Face Image Encryption Based on Feature with Optimization Using Secure Crypto General Adversarial Neural Network and...

Area of Science:

• Computer Science
• Cryptography
• Biometrics

Background:

• Increasing demand for data security drives advancements in encryption technologies.
• Biometric encryption, particularly using facial traits, is crucial for secure and convenient data access.
• Existing methods face challenges in balancing security, image quality, and efficiency.

Purpose of the Study:

• To propose a novel face feature encryption technique combining picture optimization, cryptography, and deep learning (DL).
• To enhance key security using an optical chaotic map with a 5D conservative chaotic method.
• To enable secure data mining from encrypted images via a region-of-interest (ROI) network.

Main Methods:

• Utilized a Crypto General Adversarial neural network and chaotic optical map for image encryption and decryption.
• Integrated a machine learning (ML) approach with a "hidden factor" for encryption.
• Employed a modernization network for image decryption and an ROI network for feature extraction.

Main Results:

• Achieved superior performance over existing models with a PSNR of 92%, RMSE of 85%, SSIM of 68%, and MAP of 52%.
• Demonstrated an 88% improvement in encryption speed.
• Ensured significantly improved security without compromising image quality.

Conclusions:

• The proposed face feature encryption technique offers enhanced security and privacy for biometric data.
• The method effectively balances high security, excellent image quality, and efficient processing.
• This approach facilitates privacy-preserving data mining from encrypted facial images.