A comprehensive review on secure image watermarking using cryptography and DL

The widespread distribution of digital images through cloud platforms, social media, and intelligent systems has intensified concerns regarding copyright protection, ownership verification, and content integrity. Traditional digital image watermarking techniques often exhibit limited robustness against geometric and signal-processing attacks and remain vulnerable to unauthorized watermark manipulation. Recent research has, therefore, explored hybrid approaches that integrate cryptographic techniques with Deep Learning (DL) to improve watermark security, robustness, and imperceptibility. This paper presents a comprehensive review of recent cryptography- and DL-based image watermarking methods, focusing on security enhancement strategies, robustness mechanisms, and commonly adopted performance metrics. It further examines challenges related to computational complexity, scalability, interpretability, real-time processing, and resource-constrained deployment, particularly in edge computing and Internet of Things (IoT) environments. Finally, key research gaps, open challenges, and future directions are identified to support the development of secure, robust, intelligent, and adaptive image watermarking systems.