Watermarking is designed to protect intellectual property of a digital media. We propose a method for semi-blind watermarking by introducing a novel rank reduction technique ``modified Non-Negative Matrix Factorization (NMF)'' for secure and robust watermarking. The main idea of the study is applying ``modified NMF'' algorithm to pseudo-randomly selected rectangles of the images. Our experiments show that our approach is able to detect watermark of an attacked image with rotation up to certain degrees invisible to the human's eye.
Watermarking is designed to protect intellectual property of a digital media. We propose a method for semi-blind watermarking by introducing a novel rank reduction technique ``modified Non-Negative Matrix Factorization (NMF)'' for secure and robust watermarking. The main idea of the study is applying ``modified NMF'' algorithm to pseudo-randomly selected rectangles of the images. Our experiments show that our approach is able to detect watermark of an attacked image with rotation up to certain degrees invisible to the human's eye.