Evaluating the Performance and Resistance of Dynamic AES Image Encryption to Noise and Key Sensitivity

Abstract

In an era of digital information abundance, securing visual content against potential cybersecurity breaches is crucial, given that sensitive visual data can be transmitted or stored and are often susceptible to malicious attacks. This article proposes a new novel and robust framework with AES algorithm combined with dynamically developed keys and IVs in order to provide a stronger cryptographic framework for image encryption. The most widely known framework is a collection of AES modes (ECB, CBC, CTR, etc.) with their effectiveness evaluated through a comprehensive suite of security and performance metrics. This covers traits like randomness assessment through entropy analysis, the Number of Pixel Change Rate (NPCR) and Unified Average Changing Intensity (UACI), and key sensitivity measurements between slight modifications in the encryption key and system robustness.

Real-world challenges are rigorously tested against this system through experiments, which involved introducing common disturbances such as Gaussian and salt-and-pepper noise to the encrypted data, simulating tampering and data corruption scenarios. Further, encryption and decryption efficiency are analyzed comprehensively over images with varying resolutions. obtains significant entropy values close to the theoretical maximum (≈8), high robustness factors with NPCR ≥ 99% and UACI > 33% indicating a very intensive change of pixels at the two moment sides of the reluctant encrypt BIOUNET. A third chapter delves into specific comparisons with other implementations, including ChaCha20) and RSA, demonstrating the various advantages of the new method in terms of randomness, speed, and minimum key length required for a given level of security. The proposed dynamic AES-based encryption system offers secure protection of sensitive images against unauthorized access for various applications in the real-world, as demonstrated by these results, which also prove the system to be a practical, robust, and scalable solution.