Reconstructing Trajectory of Camera for Special Effects
It is increasingly common to encounter situations where live action material supplied from film or video and a set of computer graphics models are to appear seamlessly in the given sequence. In the case of a fixed camera or simple linear movements, common compositing techniques are sufficient to fit in the models. However, if the camera undergoes some complicated motion path in 3D space, then adaptation by hand to ensure perspectively correct alignment of the computer graphics models would take a lot of time, if it can be done at all. The ability to reconstruct a camera motion path in 3D space with respect to the real environment and to export it to any animation package will empower animators to create highly specialized visual effects.

The Camera Parameter Reconstruction Technique, CPRT, is proposed. From the input of a series of shape descriptions with their corresponding models, 3D surfaces are recovered. The camera trajectory is then reconstructed with respect to a chosen world coordinate system. The CPRT also checks the incoming list of shape descriptions to ensure that a continuous camera trajectory can be reconstructed. It provides a report to alert the users if there is any missing data needed in any particular frames for the reconstruction. Result from CPRT can be used for augmenting real images.

Computation of Depth Map for Re-Rendering
The objective of this project is to compute depth maps based on given image sequences and to perform re-rendering. A depth map is a 2D function that gives the depth (with respect to the viewpoint) of an object point as a function of the image coordinates. A surface can be obtained after an appropriate model is fitted to the computed depth map. The surface extracted can thus be rendered for visualisation. However, the same visualisation can be obtained directly from the depth map without intermediate surface fitting. Recently a new approach to rendering has emerged: image-based rendering. Image-based rendering systems generate different views of an environment from a set of pre-acquired imagery. In an image-based rendering system, the model consists of a set of images of a scene and their corresponding depth maps. When the depth of every point in an image is known, the image can be rendered from any nearby point of view by projecting the pixels of the image to their proper 3D locations and reprojecting them onto a new image plane. Thus, a new image of the scene is created by warping the images according to their depth maps. A principal attraction of image-based rendering is that it offers a method of rendering arbitrarily complex scenes with a constant amount of computation required per pixel.

Virtual Studio
Just as many TV news-rooms are getting comfortable with computers, a new generation of technology is arriving: virtual set. Virtual set technology allows the broadcasters to use the same blue room as the site of several different shows with different sets, and to switch from one to another virtually instantaneously. Virtual sets also have the advantage of requiring less expensive employee time for set-up, take-down and maintenance, and have the ability to be updated quickly. One of the key problems of such technology is the tracking of the camera's movement and orientation. Currently, this is usually done mechanically. In this project, the tracking of the camera will be done using a CCD camera. A grided wall will be used as the backdrop of the news-casting room. By analysing the grid pattern of the wall, the technique is to recover the camera's pose. Based on this information, the system will then generate the required graphical background (set) that is synchronized with the camera's motion.

Interactive Virtual Studio
The interactive virtual studio (IVS) is an augmented reality system that is capable of creating 3D virtual objects dynamically as the camera moves through the real environment. It enables virtual objects to blend in completely with the real environment. The underlying algorithm is based on simulated motion recovery technique (SMRT). The SMRT is used to extract 3D information from 2D image. With this information, IVS can then render virtual objects into the real environment. This project is a continuation of previous projects; focusing on adding enhanced features and making the system more interactive.

3D Augmented Reality and Profilometry (Seah Hock Soon and Anand Asundi)
With rapid advances in optical metrology and computer technology, it is possible to obtain 3D information of objects using non-contact methods with good accuracy. 3D visualization provides a vivid and easily interpretable scene information as compared 2D imaging. This combination has application in many fields, such as augmented reality and product reengineering. Profiles of 2.5D objects and 360 degrees around a rotating object has been demonstrated and found to be quite successful. However, for augmented reality, 3D profiling across a scene would be useful. The scene could be tracked along a straight line or a circular path or any random curve. Since the scene does not shift the lighting and imaging has to be moving. Thus there is need to calibrate the camera location at each stage. This is one area of investigation. Secondly in product reengineering smaller object need to be profiled. What are the associated problems and solutions to this need to be explored.