Color Image Watermarking with Watermark Authentication against False Positive Detection Using SVD

Abstract

With growing need of exchanging digital information, digital watermarking has been an important tool to help content creators prove ownership if there is any dispute arising due to copyrights infringement. Robustness, imperceptibility and security are most imperative attributes for any efficient watermarking scheme. This requires the embedded watermark to be resilient and retrievable even if the content undergoes one or more signal processing operations, in part or as a whole. Some common image processing operations to be tackled are compression, noise addition, rotation, filtering, geometric distortions, scaling, etc. This paper presents a color image watermarking technique based on Discrete Wavelet Transform (DWT) and Singular Value Decomposition (SVD) to eliminate false positive detection of watermark. In this paper, either of the two copies of digital signature of the Eigen-vectors of the claimed watermark is used to authenticate itself before using it for extraction. The color image is represented in Hue, Saturation and Value (HSV) model. The watermark and its signature is inserted in 1-level and 4-level DWT coefficients respectively, of the value component of the colored image. Peak Signal to Noise Ratio (PSNR) of 55.3968 dB is achieved for the color Lena image and the watermark image is recovered with Correlation Coefficient (CC) as 0.9197.