Published Paper Details:

Digital image watermarking is still an important method for copyright protection, ownership proof, and image authentication. Among many approaches, frequency-domain watermarking has remained popular because it usually gives better balance between image quality and robustness than simple spatial-domain methods [1][2]. This paper presents a systematic review of classical frequency-domain watermarking with focus on DCT, DWT, DFT, and SVD-based hybrid methods. It also studies how these traditional methods perform when images are changed by modern manipulation tools such as AI-based enhancement, inpainting, semantic editing, and watermark removal systems [3][4]. The review combines literature analysis with an experimental synthesis based on published benchmark results. The main observation is that classical methods still perform well against common attacks like compression, filtering, resizing, and noise, but their strength becomes limited when image content is strongly regenerated or semantically modified by modern tools [1][3].