftone elements. Figure 11 depicts two types of tree halftone elements, a smaller one leaning left and a larger one leaning right. At first glance, these tree elements seem identical except for rotation and size. However, on closer inspection, a shape difference can be found: some of the left-leaning trees have a notch placed at the bottom of the right side of the trunk. These differences are subtle but important.

Figure 11: Security halftone pattern in a Tunisian 50 dinar banknote. This example showcases shape rotation and minor variations to different iterations of the tree elements.
Figure 12: Security halftone pattern showing a transition between shapes. This pattern contains a large number of transitional shapes instead of a few discrete shapes.

In Figure 10, most halftone elements appear in the same orientation throughout the design (with the possible exception of the pentagons that contain two dots, which may be rotated). In Figure 11, a basic tree shape has been rotated and resized to make two distinct tree images, one of which contains a small notch in the trunk. This suggests more strategies that could make security halftone designs even more difficult to replicate. By including arbitrary rotation of the halftone element as a design strategy, can the complexity of the halftone pattern and the overall security of the document be increased? Further, is it possible to use several different versions of the same element, each subtly or obviously different from the others, or otherwise expand the quantity of shapes? This could be done either by a deliberate modification of each element (such as the trees in Figure 11), or by an automated process that applies varying warp effects to distort the original shapes. As a result, no two elements of the same type would ever be identical in shape, even before the halftone process modified the line widths. Such a strategy would use halftone techniques to generate what is ultimately a highly complex line art pattern that would be hard for an artist to generate manually, whether that artist is a security designer or a counterfeiter.

Finally, the prior examples show halftone patterns derived from a finite set of shapes. In contrast, consider an AM halftone generated from one pattern of shapes transitioning into another pattern of shapes. The design shown in Figure 12 is a simple example to demonstrate the concept: a pattern of triangles that transforms into a pattern of cubes across the width of the image. These patterns need not be made of such simple shapes, and it is possible to imagine more complex examples. Consider a security document containing a transition between the halftone shown in Figure 10 to the halftone shown in Figure 11, and the complexity that could be added by using more shapes, in more complex arrangements, than the simple triangle and cube shown in Figure 12.

The patterns shown in Figures 5 through 12 demonstrate that halftone technologies do play a significant role in security design. Though all avoid the use of dot patterns to keep the security artwork unlike the dot halftones that are typical in commercial printing, complex examples also use the halftone process to defeat those counterfeiters that attempt to rebuild artwork. Paradoxically, the intricacy of these examples exempli­fies how an automated halftone software process can facilitate production of line art of such complexity that it becomes difficult to replicate. The conclusion is that halftones, used appropriately and deliberately, can contribute significantly to counterfeit deterrence.

Conclusion
Security halftones clearly play an increasing role in modern security documents and can complement traditional line art techniques. Still, the reader may ask why it is important to prevent counterfeiters from replicating artwork, if alternative simpler pathways to counterfeiting remain available. The shortcut paths of scanning, digitally manipulating and printing artwork must also be denied to counterfeiters, and security halftones are only one component of an overall counterfeit deterrence design strategy. Like halftones, targeted use of colour and split fountain printing can work to prevent counterfeiters from reverse-engineering printing plate images, which complicates replication of document artwork and can drive counterfeiters toward process colour simulations. Then the question becomes how to make the process colour simulations appear as poor as possible, which can be facilitated with expanded colour gamut and split fountain techniques. The authors will consider these strategies in subsequent papers.

MORE ABOUT THE AUTHORS

Joel Zlotnick is employed by the US Department of State, Bureau of Consular Affairs, Counterfeit Deterrence Laboratory as a physical scientist. He conducts research on how design strategies can help maximise the security value of printing technologies and security features, and develops training programmes on counterfeit detection.

Jordan Brough is employed by the Homeland Security Investigations Forensic Laboratory as a forensic document examiner, specialising in adversarial analysis and counterfeit deterrence. Jordan spends his time examining suspect documents and consulting with United States security document designers.

Troy Eberhardt is employed by the US Immigration & Customs Enforcement Homeland Security Investigations Forensic Laboratory. He supervises the Research and Development Section at the laboratory, which specialises in identifying and mitigating vulnerabilities within travel documents.