DataGlyph

Overview

DataGlyph is a 2D data-encoding scheme developed at Xerox PARC that represents bits as tiny 45-degree diagonal strokes — a stroke leaning one way is a 0, leaning the other way is a 1 ^{1, 2}. Printed densely, the strokes blend into what looks like a uniform grey halftone tint, so a DataGlyph can hide in the background of a document, masquerade as a shaded image, or form a decorative border while invisibly carrying machine-readable data ¹. The binary payload is protected with Reed–Solomon error correction, giving DataGlyphs robustness against print damage and partial occlusion ¹.

History

The core idea was conceived by Rob Tow in 1988 at Xerox PARC's Electronic Document Lab ². Tow's insight was to repurpose the halftone dots used in digital printing as data carriers: "you could make the little dot lean left to represent a zero, and to the right to represent a one" ². He chose the name "Glyph" for its typographic associations, and Xerox marketing later branded the result DataGlyph ².

The technology was patented through the 1990s ²:

U.S. 5,091,966 (Feb 1992) — Bloomberg & Tow — adaptive scaling for glyph decoding ².
U.S. 5,315,098 (May 1994) — Tow — embedding digital data in halftone images ².
U.S. 6,076,738 (Jun 2000) — Bloomberg, Hecht, Flores & Tow — self-clocking glyph shape codes ².

DataGlyph became a licensed Xerox product and was incorporated into Xerox's PaperWorks system, which encoded document metadata directly onto printed pages ². (David L. Hecht at PARC is associated with the later self-clocking glyph-code work.)

Technical specification

DataGlyph encodes bits in the orientation of small diagonal marks ^{1, 2}:

Dibit marks — each glyph is a short line at 45°; a downward slope = 0, an upward slope = 1 ¹.
Halftone camouflage — printed densely, the field of glyphs reads as a continuous grey area, "akin to halftone," so it can be tuned to resemble an image, sit behind text, or act as a page border ¹.
Reed–Solomon error correction — the bit stream is encoded with Reed–Solomon, giving substantial redundancy (a reported redundancy on the order of ~30 %) so the code survives smudging, folds, and partial damage ¹.
Self-clocking glyph codes — later variants use a self-clocking structure so the reader can recover the grid lattice and synchronise without a separate timing track ².
Density — high for a printed code: figures of around 155 bytes per cm² at 600 dpi are cited, and digital cards using DataGlyphs reportedly stored on the order of 3 kilobytes — far above a magnetic stripe ¹.

Use cases

Embedded document metadata — Xerox PaperWorks used DataGlyphs to carry routing, indexing, and form metadata invisibly on the printed page ².
Secure / value documents — the halftone camouflage plus Reed–Solomon robustness suited ID cards, tickets, and certificates where data should be present but not obtrusive ¹.
Self-describing forms — a glyph block printed on a form lets a scanner recover machine-readable instructions about that form ^{1, 2}.
Decorative / steganographic data carrying — glyphs tuned to look like shading or borders embed data without a visible barcode ¹.

Implementations

DataGlyph is proprietary to Xerox; there is no open standard or widely available open-source implementation ^{1, 2}:

Xerox DataGlyph toolkit / PaperWorks — the original licensed encoding/decoding software ².
Research decoders — academic work exists on reading DataGlyphs from camera images (e.g. grayscale-reader research), but these are studies rather than productised libraries ¹.

State honestly: DataGlyph never developed the broad open-source tooling ecosystem that QR or Data Matrix enjoy; it remained largely a Xerox-controlled technology.

Comparison

vs. QR Code / Data Matrix — QR and Data Matrix are overt, standardised (ISO), royalty-free, and ubiquitously supported by free libraries. DataGlyph is covert (it hides as halftone shading), proprietary, and lacks open tooling. Where QR/DM advertise their presence with finder patterns, DataGlyph's design goal is the opposite — to be visually unobtrusive while still carrying meaningful data.

vs. digital watermarking (e.g. Digimarc Barcode) — both aim for unobtrusive on-page data. A digital watermark is truly imperceptible (modulating existing imagery), whereas a DataGlyph is a visible-but-camouflaged texture of strokes; DataGlyph carries comparatively high capacity and structured error correction at the cost of being a discrete glyph field rather than an invisible signal spread through artwork.

vs. 1D barcodes — DataGlyph offers far higher density and Reed–Solomon robustness, plus the ability to blend into a document, but at the price of specialised, non-open readers.

Fun facts

The whole scheme grew from one elegant observation by Rob Tow in 1988: the halftone dots already sprayed across every printed image could simply be made to lean left or right to carry bits, turning ordinary printed shading into a hidden data layer ².

DataGlyphs can be tuned to look like part of an image, the background behind text, or a decorative border — so a page can carry kilobytes of machine-readable data without anything that looks like a barcode appearing on it at all ¹.

Status

Legacy. DataGlyph was an influential Xerox PARC technology of the 1990s that shipped commercially (PaperWorks) but never achieved the open-standard, mass-library adoption of QR Code or Data Matrix ^{1, 2}. It is best regarded as a historically important, largely superseded proprietary scheme: its ideas (camouflaged high-density on-page data, self-clocking glyph lattices) live on conceptually, but it is not an actively growing standard today.