A new optimization based color image watermarking using non-negative matrix factorization in discrete cosine transform domain

Abstract

Although different color image watermarking schemes are developed recently in the literature, rare of them use the color image as a watermark due to the low perceptual appearance of the watermarked image. So in this article, a novel optimization-based color image watermarking is introduced in discrete cosine transform (DCT) and non-negative matrix factorization (NNMF) for improving the imperceptibility, robustness, embedding capacity, and security requirements simultaneously. Firstly, the color cover image is divided into different blocks in each of the dimensions [8, 8]. The DC coefficient of each block is merged to generate the DC matrix, which is further factorized by NNMF to obtain the basis and coefficient matrix. The color watermark is secretly embedded into the basis matrix of each color channel. The multi-objective artificial bee colony (MOABC) is used to enhance the robustness and perceptual quality, by providing the optimal multi embedding strength factor. To improve security, a secret key is created from the watermarked image, which is later required to successfully extract the correct watermark. The experiments are performed on variety of images taken from four different databases and results of perceptual quality and robustness confirm that the proposed DCT-NNMF-MOABC watermarking algorithm outperforms the latest schemes present in the state-of-the-art.