Watermarking Techniques for Biometric Security Enhanced by Machine Learning Models: A Systematic Literature Review

Abstract

Abstract—The project develops a biometric watermarking system that encrypts iris and fingerprint images using the Rubik algorithm, producing a unique secure watermark. Convolutional Neural Networks (CNN) analyse the watermark to differentiate genuine biometric features from forgeries, enhancing document authentication security. Adaptive learning enables continuous improvement in detection capabilities while providing robust protection against fraudulent access attempts. This system combines advanced encryption techniques and machine learning to ensure the integrity of biometric data, reinforce authentication processes, and address potential security threats effectively. It represents a significant advancement in safeguarding sensitive information. Keywords—Biometric watermarking, Machine learning, Rubik algorithm, Convolutional Neural Networks (CNN), Document authentication, Adaptive learning