Digital Watermarking Techniques for Images: Survey

Introduction

Digital watermarking is defined as the imperceptibly altering a work in order to embed information about that work. In the recent years copyright-protection of digital content became a serious problem due to rapid development in technology. Watermarking is one of the alternatives to copyright-protection problem. Digital watermarking can be classified as visible and invisible (Cox, Miller & Bloom, 2002). Invisible watermarks are more secure and robust than visible watermarks. The main characteristics (Cox et al., 2002) of Digital watermark are robustness, imperceptible and security. Depending on the ability of the watermark to withstand normal signal processing operations, digital watermarking can be categorized as robust, fragile and semi-fragile watermarking. Robust watermarks (Cox, Kilian, Leighton & Shamoon, 1997) are detectable even after some image processing operations has been performed on the watermarked image such as image scaling, bending, cropping, and so on. Robust watermarks are mainly used for copyright protection. Fragile watermarks (Wong, 1998) became invalid even if a slight modification is done to the watermarked image. Fragile watermarks are mainly used for authentication purpose. Semi-fragile watermarks allow some acceptable distortion to the watermarked image. Beyond this acceptance level if any modification is done to the watermarked image, the watermark will not be detected. Watermarking technique can be categorized as blind, semi-blind or informed (non-blind) (Katzenbeisser & Petitcolas (Ed.), 2000) on the basis of whether the original host image is required or not during the watermark detection.

Generally digital watermarking of images can be performed in two ways, one in spatial domain and the other in frequency domain. In the spatial domain (Bruyndonckx, Quisquater & Macq, 1995) the watermark is embedded into a host image by changing the gray levels of some pixels in the host image. On the contrary, in frequency domain (Huang, Shi & Shi, 2000) the host image is transformed into the frequency domain by using Discrete Cosine Transform (DCT), Discrete Fourier Transform (DFT) or Discrete Wavelet Transform (DWT). In addition to these methods there are several other methods to perform digital image watermarking (Cox et al., 2002), such as singular value decomposition (SVD) (Chang, Tsai & Lin, 2005), spread spectrum watermarking(Cox et al., 1997) and vector quantization (Wang, Pan, Jain & Huang, 2004; Huang, Wang & Pan, 2001) etc. There are different quality measures used for digital image watermarking. Out of these mostly used are peak signal to noise ratio (PSNR) and normalized correlation (NC) (Shih & Wu, 2005). There are so many quality measures used other than these two such as bit correct ratio (BCR), mean absolute error (MAE) etc. The applications (Cox et al., 2002) of digital watermarking include Proof of ownership, Ownership identification, Transaction tracking, Content authentication and Copy control etc.

In this paper the review is conducted on watermarking of digital images considering both the gray-scale and color images excluding other multimedia. The review is conducted based on three intelligent techniques and independent component analysis applied to digital image watermarking. The rest of the paper is organized as follows: Section 2 presents an overview of the intelligent techniques and independent component analysis. Section 3 reviews papers discusses the literature survey of various techniques. Section 4 discusses insights of the review and section 5 concludes the review with future directions.