Аннотация
The purpose of this study is to develop a secure biometric data protection system that integrates artificial intelligence and cryptographic techniques to enhance data privacy and integrity. The proposed system employs a convolutional neural network (CNN) to extract features from multimodal biometric inputs, including facial images, retinal scans, and fingerprints. A generative adversarial network (GAN) is then trained on these features to produce synthetic biometric representations that closely mimic real data. To ensure confidentiality and authenticity, the El-Gamal cryptosystem is used to encrypt both the features and the biometric images, while a digital signature mechanism based on the SHA-256 hash function secures the data against tampering. The CNN achieved a classification accuracy above 99.8%, with GAN training stabilizing at a discriminator loss of approximately 0.3 and a generator loss of around 4.0. The encryption and signature verification processes demonstrated consistent success, confirming the robustness of the pipeline. This research concludes that the integrated approach is effective in safeguarding biometric data against forgery and unauthorized access. Practically, it provides a viable solution for secure transmission and storage in biometric authentication systems, with strong potential for deployment in high-security environments.