Scribd
Upload
3 views

ChaSAM an Architecture Based on Perceptual Hashing for Image Detection in Computer Forensics

2024 IEEE Transaction on All Domains For More Details::Contact::K.Manjunath - 09535866270 http://www.tmksinfotech.com and http://www.bemtechprojects.com 2024 and 2025 IEEE Projects@ TMKS Infotech,Bangalore

Uploaded by

Manju Nath
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
3 views

ChaSAM an Architecture Based on Perceptual Hashing for Image Detection in Computer Forensics

2024 IEEE Transaction on All Domains For More Details::Contact::K.Manjunath - 09535866270 http://www.tmksinfotech.com and http://www.bemtechprojects.com 2024 and 2025 IEEE Projects@ TMKS Infotech,Bangalore

Uploaded by

Manju Nath
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 8

ChaSAM An Architecture Based on Perceptual

Hashing for Image Detection in Computer Forensics


ABSTRACT

The growing prevalence of digital crimes, especially those involving Child Sexual
Abuse Material (CSAM) and revenge pornography, highlights the need for advanced
forensic techniques to identify and analyze illicit content. While cryptographic
hashing is commonly used in computer forensics, its effectiveness is often challenged
because criminals can modify original information to create a new cryptographic
hash. Perceptual hashes address this problem by focusing on the visual identity of the
file rather than its bit-by-bit representation. This study introduces ChaSAM
Forensics, a methodology that efficiently identifies illicit material using perceptual
hashing techniques to track and identify illicit content, with a focus on child abuse
material. Two new perceptual hashing algorithms, chHash and domiHash, were
designed for integration into ChaSAM. The results showed that, under the tested
conditions, the proposed chHash algorithm was more accurate than the established
pHash algorithm when applied in a single iteration. Combinations of algorithms in
two iterations were also assessed.

EXISTING SYSTEM

An overview of perceptual hashing was presented by Farid [35], which addressed the
persistence of harmful online content such as terror-related materials and child sexual
abuse imagery. The paper reviews the mechanisms, advantages, disadvantages, and
real-world applications of perceptual hashing.

Samanta and Jain [22] analyzed the detection of image manipulations using
perceptual hash algorithms and discussed the efficiency of the algorithms in
detecting image similarity. They mentioned that simple modifications such as
changes in color, brightness, and texture are perfectly detected by the algorithms.
However, the same does not occur when drastic changes are present, such as
cropping, rotations, and mirroring.

McKeown and Buchanan [36] presented a large-scale evaluation of several


perceptual hashing algorithms against various image modifications to understand the
effectiveness of these algorithms in detecting similar images despite these
modifications. The study provided insights into the strengths and weaknesses of
popular perceptual hashing algorithms against content-preserving modifications.

Zheng et al [37] proposed MR-DCAE, a deep convolutional autoencoder based on


manifold regularization, which aims to improve the detection of unauthorized
transmissions. This method uses manifold regularization to preserve the intrinsic
structure of the data, providing more accurate and robust detection. The approach has
demonstrated superior effectiveness compared to traditional techniques, especially in
scenarios with noisy and complex data.

Zheng et al. [38] propuseram um método de classificação de imagens de constelação


em tempo real utilizando a rede leve MobileViT. Esta abordagem combina a
eficiência computacional do MobileNet com a capacidade de modelagem de visões
globais do ViT (Vision Transformer), resultando em um modelo que oferece alta
precisão de classificação enquanto mantém baixos requisitos de processamento.
Torres [25] addresses the television and film industry’s difficulty in protecting their
media content against digital pirates. His work involved detecting alterations in video
files transmitted by cable TVs. According to the author, this analysis resulted in a
significant computational processing problem, as it required the computational
analysis of each frame of a video that has 30 frames per second, resulting in
thousands of static images to be processed and compared in search of possible
channels on the Internet that are broadcasting the content without legal authorization.

Faria [15] argues that the analysis of files performed by computer experts is limited
to the hash-to-hash comparison of cryptographic algorithms, mainly in examinations
where files of child pornography are sought, and there is a lack of methods that allow
the computational expert to compare similar files. He also mentions free software
tools and general-purpose similarity hashes; however, they are not very performative.

The work by Menezes and Silva [26] addressed what the authors called the problem
of the modern world, which is the accumulation of various identical or similar files at
the expense of disk storage space. The author argues that programs can compare files
for similarity; some are available for free download on the Internet. However, he
points out that these programs are slow.

In light of this problem, the work of Santos [8] discusses the exponential increase in
crimes related to the production, storage, and sharing of child sexual abuse material
on the internet, which, after police operations, will be subject to forensic
examination to seek the criminal materiality, thus requiring computational methods
capable of supporting the work of the criminal expert.

Hao et al. [39] examined the robustness of perceptual hashing algorithms used in
digital forensics and cybercrime studies. The study highlighted the vulnerabilities of
perceptual hashing algorithms and demonstrated the need for more robust designs to
protect against such attacks It is observed, therefore, that the studies converge to the
increasing existence of devices and files to be analyzed by criminal experts, with the
aggravating factor that many of these files undergo some anti-forensic technique for
their malicious manipulation, aiming to cover up or produce adverse results from the
truth.
Disadvantages
 An existing system didn’t explore Implementation of ESTABLISHED
PERCEPTUAL HASHING ALGORITHMS.
 An existing system didn't implement Architecture Based on Perceptual
Hashing for Image Detection.

Proposed System

This paper presents a forensic methodology named ChaSAM Forensics and the
chHash algorithm, which consists of a computational architecture capable of
identifying images based on their similarities, particularly in the search for CSAM
files and revenge pornography on computer devices seized in police operations. To
achieve this, ChaSAMsearches for files within the target folder only if they have
similar counterparts in the candidate files folder within the source folder.
Additionally, we propose two perceptual hashes to minimize false negative results.

The main contribution of this work lies in the development and integration of two
novel perceptual hashing algorithms, chHash and domiHash, within the ChaSAM
Forensics tool. Unlike existing perceptual hash algorithms, chHash is particularly
effective in identifying similar images while minimizing false positives, especially
when used in combination with dHash or dHash-v algorithms. This dual-iteration
approach significantly enhances accuracy and efficiency in detecting manipulated
images, such as those found in CSAM. Additionally, the use of parallel processing
routines within ChaSAM further improves performance, making it a robust tool for
forensic investigations involving large volumes of digital images.
Advantages

• Designing a forensic analysis tool for processing and identifying modified or


adulterated images;
• Designing and evaluating two perceptual hashing algorithms;
• Building a test image database and using it in the simulations for evaluating the
proposed methodology.

SYSTEM REQUIREMENTS

➢ H/W System Configuration:-

➢ Processor - Pentium –IV


➢ RAM - 4 GB (min)
➢ Hard Disk - 20 GB
➢ Key Board - Standard Windows Keyboard
➢ Mouse - Two or Three Button Mouse
➢ Monitor - SVGA

Software Requirements:
 Operating System - Windows XP
 Coding Language - Java/J2EE(JSP,Servlet)
 Front End - J2EE
 Back End - MySQL

You might also like