Performance Analysis of Wavelet-Based Image Watermarking System for Improved Robustness

Abstract

Digital Watermarking is one of the proposed solutions for copyright protection of digital images. In this paper, we designed and analyzed a new image watermarking system in the wavelet domain using Non-Blind method. The watermark is inserted invisibly into the original image using 2D-DWT. The chosen sub-bands of wavelet for our work are LH, HL, HH and combined HL-HH. In the proposed scheme, mask matrix is generated using the original image with the aid of MD5 algorithm and generated random matrix. The generated hash value from MD5 algorithm is a seed for random matrix generation. This acts as a key for embedding. Since our method is non-blind, original image is required for extraction. Therefore, same hash value (seed) should be used as key for watermark extraction. The generated mask matrix changes the wavelet coefficients as a part of embedding with respect to pixel values of watermark image. The proposed method has two level securities due to two relationship degree factors r1, r2 for mask matrix generation in embedding. The analysis of variation of Gain factor (β) (with different host images and watermark images) for the PSNR of embedded image and correlation of detected watermark is being done. Also, how the variation of detection threshold factor δ affect the detected watermark is being studied. The results of the experimentation using attacks demonstrate the robustness and efficacy of the proposed watermarking scheme. Ownership proof could be established under various hostile attacks.