[attachment=11099]
1. Abstract
In recent time watermarking technique becomes a potential solution for copyright protection, authentication and integrity verification of digital media. Among the widely used watermarking techniques, spread spectrum modulation based method becomes appealing due to its inherent advantage of greater robustness and is used widely for various applications. Some watermarking applications, for example, digital television broadcasting, internet protocol television (IP-TV), etc. essentially demand development of low cost watermark algorithms in order to implement in real-time environment. This paper proposes a block based multiple bit spatial domain spread spectrum image watermarking scheme where a gray scale watermark image is represented by less number of binary digits using novel channel coding and spatial bi-phase modulation principle. VLSI implementation using Field Programmable Gate Array (FPGA) has been developed for the algorithm and circuit can be integrated into the existing digital still camera framework. The proposed image watermarking algorithm may be applied for authentication as well as secured communication in real time environment.
2. Main Aim Of The Project
3. Problem Statement
4. Literature Survey
4.1 Data Hiding in Image: Fundamental Issues and Solutions
IN THE recent decade, new devices and powerful software have made it possible for consumers worldwide to access, create, and manipulate multimedia data. Internet and wireless networks offer ubiquitous channels to deliver and to exchange such multimedia information. However, the potential offered by the information technology era cannot be fully realized without the guarantee on the security and protection of multimedia data. Digital watermarking and data hiding1 are schemes to embed secondary data in digital media for a variety of applications, including ownership protection, authentication, access control, and annotation. Data hiding is also found to be useful to send side information in multimedia communication for achieving additional functionalities or enhancing performance.
Data hiding can be considered as a communication problem where the embedded data is the signal to be transmitted. A fundamental problem is the embedding capacity. That is, how many bits can be embedded in a host signal. The answer depends on the required robustness. Earlier works regarding the embedding capacity focused on spread spectrum additive watermarking, by which a noise-like watermark is added to a host image and is later detected via a correlator . This embedding can be modeled as communication over a channel with additive-white Gaussian noise (AWGN) . Other researchers studied the bounds of embedding capacity under blind detection. Zero-error capacity has been studied for a watermark- based authentication system under magnitude-bounded noise , using the principles originally proposed by Shannon , Costa showed that the channel capacity under two additive Gaussian noises with one known to the sender equals to the capacity in the absence of the known noise. This result has been incorporated in information theoretical formulations of data hiding.
The gap between the theoretical embedding capacity in data hiding and what is achievable in practice can be bridged by investigation of such issues as basic embedding mechanisms for embedding one bit and modulation/ multiplexing techniques for embedding multiple bits. The following problems require particular attention:
• Distortion: The distortion introduced by watermarking must be imperceptibly small for commercial or artisti reasons. However, an adversary intending to obliterate the watermark may be willing to tolerate certain degree of visible artifacts. Therefore, the distortions by embedding and by attack are often asymmetric, leading to a wide range of possible watermark-to-noise ratio.
• Actual noise conditions: An embedding system is generally designed to survive certain noise conditions. The watermarked data may encounter a variety of legitimate processing and malicious attacks, so the actual noise can vary significantly. Targeting conservatively at surviving severe noise would lead to the waste of actual payload, while targeting aggressively at light noise could result in the corruption of embedded bits. In addition, some bits, such as the ownership information and control information, are required to be more robust.
• Uneven distribution of embedding capability: The amount of data that can be embedded often vary widely from region to region in image and video. This uneven embedding capacity causes serious difficulty to high-rate embedding.
4.1.1 PRELIMINARIES
In this section, we review a few concepts and principles of data hiding that will be used throughout the discussion in this project.
A. A Data Hiding Framework
A typical data hiding framework is illustrated in Fig 4.1.1 Starting with an original digital media (Io), which is also known as the host media or cover media, the embedding module inserts in it a set of secondary data (b ), which is referred to as embedded data or watermark, to obtain the marked media (I1 ). The insertion or embedding is done such that is I1 perceptually identical to Io, . The difference between I1 and Io is the distortion introduced by the embedding process. In most cases, the embedded data is a collection of bits, which may come from an encoded character string, from a pattern, or from some executable agents, depending on the application.
The embedded data will be extracted from the marked media I1 by a detector, often after In Type-I embedding, the secondary ata, possibly encoded, modulated, and/or scaled, is added to the host signal, as illustrated in Fig. 4.1.1.3 The addition can be performed in a specific domain or on specific features. To embed one bit , the difference between marked signal and the original host signal is a function of , i.e., . can be a major noise source in detection. Although it is possible to detect directly from the knowledge of will enhance detection performance by eliminating the interference. Additive spread spectrum watermarking is an example of Type-I. has gone through various processing and attacks. The input to the detector is referred to as test media , and the extracted data from I2 is denoted by . The difference between I2 and I1 is called noise. In such applications as ownership protection, fingerprinting, and access control, accurate decoding of hidden data from distorted test media is preferred. In other applications such as authentication and annotation, robustness is not critical.
The key elements in many data hiding systems include
• a perceptual model that ensures imperceptibility;
• a mechanism for embedding one bit;
• techniques for embedding multiple bits via appropriate modulation/multiplexing;
• what data to embed;
• how to handle the parts of host media in which it is difficult to embed data;
• how to enhance robustness and security.
We can view these elements through a layered structure shown in Fig 4.1.1.2 analogous to that in communications. The lower layers deal with how one or multiple bits are embedded imperceptibly in the host media. Upper layers for achieving additional functionalities can be built on top of these lower layers.
B. Two Basic Embedding Mechanisms
The embedding of one bit in host media is basic to every data hiding system. Almost all embedding approaches belong