Image and video watermarking for wireless multimedia sensor networks

Imran, Noreen
Seet, Boon-Chong
Fong, Alvis
Item type
Degree name
Doctor of Philosophy
Journal Title
Journal ISSN
Volume Title
Auckland University of Technology

Enormous technological growth in wireless communications and CMOS-sensor electronics has enabled the deployment of low-cost, battery-powered multifunctional embedded camera sensor at remote locations. These tiny gadgets adheres a major improvement over traditional wireless sensor networks (WSNs) by accumulating multimedia data into the system. Wireless multimedia sensor networks (WMSNs) are expected to be the solution of many stimulating range of applications such as critical infrastructure surveillance, person locator services, theft control, environmental, health monitoring systems and much more. Many of these applications have mission-critical tasks, which may process the received multimedia content for decision making purpose and thus require that security be considered as a vital concern. Inadequate use of multimedia content that involve tampering or forgery may cause manipulated information distribution that leads to unwanted consequences.

Watermarking, a more flexible and lower complexity solution, is able to ensure that the semantic meaning of the digital content has not been modified by illegitimate sources, while being sustainable against wireless channel errors, lossy compression and other signal processing primitives. Attributed to our literature review, we found that there still exists a considerable research gap in designing security solutions for WMSNs. Therefore, the aim of this thesis was to develop a digital watermarking system for visual information (image/video), complaint with the design requirements of WMSNs based applications. The resource-limited camera sensor nodes need to encode their own raw video data before transmitting them to the sink (or intermediate relay nodes), which makes it essential for sensor nodes to have a video coding paradigm that exploits simple encoding and complex decoding architecture.

The thesis is structured into seven chapters. The first three chapters provide the introduction, background, and literature analysis. The fourth chapter deals with comparison, analysis, and the selection of an appropriate DVC video codec for a given WMSN application. It provides an insight about the computational (encoding/decoding) complexity, energy consumption, node and network lifetime, processing and memory requirements, and the quality of reconstruction of these video codecs. In chapter five, we propose an enhanced semi-oblivious energy-aware adaptive watermarking scheme for WMSNs, which considered key characteristics such as the embedding capacity, security, imperceptibility, computation, and communication energy requirements. We evaluated the distortion in cover image due to watermark redundancies, the number of embedding locations with respect to two channel adaptive parameters, and the impact of compression of cover image on the correctness of extracted watermark. In addition, we investigated the robustness of the scheme against statistical analysis attacks. Chapter 6 presents a novel, energy-efficient, low-complexity, blind, and imperceptible video watermarking scheme based on transform domain Wyner-Ziv (WZ) coding, which builds on the principles of DVC. It gives an insight about the practical implementation of the proposed scheme on a fully functional WZ codec and its evaluation using real video sequences captured from embedded video camera sensors. In addition, the derived analytical models are used to examine the energy consumption, feedback requests, and rate-distortion performance, embedding capacity, imperceptibility, node and network lifetime in reference to the proposed scheme. Finally, we concluded our work in seventh chapter.

Image and video watermarking , Wireless multimedia sensor networks
Publisher's version
Rights statement