My research interests include information security and forensics, media-sharing social networks, digital signal processing and communications.
The recent popularity of networked multimedia has made it a critical issue to protect the content security and intellectual rights. With a unique label ("fingerprint") embedded in the multimedia data to identify each recipient, digital fingerprinting has become an emerging forensic tool to track the usage of multimedia content and trace the source of illicit copies. However, the global nature of Internet has enabled a group of users to collectively mount attacks and effectively remove the identifying fingerprints. These attacks, known as collusion, pose serious threats to multimedia forensic systems. We investigate this new challenge of multi-user collusion in multimedia fingerprinting, and our research has resulted in the design of robust multimedia fingerprints as well as reliable and trustworthy forensic tools for traitor tracing.
Behavior dynamics in media-sharing social networks
In the past decades, we have witnessed the emergence of large-scale media-sharing social network communities, for example, Napster, flickr and YouTube, and the Internet traffic has shifted dramatically from HTML text pages to multimedia file sharing. This poses new challenges to the efficient, scalable and robust sharing of multimedia over large and heterogeneous networks. It also significantly affects the copyright industries and raises critical issues of protecting the intellectual property rights of multimedia. With millions of users worldwide, a crucial issue in research in media-sharing social networks is to understand the user dynamics that influence human's behavior and analyze the impact of human factors on multimedia systems. Such an investigation provides fundamental guidelines to the systematic design of multimedia systems and helps develop better technologies to offer secure and personalized services. Our work provides a general framework to model user dynamics and investigates a few major methodologies from signal processing to analyze the impact of human factors on media-sharing social networks.
Adaptive compressed video sensing
We provide advanced signal processing tools to achieve high efficiency, high quality and high resolution in video acquisition, and introduce the novel concept of adaptive compressed sensing for maximum frame rate video acquisition. This work enables the acquisition of a high-resolution video sequence below the Nyquist rate using just one or a few sensors, and it can significantly increase the video frame rate without increasing the sampling frequency of the sensors. The proposed video acquisition framework automatically analyzes the motion and the complexity of the scene, selects the maximum achievable frame rate, adjusts the acquisition process to support a constant play speed of the reconstructed video, and achieves high perceptual quality and sampling efficiency. This invention has a broad range of applications, e.g., terahertz imaging for surveillance, satellite imaging, medical imaging, television transmissions and many others.
My research is funded by Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant and Strategic Project Grants, and Glenrose Rehabilitation Hospital Clinical Research Fund.