Improved security of the Forensic Images using Encryption and Integrity Verification

Abstract

Digital forensics plays a critical part in any criminal investigations and cybersecurity incidents. As digital wound images are main evidence in investigations, safeguarding their truthfulness and protecting them from any tampering or illegal modification is paramount to maintaining the reliability of the embedded data and the precision of investigative consequences. Despite advances, existing methods continue to address forensic protection from isolated viewpoints rather than within a comprehensive model, focusing on aims such as encryption methods without concurrently addressing the balancing necessities of integrity, confidentiality, and authenticity. Current frameworks handle wound image classification, authentication, and secure storage as independent processes. Additionally, many classification methods assume digital wound images are secure and seldom combine schemes to validate whether digital wound images were tampered with before analysis. This paper suggests a unified end-to-end framework that integrates deep learning techniques and an encryption method within a single model to enhance the truthfulness and accuracy of forensic digital wound images. The principal objective is to employ learning techniques to classify digital wound images, authenticate the original images through accurate localization before any encryption method, and finally secure the verified evidence based on a chaos method. This combination guarantees integrity, confidentiality, and authenticity during forensic investigation. Experimental results demonstrate that the framework achieves a tampering discovery accuracy exceeding 99.6% for all tested data types, with recall reaching 1.0. The wound image classification achieved a weighted average accuracy of 91%. Also, the encryption achieved high security with the best entropy value of 7.9998. The framework confirmed high-speed performance, making it very appropriate for any real-time applications.