Copy Detection Pattern

Overview

A Copy Detection Pattern (CDP) is a small random (maximum-entropy) digital image printed on a document, label, or product so that copies can be told apart from originals ¹. It works on an information-loss principle: each print-and-scan cycle degrades fine detail, and because making a counterfeit requires an extra scan-and-print pass, a copied CDP carries measurably less information than a genuine first-generation print ¹. Authentication is performed by scanning the CDP with an image scanner or an ordinary smartphone camera and measuring how much detail survives ¹. CDPs are frequently embedded inside or alongside a QR Code so that a single scan both reads the product identifier and authenticates the print ¹.

History

Copy Detection Patterns were introduced by Justin Picard around 2004, notably in the paper "Digital authentication with copy-detection patterns" (SPIE, 2004) ¹. Earlier related patent work on visible authentication patterns for printed documents (J. Picard and J. Zhao) dates to about 2002 ². The underlying insight — that the print-scan channel is a lossy, hard-to-invert process and that copying compounds that loss — became the foundation for a class of anti-counterfeiting graphics ¹.

CDPs were deployed early in security documents, including identity-style credentials and event badges (e.g. 2006 FIFA World Cup badges), and later moved into pharmaceutical and product packaging ¹. The technology was commercialised most prominently by Scantrust, which prints a CDP inside a secure QR code so brands can verify authenticity with a phone ^{3, 1}.

Security research has continued to probe CDPs: from 2019 onward, studies (Taran, Yadav, and others) examined machine-learning "clonability" attacks that attempt to reconstruct the digital CDP from a scan and reprint a higher-fidelity copy, prompting work on stronger detection metrics and provable guarantees ^{1, 4}.

Technical specification

A CDP is engineered to be hard to reproduce and easy to verify ¹:

Maximum-entropy image — a small random or pseudo-random pattern (e.g. a dense field of black/white pixels) designed to pack the maximum amount of fine detail, so that any loss of detail is conspicuous ¹.
Information-loss detection — every print or scan loses information about the original digital image; a counterfeit needs an additional capture-and-reprint cycle, so it is further from the digital original and measurably lower in information content ¹.
Verification metric — the detector compares the scanned pattern against the known digital reference (or a derived statistic) and computes a similarity/correlation score; a genuine first-generation print scores high, a copy scores low ¹.
Public algorithm, secret reference — the detection algorithm can be public; security rests on keeping the digital CDP itself (or its generation key) secret, since an attacker who lacks the high-resolution digital original must work from a degraded scan ¹.
Smartphone-readable — modern CDPs are tuned so a consumer-grade phone camera can capture enough detail to authenticate, not just a flatbed scanner ^{1, 3}.
QR integration — high-entropy secure graphics are commonly embedded within a standard QR code (e.g. inside the finder/data zone), so one scan yields both the identifier and the authenticity verdict ^{1, 3}.

Use cases

Brand protection / anti-counterfeit — the core use: proving a label or package is a genuine first-generation print rather than a scanned-and-reprinted copy ^{1, 3}.
Pharmaceutical packaging — authenticating medicine packs against counterfeits ¹.
Secure documents and credentials — identity documents and access badges (e.g. 2006 FIFA World Cup badges) ¹.
Product authentication via IoT / track-and-trace — combined with serialized QR for both identity and authenticity in supply-chain systems ^{1, 3}.

Implementations

CDPs are mostly delivered through commercial platforms, with academic datasets available ^{1, 3}:

Scantrust — the leading commercial provider; prints a CDP inside a secure QR code and authenticates via a mobile app/SDK and cloud verification ³.
Academic datasets / tooling — research groups (and Scantrust-shared datasets) publish CDP image sets for studying detection and clonability, but these are research artifacts rather than a turnkey open standard ^{1, 4}.

State honestly: there is no single open CDP standard; generation, the secret digital reference, and the verification service are typically held by the vendor. The general principle is public, but production-grade CDP authentication is a proprietary/managed offering.

Comparison

vs. QR Code alone — a plain QR carries an identifier but is trivially photocopied; the symbol scans identically whether genuine or copied. A CDP embedded in or beside the QR adds a copy-resistance layer the QR lacks, so the same scan can both identify and authenticate ¹.

vs. holograms — holograms are overt, visually striking deterrents but are themselves widely counterfeited and require human inspection. A CDP is a flat printed graphic verified automatically by a camera + algorithm, giving an objective genuine/copy decision rather than relying on a person to judge a hologram's quality ³.

vs. serialization (unique serial numbers / track-and-trace) — serialization proves an item's identity and detects duplicates only if the back-end sees the same serial twice; it does not, by itself, prove the physical print is original. A CDP complements serialization by detecting that a given printed instance is a scanned copy even if the serial looks valid ^{1, 3}.

Fun facts

The security model is unusual: the detection algorithm can be completely public — secrecy lives only in the high-resolution digital original of the random pattern. A counterfeiter who can only photograph or scan a genuine label is already working from a degraded copy and cannot recover the full-detail digital source, which is exactly what the detector checks for ¹.

CDPs were used on 2006 FIFA World Cup badges, an early high-profile deployment of the print-and-scan information-loss idea in a real-world security-credential setting ¹.

Status

Active and commercially deployed, with an ongoing security arms race ^{1, 3}. CDPs are in production use for brand protection, pharma, and secure documents, most visibly through Scantrust's secure-QR product ³. Academic work continues on machine-learning attacks that try to clone CDPs and on stronger, provable detection guarantees, so the technology is evolving rather than static ^{1, 4}.