Digital watermarking in the singular vector domain

TL;DR

This paper introduces a novel digital watermarking method that embeds data into the singular vectors of images using SVD, offering imperceptibility and robustness against common image attacks.

Contribution

The paper proposes a new watermarking technique leveraging SVD of images, enhancing imperceptibility and robustness compared to traditional spatial or transform domain methods.

Findings

Effective in grey scale and color images

Resistant to noise and JPEG compression attacks

Maintains image quality while embedding watermarks

Abstract

Many current watermarking algorithms insert data in the spatial or transform domains like the discrete cosine, the discrete Fourier, and the discrete wavelet transforms. In this paper, we present a data-hiding algorithm that exploits the singular value decomposition (SVD) representation of the data. We compute the SVD of the host image and the watermark and embed the watermark in the singular vectors of the host image. The proposed method leads to an imperceptible scheme for digital images, both in grey scale and color and is quite robust against attacks like noise and JPEG compression.