Improved digital image watermarking using discrete wavelet transform technique

Abstract

Digital watermarking is a huge research area which is progressively growing. It covers theoretical studies, novel techniques, attacks and performance analysis. Over the years, the digital watermarking community has focused on developing new techniques for watermark embedding and detection. Analysis of these techniques leads to methods for attacks and countermeasures which are used to discover faults and limitations in applications, encouraging the development of better ones. Digital watermarking is distinctive depending on its techniques and applications. In general, digital watermarking is distinguished according to media type (image, audio, video and etc.), visibility (visible and invisible), robustness level (fragile, semi-fragile and robust) and the need for original data (blind, semi-blind and non-blind). The scope of this research is invisible digital image watermarking for gray-scale images using Discrete Wavelet Transform (DWT). The system consists of watermark embedding, attacks and watermark extraction. A new method called subband matching was proposed and implemented. For this method, the 1-level DWT coefficients of the watermark image were embedded in 2-level DWT coefficients of the cover image for the same matching subbands. An improved method was also introduced, known as selective subband matching. Modification was made to the algorithms where only selected matching subbands were used in embedding and extracting the watermark in this improved method. These methods were compared and tested against attacks in MATLAB and by using standard benchmark, Stirmark. Experimental results of the proposed methods’ performance were analyzed using Peak Signal to Noise Ratio (PSNR) calculations and Structural SIMilarity (SSIM) index for watermark imperceptibility and robustness, respectively. The improvement could be seen in quality of the watermarked image (imperceptibility) and of extracted watermark (robustness).