Digital-Image-Watermarking

Digital Image Watermarking method based on hybrid DWT-HD-SVD Technique: Attacks, PSNR, SSIM, NC

About

Watermarking is identified as a major means to achieve copyright protection. It is technology of information hiding which is used to hide dedicated data in digital media like images, musics, videos etc. The main goal of this project is to provide basic watermark toolbox for researchers to evaluate watermarking methods under various attacks.

The simulation is based on the IEEE Access journal, An Optimized Image Watermarking Method Based on HD and SVD in DWT Domain. The following methods are used:

Discrete Wavelet Transformation (DWT)
Hessenberg Decomposition (HD)
Singular Value decomposition (SVD)

First, in the embedding process, the host image is modified to embed watermark image. Different attacks are applied to evaluate the robustness and invisibility of the proposed method by considering peak signal-to-noise ratio (PSNR), and structural similarity (SSIM). Finally, the watermark image is extracted, and the robustness is evaluated considering normalized correlation (NC).

Attacks

The following attacks are considred:

Attack	Specification
Filter attack	Median filter (3 × 3)
	Gaussian Low-pass filter (3 × 3)
	Average filter (3 × 3)
Noise attack	Gaussian noise ( var = 0.001)
	Salt & peppers noise (0.001)
	Speckle noise (0.001)
Compression attack	JPEG compression (QF=50)
	JPEG2000 compression (CR=12)
Histogram equalization (HE)	HE attack
Motion Blur attack	Motion blur (theta = 4, Len = 7)
Sharpening attack	Sharpening (strength = 0.8)
Rotation attack	Rotation (2 degree)

Getting started

In order to make the copy of the repo, please fork it; otherwise simply Clone or Download it to your local device.
Run main.m file in MATLAB. It is recommended to run it section by section; otherwise you would need to wait whole code to run.
The initial sections merely, run code to illustrate the proposed watermarking method.
The mid sections, plots NC, PSNR, SSIM values for each alpha (Fig. 5, 6 and 7 paper).
Last sections, evaluates the invisibility and robustness of the watermarked image and extracted watermark logo, for different watermark image size under various attacks (fig. 8, 9, and 10 paper).
Finally, the NC value of different attack parameters are evaluated for each attack (fig. 11 paper).