2021 IEEE International Conference on Blockchain (Blockchain)
Photrace: A Blockchain-Based Traceability System
for Photographs on the Internet
Tatsuya Igarashi, Takabayashi Kazuhiko,
Yoshiyuki Kobayashi, Hiroshi Kuno
Sony Group Corporation, Japan
Email:{tatsuya.igarashi, kazuhiko.takabayashi,
yoshiyuki.ky.kobayashi, hiroshi.kuno}@sony.com
2021 IEEE International Conference on Blockchain (Blockchain) | 978-1-6654-1760-0/21/$31.00 ©2021 IEEE | DOI: 10.1109/BLOCKCHAIN53845.2021.00089
Abstract— This paper proposes a blockchain-based
traceability system of photos. Thanks to a novel scheme that
securely chains certificates of image data originating from a
trusted origin certificate, the system authenticates the
provenance of original and converted images, e.g., which digital
camera created it. The system requires a Public Key
Infrastructure (PKI) of digital cameras to authenticate the
trusted origin certificate signed by the certified digital camera.
It does not require a PKI of image conversion software. The
proposed solution is suitable for the traceability system of
photos on the Internet as the scheme of chaining certificates
applies to any image conversion software, including open
source-based photo editing applications and cloud server
software that converts image data of photos. We have
implemented a prototype of the traceability system using
Ethereum as a permissioned blockchain and evaluated the
system's feasibility.
Keywords—blockchain, photograph, image data, digital
camera, traceability, certificate, PKI, copyright, fake media, NFT.
I. INTRODUCTION
For the last decade, social networking services for photos
and videos have become popular due to the spread of
smartphones. The photo and video markets are also growing,
and recently the Non-Fungible Token (NFT) market is
booming. Unfortunately, infringement of copyright and
dis/misinformation on the Internet are unsolved problems.
Emerging AI technologies worsen the issue by facilitating
image alteration such as deep fakes.Meanwhile, blockchain
technology got much attention with the advent of Bitcoin [1].
The immutability of blockchain is suitable for applications
such as copyright management, fake media detection, or
verifying digital assets’ authenticity. For example, Binded
(formerly Blockai) [2] is one of the early high-profile
copyright management services that use the Bitcoin network.
Most NFT markets use blockchain, such as Ethereum [3].
ERC-721 [4] defines a standard for smart contracts to manage
the token ID, e.g., a cryptographic hash value, for digital assets
and to transfer the ownership of the NFT. However, anyone
can record a token ID on the blockchain on a first-come-first-
served basis. It cannot prove the provenance of a digital asset,
e.g., which digital camera created a photo and who the creator
is.
We have studied a blockchain-based traceability system to
prove the provenance of photos created by digital cameras. It
is similar to the ones used by agricultural products [5]. The
blockchain manages the information about the provenance of
photos, i.e. traceability information that traces which digital
978-1-6654-1760-0/21/$31.00 ©2021 IEEE
DOI 10.1109/Blockchain53845.2021.00089
Authorized licensed use limited to: Middlesex University. Downloaded on November 28,2023 at 14:30:13 UTC from IEEE Xplore. Restrictions apply.
Eric Diehl
Sony Pictures Entertainment Inc., USA
Email: Eric_Diehl@spe.sony.com
camera captured the photo, who is the creator, and who edited
it. A consortium comprising digital camera manufacturers
running a private Ethereum, as Enterprise Ethereum Alliance
[6], could drive the use of the open-sourced Ethereum in the
industry.
A technical challenge is to prove that a converted image is
derived from an original photo taken by a camera because
anyone can copy and modify photos. Image authentication
algorithms [7] [8] [9] [10] have been suggested to verify the
authenticity and integrity of photos. They use digital
watermarking technology to embed a digital signature or
secret information into an image. They have a technical issue
about the robustness of watermarking against image
conversions. Furthermore, the computing cost for integrating
a watermark in image data and detecting watermarking in the
image data is high. Digital watermarking is used only in rare
use cases.
In another solution, a digital camera and every image
conversion software generate a digital signature of the image
data and attach it to the image file. The Coalition for Content
Provenance and Authenticity (C2PA) [11] will standardize an
interoperable format of certificates to verify and authenticate
media provenance. The Content Authenticity Initiative (CAI),
jointly working with the C2PA published a white paper [12].
CAI proposes that the photo editing application records all the
editing actions in certificates to prove that the edited photo
derives from the original photo. It uses a PKI encompassing
all digital cameras and photo editing software to verify a
certificate attached to an image file. This approach is practical
and robust if every image editing software supports the
scheme and some authorities manage this PKI. The solution
may be suitable for photos in a production workflow that uses
a limited set of editing applications. Image conversion
software may modify photos on the Internet, including a photo
editing application, such as mobile applications and cloud
server software. It is unrealistic to hope that each instance
may have a unique signing key issued by a certification
authority.
This paper describes Photrace: a blockchain-based
traceability system of photos. It verifies the certificates of
images signed by a digital camera and image conversion
software. It uses an approach similar to CAI’s one. Our
proposed solution requires managing a PKI for digital cameras
but not a PKI for image conversion software. The followings
are the main features of Photrace.
590
A. The novel scheme of Off-ChainCERTs
Off-ChainCERTs is a novel scheme that securely chains a
trusted origin certificate issued by a digital camera and
successive certificates of image conversions. The basic
concept of Off-ChainCERTs is as follows; A digital camera
signs the captured image in a certificate and appends it to the
captured image file. An image conversion software does not
have a unique signing key pair. It signs a new certificate with
the private key present in the source image file. This makes a
chain of certificates, aka Off-ChainCERTs. The PKI of image
conversion software is not necessary to verify Off-
ChainCERTs. The private key recorded in a source image file
does not carry after the conversion. Another private key is
generated for the converted image file. Thus, nobody can
alter the history of image conversions recorded in Off-
ChainCERTs unless it has all the past image files. An Off-
ChainCERTs can prove that a source image file is available
when the image conversion software converts the source
image file. It traces back the history of conversion and
identifies the original image captured by a digital camera. Off-
ChainCERTs does not prevent unauthorized copy or
modification of photos.
B. The Blockchain-based Traceability system
The blockchain implements a smart contract that runs the
PKI of digital cameras and manages the traceability and
authorship of photos. If a photo creator wants to register a
shot, they use their blockchain client to invoke a transaction
including Off-ChainCERTs. The smart contract checks the
integrity of the Off-ChainCERTs and verifies the signature of
the trusted origin certificate using the PKI of digital cameras.
If the transaction’s issuer is the camera owner, it records that
the creator of the image file is the camera owner. Only the
creator of the original image file or the other users authorized
by the creator can register an image file derived from the
original image to the traceability system and claim the
authorship of the photo. Though anyone can generate Off-
ChainCERTs for a converted image file if a source file is
available, the blockchain securely keeps tracking the
traceability and authorship of image files originating from an
original image taken by a digital camera.
C. Original verification of a converted image with
metadata of the original image.
With the Off-ChainCERTs and the blockchain-based
traceability system, it is possible to trace the authorship of an
image file originating from its camera owner. However, it
cannot prevent the image creator and authorized users from
maliciously manipulating it or replacing it with another image
utterly different from the original image. So, the system
verifies the modified image with the metadata of the original
image that the digital camera added in the Off-ChainCERTs.
A digital camera records the metadata information such as a
thumbnail image, 3D depth-sensing information in the Off-
ChainCERTs. The original verification uses the metadata of
the original image generated by a digital camera instead of
information added by an image conversion software.
The remaining sections are organized as follows. Section
II describes the related works. Section III provides an
overview of the blockchain-based traceability system. Section
IV describes the novel scheme of Off-ChainCERTs. Section
V describes the implementation and evaluation of the
prototype. Section VI discusses concerns of this proposal, and
Section VII concludes this paper.
591
Authorized licensed use limited to: Middlesex University. Downloaded on November 28,2023 at 14:30:13 UTC from IEEE Xplore. Restrictions apply.
II. RELATED WORKS
A framework of decentralized rights management system
based on blockchain was introduced by Fujimura et al. [13].
This framework uses the DRM (Digital Right Management)
technology for distributing protected content. This framework
is not suitable for copyright management of unprotected
photos.
A combination of images authentication algorithms using
digital watermarking technology [7] [8] [9] [10] and
blockchain technology attempts to solve the issue of
traceability of photos. Robert-Alexandru Dobre et al. [14]
proposed a method that an ID of signature is embedded in the
compression data of JPEG. The blockchain verifies the
signature of the ID extracted from the JPEG file.
Authentication of Media via Provenance proposed by Paul
England et al. [15] uses fragile watermarking to integrate the
signature ID in an image data. These solutions have the same
issues as the watermarking described in Section I.
Rishabh Mehta et al. [16] proposed that the blockchain
records perceptual hashes instead of cryptographic hashes.
Perceptual hash represents the perceptual similarity of images
and can be used to search copyright infringement. It cannot
prove the provenance of images. Perceptional hashes may be
used to compare a converted image with an origin thumbnail
image at the original verification of our proposal. But a more
accurate method is preferred.
The JPEG Committee standardized JPEG Privacy and
Security and explored media blockchain applications based on
the standard [17]. The standard defines a format to verify the
integrity and authenticity of a JPEG file. However, it does not
address the traceability issue.
III. PHOTRACE
Photrace is a blockchain-based traceability system for
photos. It supports many use cases to verify the authenticity of
photos and manage their authorship. Figure 1illustrates such
a use case, where an end-user confirms the provenance of a
photo before purchasing it on a photo marketplace. This
section describes how Photrace supports this scenario.
The actors of the use case are a photographer (Alice), the
photo marketplace, an end-user of the system (Bob), the photo
verification service, and the consortium blockchain. Hardware
devices and software that process images are called Entities.
Each digital camera has a unique public/private key pair
issued by its manufacturer. A smart contract (PKI for digital
cameras) on the blockchain registers the corresponding
certificate of the public key and the camera’s owner
information. The procedure to register photos with Off-
ChainCERTs is described in Section IV.
The following explains the use case about how an end-
user can verify the provenance of photos on a photo
marketplace. Please note that the detail of copyright
management is out of the scope of this paper.
1) Alice takes a photo with her digital camera and edits it
using a photo editing application. The digital camera and the
photo editing application append Off-ChainCERTs to the
image files, Image File0 and Image File1. She uses her BC
(Blockchain) client, aka a Blockchain Wallet, to send a
transaction including the Off-ChainCERTs of Image File1 to
register the photos to the traceability system.
 Figure 1 An example of a use case
2) The blockchain verifies the Off-ChainCERTs. If Alice
is the owner of the digital camera, the blockchain records IDs
of all photos recorded in the Off-ChainCERTs with Alice’s
BC address as the author.
3) Alice uploads the edited photo to a photo marketplace.
She also sends a blockchain transaction to authorize that the
photo marketplace uses the photo.
4) The photo marketplace resizes the upload photo with
image conversion software. The software appends the newly
generated Off-ChainCERTs to the resized Image File2. It
sends a transaction including the new Off-ChainCERTs to
register the resized photo to Photrace.
5) The blockchain verifies the Off-ChainCERTs. If the
transaction’s issuer is a user who is authorized by the creator
of the photo, the blockchain records the ID of the resized
photo with the BC address of the user.
6) The photo marketplace uses the resized photo on its
website. It also embeds the link to the photo verification
service on the web page.
7) Bob accesses the web page on the photo market site.
Before purchasing a photo, he wants to check the provenance
of the photo and clicks the link to show the web page of the
photo verification service
8) The photo verification service retrieves the photo
from the photo marketplace and queries the traceability
information of the photo by specifying the ID of the photo in
the Off-ChainCERTs.
9) The blockchain checks if the ID of the photo is
recorded in the blockchain and returns the corresponding
traceability information, including the issuer of the original
photo, i.e., Alice.
592
Authorized licensed use limited to: Middlesex University. Downloaded on November 28,2023 at 14:30:13 UTC from IEEE Xplore. Restrictions apply.
10) The photo verification service gets traceability
information from the blockchain. It performs the original
verification that verifies the difference between the converted
image of the photo on the marketplace and the thumbnail
image of the original photo, recorded in the Off-ChainCERTs.
It provides verification information including the traceability
information and the result of the original verification, e.g.
showing both the converted image and the thumbnail image
of the original image.
11) Bob checks if the author of the photo matches the
photographer’s information in the marketplace and if the
author authorizes the marketplace to use the photo. He may
decide whether to purchase the photo.
The core component of Photrace is the blockchain which
implements the smart contract of traceability management and
the smart contract of the PKI for digital cameras. The photo
verification service with the blockchain can be used for other
use cases such as NFT marketplace and Fake Media detection.
A browser or smartphone application may display traceability
information from the blockchain if it has a BC client. In this
case, the application supports the original verification of the
converted image and shows the result to the end-user as well
as the photo verification service.
 Figure 2 An example of Off-ChainCERTs

IV. OFF-CHAINCERTS
This section explains the proposed scheme of generating
Off-ChainCERTs that securely chains a trust origin certificate
issued by a digital camera and successive certificates of image
conversions published by the image conversion software.
Figure 2 shows an example of Off-ChainCERTs, where
‫ݕݐ݅ݐ݊ܧ‬஺ is a digital camera that creates an original image file
‫݈݁݅ܨ‬଴ and ‫ݕݐ݅ݐ݊ܧ‬஻ is the image conversion software that
converts the original file ‫݈݁݅ܨ‬଴ to the converted image file
‫݈݁݅ܨ‬ଵ .
A. Cryptographic material
Every digital camera holds the following material:
x A unique private key ‫ܭ݅ݎ݌‬௘௡௧௜௧௬ି஺ and the corresponding
X509 certificate ‫ݕ݅ݐ݊ܧݐݎ݁ܥ‬஺ issued and signed by the
manufacturer.
x A manufacturer-defined device identifier ݁‫ܦܫ‬஺
Image conversion software does not hold cryptographic
material.
B. cCERT
Each entity generates a data structure cCERT. The data
structure ܿ‫ܴܶܧܥ‬௜ , where i >= 0, comprises the following
information:
x A data identifier ݀‫ܦܫ‬௜ , an operation identifier ‫ܦܫ݋‬௜
x A public key ‫ܭܾݑ݌‬ௗ௔௧௔೔ whose corresponding private key
is ‫ܭ݅ݎ݌‬ௗ௔௧௔೔
defines ‫ܦܫ݋‬଴ as the device identifier ݁‫ܦܫ‬஺ . Then, ‫ݕݐ݅ݐ݊ܧ‬஺
calculates the data identifier ݀‫ܦܫ‬଴ using (1) with i=0.
 ݀‫ܦܫ‬௜ ൌ ݄ܽ‫݄ݏ‬ሺ‫ܽݐܽܦ‬௜ ‫ܦܫ݋ צ‬௜ ሻ 
Where hash is a cryptographic hash function and || is the
concatenation operator. Eventually, ‫ݕݐ݅ݐ݊ܧ‬஺ calculates the
digital signature ܵௗ௔௧௔బ of all generated data using (2) and (3)
with i=0.
݉‫ܽܪ‬௜ ൌ ݄ܽ‫݄ݏ‬൫݀‫ܦܫ‬௜ ‫ܦܫ݋ צ‬௜ ‫ܭܾݑ݌ צ‬ௗ௔௧௔೔ ‫ܽݐܽܦܽݐ݁݉ צ‬௜ ൯  
ܵௗ௔௧௔బ ൌ ‫݊݃݅ݏ‬൛௣௥௜௄౛౤౪౟౪౯షఽ ൟ ሺ݉‫ܽܪ‬௜ ሻ 
Once the image conversion software ‫ݕݐ݅ݐ݊ܧ‬஻ modified
the image ‫ܽݐܽܦ‬௜ିଵ to produce ‫ܽݐܽܦ‬௜ , it performs similar
operations to generate its cCERT. It generates a new
public/private key pair ‫ܭܾݑ݌‬ௗ௔௧௔೔ and ‫ܭܾݑ݌‬ௗ௔௧௔೔ . It
potentially creates a set of ancillary data ݉݁‫ܽݐܽܦܽݐ‬௜ . It
generates the operation identifier ‫ܦܫ݋‬௜ using (4) with i > 0.
‫ܦܫ݋‬௜ ൌ ݄ܽ‫݄ݏ‬൫‫ܭܾݑ݌‬ௗ௔௧௔౟ ‫ܦܫ݀ צ‬௜ିଵ ൯ 
Then, ‫ݕݐ݅ݐ݊ܧ‬஻ calculates the data identifier ݀‫ܦܫ‬௜ using
(1). Eventually, it calculates the digital signature ܵௗ௔௧௔೔ using
(2) and (5).
ܵௗ௔௧௔೔ ൌ ‫݊݃݅ݏ‬ቄ௣௥௜௄ ሺ݉‫ܽܪ‬௜ ሻ 
೏ೌ೟ೌ೔షభ ቅ

x Optional ancillary data ݉݁‫ܽݐܽܦܽݐ‬௜ ; Off-ChainCERTs C. Chaining cCert certificates

does not define the ancillary data’s semantics. The
‫݈݁݅ܨ‬଴ consists of the generated data set ‫ܽݐܽܦ‬଴ and the
information is application-specific. For instance, it may
trace data set ܶ‫ܽݐܽܦ݁ܿܽݎ‬଴ . ܶ‫ܽݐܽܦ݁ܿܽݎ‬଴ consists of the
hold a thumbnail of the original picture.
certificate ܿ‫ܴܶܧܥ‬଴ and the private key ‫ܭ݅ݎ݌‬ௗ௔௧௔బ generated
Once the digital camera ‫ݕݐ݅ݐ݊ܧ‬஺ captured the image for ‫ܽݐܽܦ‬଴ .
‫ܽݐܽܦ‬଴ by obtaining image sensor readings, it generates a new
public/private key pair ‫ܭܾݑ݌‬ௗ௔௧௔೚ and ‫ܭܾݑ݌‬ௗ௔௧௔బ . It ‫݈݁݅ܨ‬ଵ consists of the converted data set ‫ܽݐܽܦ‬ଵ and the
trace data set ܶ‫ܽݐܽܦ݁ܿܽݎ‬ଵ . ܶ‫ܽݐܽܦ݁ܿܽݎ‬ଵ consists of the
potentially creates a set of ancillary data ݉݁‫ܽݐܽܦܽݐ‬଴ . It

593

Authorized licensed use limited to: Middlesex University. Downloaded on November 28,2023 at 14:30:13 UTC from IEEE Xplore. Restrictions apply.
certificates ܿ‫ܴܶܧܥ‬଴ , ܿ‫ܴܶܧܥ‬ଵ and the private key ‫ܭ݅ݎ݌‬ௗ௔௧௔భ
generated for ‫ܽݐܽܦ‬ଵ .
The signature using the camera’s secret private key
prevents the alteration. The blockchain using the device
identifier ݁‫ܦܫ‬஺ extracts the registered certificate ‫ݕ݅ݐ݊ܧݐݎ݁ܥ‬஺
from the PKI and verifies the validity of ܵௗ௔௧௔బ .
Similarly, the signature using the private key passed in the
previous ܶ‫ܽݐܽܦ݁ܿܽݎ‬௜ିଵ prevents the subsequent certificates’
alteration. The chaining is protected because the signature
ܵௗ௔௧௔೔ uses the private key ‫ܭ݅ݎ݌‬ௗ௔௧௔೔షభ whose public key is
protected in the previous certificate ܿ‫ܴܶܧܥ‬௜ିଵ .
V. IMPLEMENTATION AND EVALUATION
To prove the feasibility of Photrace, we have implemented
a prototype using Ethereum as a permissioned blockchain.
The blockchain has the issue of scalability. We measured the
maximum throughput of transactions to evaluate the benefit of
using Off-ChainCERTs by comparing the cases that a
transaction that includes a single certificate with one that
includes a chain of multiple certificates.
A. Structure of TraceData
TABLE I. and TABLE I. show the structure of a trace
data TraceData stored in a file and a chain of certificates
cCERT embedded in TraceData, respectively. The hash
function is SHA-256. The digital signature uses ECDSA with
the elliptic curve secp256k1. The Solidity of Ethereum
supports both algorithms.

TABLE I. STRUCTURE OF TRACEDATA

Length
Field Description
(bytes)
ܿ‫ܴܶܧܥ‬௜ N A chain of the certificates
Private key of a key pair generated for
‫ܭ݅ݎ݌‬ௗ௔௧௔೔ 32
‫ܽݐܽܦ‬௜ .

information is registered in Data Records on the blockchain.

TABLE II. STRUCTURE OF cCERT Otherwise, it returns an error.
Field Length Description Algorithm1 shows a pseudo code of the function
Length 2 Total bytes of the following fields VerifyCerts() to verify a chain of certificates recursively. It is
݀‫ܦܫ‬௜ 32 Data ID
called by RegisterData(). Its input includes an array of Data
‫ܦܫ݋‬௜ 32 Operation ID
‫ܭܾݑ݌‬ௗ௔௧௔೔ 33 Public Key of a key pair generated for
Record rArray which is initially empty, cCERT which is
‫ܽݐܽܦ‬௜ . (SEC1 format) initially specified in RegisterData(), userRec which is
Length of 2 Total bytes of metaData field, i.e. M, If UserRecord bound to the user’s blockchain address and dID
݉݁‫ܽݐܽܦܽݐ‬௜ this equals zero, ݉݁‫ܽݐܽܦܽݐ‬௜ does not of source data sinkID which initially is empty. First, it creates
exist.
a new DataRecord from the traceability information of
݉݁‫ܽݐܽܦܽݐ‬௜ M Metadata of ‫ܽݐܽܦ‬௜ In the image file, the
thumbnail of an original image, i.e. cCERT. Second, it verifies the integrity of IDs that they
‫ܽݐܽܦ‬଴ must be recorded by the Entity. conform to the calculations of generating the identifications.
ܵௗ௔௧௔೔ 65 Signature signed by a private key of the Third, it verifies an Entity’s signature, and then if there is no
source data set ‫ܽݐܽܦ‬௜ିଵ , (v, r, s format) more source data set, i.e., if a certificate is the origin of trust,
ܿ‫ܴܶܧܥ‬௜ିଵ N cCERT of the source data ‫ܽݐܽܦ‬௜ିଵ
it returns rArray as the successful result. Last, it verifies
signatures of all its source data sets, while it is calling
B. Smart Contract VerifyCerts() with cCERT of the source data set. It is
The functions of the blockchain are implemented as a recursively called by itself until all certificates in cCERT are
Smart Contract written in Solidity of Ethereum. The following verified. If VerifyCerts() finally succeeds, it returns rArray
describes the key function RegisterData() that registers the which contains to all DataRecords be recorded in the
traceability information of photos on the blockchain. blockchain. Otherwise returns NULL.
RegisterData() has as an input cCERT that is cCERTi C. Measurement
stored in TraceData. It checks the validity of the chain of Figure 3 shows the result of measuring the maximum
certificates. If all verifications succeed, all the traceability transaction throughput of the system. The bar graphs in
transaction per second(tps) represent the maximum

594

Authorized licensed use limited to: Middlesex University. Downloaded on November 28,2023 at 14:30:13 UTC from IEEE Xplore. Restrictions apply.
 300 300
transaction per second (tps)
250 250
200 200
tps
rps
150 150
100 100
50 50
0 0
1 2 3 4 5 6
N (number of certificates in Off-ChainCERTs)
Figure 3 Max throughput using Off-ChainCERTs
transaction throughput values of the blockchain system, where
N represents the number of chained certificates. The line
graph in registration per second(rps) represents the number of
certificates registered by a single transaction, where each
value is calculated by tps x N.
We prepared three AWS EC2 x2.large instances
having 8 vCPUs, Intel® Xeon® CPU E5-2686 v4 @ 2.30GHz
processors and 32GiB RAM, installed Ubuntu 18.04 LTS and
go-ethereum 1.9.10. Three nodes were connected via a high
bandwidth network. The consensus algorithm is Clique [18].
The block generation period is one second. Two sealers are
assigned. We used Hyperledger Caliper [19] to measure the
transaction throughput of Ethereum. 1,000 transactions are
sent with a given send rate in each measurement. We
determined that the maximum is where transaction throughput
becomes the limit with significant latency, by increasing the
send rate step by step.
D. Evaluation
We evaluated the system throughput when registering
traceability information on the blockchain using the Off-
ChainCERTs. As shown in Figure 3, when N increases from
1 to 6, the tps value decreases from 142 to 41 (71% down).
Conversely, the rps value increases from 142 to 246 (73% up).
We think that the overhead of processing multiple transactions
among the blockchain nodes is dominant compared with the
overhead of verifying multiple certificates in a transaction by
each node. We confirmed that there is no-demerit of using
Off-ChainCERTs. Instead, it could improve the throughput of
registering the traceability information on the blockchain by
increasing N.
We estimate the throughput required for the traceability
system of photos taken by a digital camera. From the CIPA
report [20], we assume that 50 million digital cameras,
excluding mobile phones, are available in the world. If a
digital camera captures 1,000 photos per year, 50 billion
photos are generated per year. If only 4 % of original photos
are registered to the blockchain, the average throughput
required for the blockchain becomes 63 tps. Our prototype
achieves 142 tps in the case of N=1. Thus, it meets the
requirement. In the case that a photo is edited twice from an
original photo, the number of certificates for the edited photo
becomes triple. If Off-ChainCERTs is not used, i.e. a single
transaction registers a single certificate, the average
throughput required for the blockchain is about 186 tps. It
exceeds the maximum measured transaction throughput.
However, if Off-ChainCERTs is used, our prototype achieves
max 65 tps and max 195 rps in the case of N=3 and meets the
595
Authorized licensed use limited to: Middlesex University. Downloaded on November 28,2023 at 14:30:13 UTC from IEEE Xplore. Restrictions apply.
requirement of rps. A further evaluation on how many photos
will be registered to the blockchain per year, how many
conversions from an original photo are made for a photo is
needed.
VI. DISCUSSION
Photrace uses the novel scheme, Off-ChainCERTs, that
generates a chain of certificates in a series of image
conversions originating from the original photo taken by a
digital camera. Off-ChainCERTs traces back a converted
photo to its original photo. Combining Off-ChainCERTs, the
blockchain-based traceability information management, and
the original verification, Photrace can prove the provenance of
photos, e.g., who is a creator of the photo and which camera
captures a photo. The widespread of photos with Off-
ChainCERTs may mitigate copyright infringement and
dis/misinformation about photos on the Internet.
Photrace does not require a PKI for image conversion
software. Nevertheless, some use cases exclusively utilize a
specific software, e.g. a surveillance camera system. Such
software would have a private and public key pair managed
by the PKI. This enables that only certificated software signs
traceability information with that private key. However,
cascading certificates with signatures signed by the software
is not sufficient. An attacker may compromise the software
and forge traceability information, e.g. removing an
intermediate certificate. Off-ChainCERTs mitigates such
attacks because the signature signed with the private key in a
source file prevents tampering with the chain of certificates.
The blockchain has the issue of transaction throughput
scalability. We assume that 50 million digital cameras are
used in the world, but the shipments of smartphones are
forecast to reach 1.38 billion units in 2021 [21]. The overall
number of photos taken is huge. For instance, more than 1,000
photos are uploaded to Instagram every second [22]. We need
to study and develop the technology to improve the transaction
scalability of this traceability system. Other blockchain
technologies, such as HyperLedger Fabric, may help.
Copyright management of photos based on Photrace and
digital cameras’ metadata for the origin verification are the
future works.
VII. CONCLUSION
In this paper, we proposed the blockchain-based traceability
system for photos on the Internet. With a novel scheme of
chaining certificates Off-ChainCERTs, the system does not
require a PKI of image conversion software. We hope that our
proposal will become the base technology to realize a
blockchain-based application such as copyright management,
fake media detection, and photo marketplace.
REFERENCES
[1] S. Nakamoto, "Bitcoin: A peer-to-peer electronic cash system,"
[Online]. Available: https://bitcoin.org/bitcoin.pdf, 24 May 2009.
[2] Binded, [Online]. Available: https://binded.com/.
[3] E. Foundation, Ethereum, [Online]. Available: https://ethereum.org/.
[4] W. Entriken, D. Shirley, J. Evans and N. Sachs, "Eip 721: Erc-721
nonfungible token standard," [Online]. Available:
https://eips.ethereum.org/EIPS/eip-721, Jan 2018.
[5] R. Kamath, "Food Traceability on Blockchain: Walmart’s Pork and
Mango Pilots with IBM," The Journal of British Blockchain
Association 1(1):1-12, July 2018, 2018.
 [6] "Enterprise Ethereum Alliance," [Online]. Available:
https://entethalliance.org/.
[7] M. Schneider and S.-F. Chang, "A robust content based digital
signature for image authentication," Image Processing, 1996.
Proceedings., International Conference onVolume: 3, 1996.
[8] C.-Y. Lin and F. Chang, "A robust image authentication method
distinguishing JPEG compression from malicious manipulation,"
IEEE Transactions on Circuits and Systems for Video Technology
11(2):153 - 168 March 2001, 2001.
[9] N. D. M. Ping Wah Wong, "Secret and public key image
watermarking schemes for image authentication and ownership
verification," IEEE Transactions on Image Processing 10(10):1593 -
1601, November 2001, 2001.
[10] C.-C. Chang, Y.-S. Hu and T.-C. Lu, "A watermarking-based image
ownership and tampering authentication scheme," Pattern
Recognition Letters Volume 27, Issue 5, April 2006, 2006.
[11] C2PA, Coalition for Content Provenance and Authenticity, [Online].
Available: https://c2pa.org/.
[12] CAI, Content Authenticity Initiative White Paper, [Online].
Available: https://contentauthenticity.org/approach, 2019.
[13] S. Fujimura, H. Watanabe, A. Nakadaira, T. Yamada, A. Akutsu
and J. Kishigam, "BRIGHT: A concept for a decentralized rights
management system based on blockchain," 2015 IEEE 5th
International Conference on Consumer Electronics - (ICCE-Berlin),
2015.
[14] R. A. Dobre, R. O. Preda, C. C. Oprea and I. Pirnog,
"Authentication of JPEG Images on the Blockchain," Conference:
2018 International Conference on Control, Artificial Intelligence,
Robotics & Optimization (ICCAIRO), 2018.
596
Authorized licensed use limited to: Middlesex University. Downloaded on November 28,2023 at 14:30:13 UTC from IEEE Xplore. Restrictions apply.
[15] H. M. E. H. J. W. S. C. F. R. B.-A. e. a. Paul England, AMP:
Authentication of Media via Provenance,
https://arxiv.org/pdf/2001.07886.pdf, 2020.
[16] N. K. S. S. R. S. Rishabh Mehta, "Decentralised Image Sharing and
Copyright Protection using Blockchain and Perceptual Hashes,"
11th International Conference on Communication Systems &
Networks (COMSNETS), 2019.
[17] F. Temmermans, D. Bhowmik, F. Pereira, T. Ebrahimi and P.
Schelkens, "Exploration of media blockchain technologies for JPEG
privacy and security," Optics, Photonics and Digital Technologies
for Imaging Applications VI, April 2020, 2020.
[18] P. Szilágyi, "EIP-225: Clique proof-of-authority consensus
protocol," [ONLINE]. Available:
https://eips.ethereum.org/EIPS/eip-225, 2017.
[19] "Hyperledger Caliper," [Online]. Available:
https://www.hyperledger.org/projects/caliper.
[20] CIPA, Quantity of Total Shipment of DSC ǏWorldwideǐ
Comparison of 2019, 2020 and 2021 :Jan.-May., [Online].
Available: https://www.cipa.jp/stats/documents/e/dw-202105_e.pdf,
2021.
[21] IDC, "2021 Smartphone Growth to Reach Its Highest Level Since
2015, According to IDC," [Online] Available:
https://www.idc.com/getdoc.jsp?containerId=prUS47770921, 2021.
[22] OMNICORE, Instagram by the Numbers: Stats, Demographics &
Fun Facts, [Online] Available at
https://www.omnicoreagency.com/instagram-
statistics/#Quick_Instagram_Statistics, Jul 8, 2021.