The research in this group span all types of digital media text, data, audio, speech, image and graphics, video and animation. In addition, we address integrated usage of these media for versatile multimedia applications as well as technical aspects of their reliable end-to-end transmission over heterogeneous networks and wireless channels.
Highlights of Research Activities TOP
In 1999, there are 10 ongoing research projects with an aggregate funding of $820k, together with 23 ongoing MEng and PhD projects. The group graduated 2 MEng students in 1999.
| Our research thrust
on digital video emphasizes international
video standards, namely ITU’s H.26x and ISO/IEC’s MPEGs, and cover a wide spectrum from analysis, compression, indexing and retrieval, to the Internet applications. In MPEG-4, our work is on encoder optimisation issues, while in MPEG-7 our endeavors are to perform effective indexing and retrieve relevant digital videos over raw video data as well as a compressed domain. In addition, we have been engaged in the development of proprietary image and video encoder and investigating the fundamental issues of digital imaging. |
We continue to participate in international MPEG standardization meetings, as well as contributing technical efforts to the international MPEG community by conducting MPEG’s Core Experiments and VM software integration. In a recent MPEG meeting, our fast motion estimation technology has been included as part of MPEG-4 standards, after more than a year of thorough tests and tough competitions. Besides the source coding, we examine the transmission aspects of digital videos, particularly on the wireless and mobile environment, by investigating fundamental issues such as error resilience, joint source and channel coding, error concealment, etc. | ||
(Left) Still Frame of Video Sequence "Singer" (Right) Fast Polygonal Approximation of the Boundary of the Extracted Video Object Using only 58 Vertices |
(Top) MPEG Generated Motion Vector Field (Bottom) Video Object Tracking based on Motion Vector Field |
| For multimedia applications, a project
on interactive multimedia in electronics engineering education is under
investigation, which aims to develop and implement interactive multimedia
instructional materials for electronics courses. The objectives of this
project are to examine the application of innovative IT to support the
effective dissemination of learning materials, and formulate guidelines
for creating interactive multimedia based materials for classroom instruction.
One of the image processing topics is to use Markov random fields to model texture images. The model integrates the concept of spatial frequency filters and random fields in order to provide an effective way for color image classification and segmentation. |
Applications are found in content
based image indexing and retrieving for medical application. For spatial
audio signal processing, our research is on various areas of the virtual
surround sound system for home entertainment and PC multimedia applications.
These include sweetspot steering and signal processing techniques to enhance
spatial sound effects. Some of our completed projects included creating
a
dual channel Dolby digital audio system based on psycho acoustical signal processing techniques, and developing a personalized 3D audio system that suits the individual listener. In speech signal processing research, the current focus is on techniques for removing background noise, which is inherently present in any speech processing system, such as voice recognition systems or vocoders. The presence of such background noise will inevitably reduce the accuracy of the voice recognition systems and increase vocoder noise. |
|||||
