M.Sc. Projects (1998- )
Virtual Ping-Pong game on Internet demands coordinating user interactions, dynamic gaming behavior, and real-time scene display. It is a real challenge to even offer multi-user gaming capability on Internet with the same demanding features. This project will address the issues in multi-user gaming environment and develop a virtual Ping-Pong game supporting the metaphor of multiple remote players. A single-user Ping-Pong game has been developed by a previous project, which can be refined and extended to a multi-user environment. The project requires Java programming skills.
It is now feasible to compose immersive and interactive virtual worlds and distribute them over the WWW for visualization. Our goal is to compose virtual campus of CUHK and allow the users to navigate the virtual campus using Netscape browser. Some virtual models of CUHK, such as main library, computer lab., conference room, and university platform, are built by previous projects. These models will be linked in a framework for interactive navigation. The main issue to be addressed in this project is to improve the navigation efficiency of large geographic models with real-time visual effect and styles. This project requires Java programming skills and knowledge of data structures.
The country parks in Hong Kong have played an important role in promoting the ecological health and public recreation in our surrounding natures. It is necessary to build up a planning support system that provides on-line visualization of multi-scale, multi-dimension and multi-source data of HK country parks. The system will utilize client-server networking communication and intelligent service capabilities (e.g. neural network/data mining) to support advanced planning of country-park development and tourism. Some preliminary work in data collection and interface design has been conducted jointly with the Dept. of Geography of CUHK.
As the fast development of networks and the wide applications of multimedia technology, a mass of digital videos have arisen. How to quickly browse the large volume of video data and how to express the important contents of videos have emerged the online challenges to be tackled. In general, the video scene cutting and the key frame abstraction are the main approaches used in video briefing displays. Here, we will work on the moving image abstract, known as moving storyboard or multimedia summary, consisting of the interesting features/moving objects selected by users. Such changing between shots can be detected by types like cut, dissolve, fade, wipe and other transitions. Methods used to detect the changing include color histogram, edges, motion, FFT feature. We will contrast and define the metrics for these methods, and examine the performance for online video applications.
Image mosaic effects are wildly applied in print media, domestic decoration and many image beautification applications. However, the current image mosaic methods are mostly based on fixed-size image tiles, simple color adjustment and irregular image segmentation, which is inaccurate and very time consuming with software coding only. In this project, we work on GPU-accelerated image mosaic using density tile replacement and lighting justification optimization keeping original image structure details and providing more expressive luminance effects. Automatic density-based replacement map segmentation and color-based region tiles replacement are performed to facilitate the mosaic process, and delicate luminance adjustment/correction optimization is applied to enhance expressive lighting effects. We facilitate the salience perception of images and similarity correlation between neighboring tiles for our image mosaics. With cartoon images as input, the experimental results have shown the efficiency and high quality performance of our image mosaic display using GPU parallelism.