A Reliable SVD based Watermarking Schem

TL;DR

This paper introduces a new SVD-based digital image watermarking scheme that enhances security and robustness by embedding watermark components in the image's principal SVD subspace, addressing previous vulnerabilities.

Contribution

The proposed method leverages the SVD subspace for watermark embedding, improving security and overcoming limitations of earlier algorithms like Liu and Tan (2002).

Findings

Watermark extraction is secure against unauthorized removal.

The scheme preserves significant image information in the SVD subspace.

It addresses vulnerabilities in previous SVD watermarking methods.

Abstract

We propose a novel scheme for watermarking of digital images based on singular value decomposition (SVD), which makes use of the fact that the SVD subspace preserves significant amount of information of an image, as compared to its singular value matrix, Zhang and Li (2005). The principal components of the watermark are embedded in the original image, leaving the detector with a complimentary set of singular vectors for watermark extraction. The above step invariably ensures that watermark extraction from the embedded watermark image, using a modified matrix, is not possible, thereby removing a major drawback of an earlier proposed algorithm by Liu and Tan (2002).