Cyber Forensic Security in Digital Multimedia Communication Using Deep Learning

Abstract

In the computing world, Cyber forensic security for digital multimedia is an instigative and gruelling field. With the rapid-fire increase in the use of digital technology, crimes moment are committed using contemporary ways that don't involve physical contact. As a result, forensic specialists are unfit to examine and dissect the data at the crime scene. A change in the disquisition ways is necessary to achieve effective disquisition of crimes involving advanced technology. The forgery of digital images compromises the authenticity and integrity of the images. In recent times, the need for the forgery detection algorithm has increased because of the rapid-growth and availability of imaging processing software and the advancements made in digital cameras. With the growing frequency of digital communication, cybercrime similar to image forgery have surfaced as significant challenges. Traditional forensic methods struggle to manage these advanced ways, challenging innovative results. To improve the detection and localization of manipulated areas in digital images, this study proposes a hybrid deep learning model that integrates Long Short-Term Memory (LSTM) networks with Convolutional Neural Networks (CNNs). Additionally, the Scale-Invariant Feature Transform (SIFT) algorithm is incorporated to enhance precision and robustness. The model effectively identifies forged regions, classifies tampering types (e.g., copy-move, image splicing), and delineates manipulated areas using bounding boxes. By generating double masks, it strengthens the accuracy of forensic analysis, contributing to advancements in digital forensic security. This exploration aims to use deep learning methods to produce a reliable frame for relating and examining these attacks, fulfilling an essential demand in the field of cyber forensics.