Implémentation et développement d’algorithmes avancés dédiés à la sécurité et à l’identification

Abstract

This doctoral thesis focuses on the development and implementation of advanced chaotic algorithms to enhance information security and identification systems. Using the inherent unpredictability of chaotic systems, the research proposes novel algorithmic models tailored for robust data encryption, particularly in the context of digital and biomedical imaging. A central contribution is the Improved Composite Chaotic Map (ICCM), a dynamic model that integrates classical quadratic and cubic chaotic behaviors, augmented by an additional control parameter to expand the keyspace and enhance flexibility. The chaotic properties of the ICCM are validated through Lyapunov exponent analysis and bifurcation diagrams, achieving a Lyapunov exponent of 21.42, which surpasses that of previous models. Incorporated into a hybrid optodigital image encryption scheme, ICCM demonstrates strong resistance to statistical, differential, and brute-force attacks. Its effectiveness is measured using performance metrics such as a high mean square error (MSE > 10), low correlation coefficients (Cr < 0.0001 − 0.0028), and a low peak signal-to-noise ratio (P SNR < 10dB), confirming its ability to obscure and protect sensitive data. In addition to general data encryption, the thesis addresses medical image security through a customized Pre-processing encryption-decryption technique for retinal images. This method utilizes dual chaotic sequences derived from the logistic map and integrates biometric information, such as fingerprints, to generate dynamic encryption keys. Comprehensive testing, including histogram analysis, sensitivity analysis, PSNR, and correlation evaluation, confirms the method’s ability to maintain image integrity while ensuring privacy and resistance to common cryptographic attacks. In general, this work contributes significantly to the field of chaos-based cryptography by introducing adaptable and secure encryption mechanisms suitable for general-purpose and domain-specific applications. It also paves the way for further innovations in data protection and biometric security