A Novel ICA-Based Extraction Scheme for Image Watermarking
Gao-Rong Zeng, Zhengding Qiu · 2008
We propose a novel blind robust watermarking scheme which embeds a digital watermark into the DCT coefficients of the host image based on human visual models (HVS). Watermark extraction utilizes independent component analysis (ICA) technique which is a blind source separation procedure without requiring the host image and the original watermark. Our method is an attempt toward the proof that the Kerckhoffs principle can be suitable in the watermarking framework. Through a series of experiments, it is shown that the proposed method is effective against the attacks, such as JPEG compression, image resizing and filtering.