Ian J. Palmer

Gender Classification Based on 3D Face Geometry Features Using SVM

In this work we have used non-linear Support Vector Machines (SVMs) for gender classification. The SVMis applied to triangular meshes representing human faces. In this work we rely on handful of 3-dimentional facial features which are extracted from the corresponding geometry meshes. The experimental results show that in our method the error rate is 17.44% on average. It is thought that the approach used to determine gender prior to face recognition would make an automatic face recognition system more efficient.

High-level interpretation in virtual environments

As virtual environments applications become more complex, there is a need to interpret user interaction in terms of high-level concepts. In this paper, the relations between conceptual representations of actions and their physical simulation in virtual worlds is investigated. A model inspired fromnatural language processing (NLP) research in the linguistic interpretation of dynamic scenes has been developed. The experiments are based on real-time animation software, which has been enhanced with a symbolic information processing layer, originally developed for NLP-based animation. We report the implementation of a high-level interpretation module that is able to recognize complex actions fromlow-level physical events in the virtual world and discuss its performance as well as directions for further developments.

A survey of partial differential equations in geometric design

Computer-aided geometric design is an area where the improvement of surface generation techniques is an everlasting demand, since faster and more accurate geometric models are required. Traditional methods for generating surfaces were initially mainly based upon interpolation algorithms. Recently, partial differential equations (PDE) were introduced as a valuable tool for geometric modelling, since they offer a number of features from which these areas can benefit. This work summarizes the uses given to PDE surfaces as a surface generation technique together with some other applications to computer graphics.

Natural language control of interactive 3D animation and computer games

In this paper we describe a fully implemented system for speech and natural language control of 3D animation and computer games. The experimental framework has features that have been emulated from the popular DOOM™ computer game. It implements an integrated parser based on a linguistic formalism tailored to the processing of the specific natural language instructions required to control a player character. This parser outputs structured message formats to the animation layer, which further interprets these messages to generate behaviours for the scene objects. We have found that interactive control significantly impacts on the behavioural interpretation of natural language semantics. Besides bringing stringent requirements in terms of response times for the natural language processing step, it determines the level of autonomy that the animated character should possess, which in turn influences the generation of behavioral scripts from natural language instructions.

Sharing attractions on the Net with VPark

We present a networked virtual environment (NVE) system and an attraction building system based on Windows NT that enables users to introduce their own scenario-based applications into a shared virtual environment. Our goal was to develop and integrate several modules into a system capable of animating realistic virtual humans in a real-time performance. This includes modeling and representing virtual humans with high realism and simulating human face and body movements in real time. Realism becomes quite important in NVEs, where the communication among participants is crucial for their sense of presence.

Urban flood risk analysis for determining optimal flood protection levels based on digital terrain model and flood spreading model

The objective of the paper is to present a new risk-analysis approach for the assessment of optimal flood protection levels in urban flood risk management, which is based on an active contour method. Although the active contour method is a very popular research topic, there has been no attempt made on deriving a model for simulating flooding and inundating to date, as far as we are aware. We have developed a flooding prototype system, which consists of two main parts: a digital terrain model and a flood simulation model. The digital terrain model is constructed using real world measurement data of GIS, in terms of digital elevation data and satellite image data. A pyramidal data arrangement structure is used for dealing with the requirements of terrain details with different resolutions. A new flooding model has been developed, which is useful for urban flood simulation. It consists of a flooding image spatial segmentation based on an active contour model, a water level calculation process, a standard gradient descent method for energy minimisation. When testing the 3D flood simulation system, the simulation results are very close to the real flood situation, and this method has faster speed and greater accuracy of simulating the inundation area in comparison to the conventional 2D flood simulation models.

A Real-time Dynamic Simulation Scheme for Large-Scale Flood Hazard Using 3D Real World Data

We propose a new dynamic simulation scheme for large-scale flood hazard modelling and prevention. The approach consists of a number of core parts: digital terrain modelling with GIS data, Nona-tree space partitions (NTSP), automatic river object recognition and registration, and a flood spreading model. The digital terrain modelling method allows the creation of a geometric real terrain model for augmented 3D environments with very large GIS data, and it can also use information gathered from aviation and satellite images with a ROAM algorithm. A spatial image segmentation scheme is described for river and flood identification and for a 3D terrain map of flooding region growth and visualisation. The region merging is then implemented by adopting flood region spreading algorithm (FRSA). Compared with the conventional methods, our approach has the advantages of being capable of realistically visualising the flooding in geometrically-real 3D environments, of handling dynamic flood behaviour in real-time and of dealing with very large-scale data modelling and visualisation.

Multimedia semantic features and image content description

Content based indexing and retrieval of images and video requires a proper semantic description for image content. The paper discusses the mapping of high level, application specific features to the visual primitives that are accessible through image processing techniques. A major difficulty is that there are currently no established methodologies for describing the contents of an image in terms of semantic features, and we suggest that semiotic approaches could be adapted to the task of image description in limited areas of expertise. After reviewing current trends in the mapping of high level to low level features, we present preliminary results that suggest a possible strategy for mapping semiotic descriptions to image processing primitives.

Automatic Reconstruction Of 3D Environment Using Real Terrain Data And Satellite Images

This paper presents a novel 3D reconstruction method for large-scale 3D environments. There are three core components of our work: dynamic terrain modelling, river and water region identification and modelling using an active contour model and primitive shape matching method. Real-time environment reconstruction is constructed using real measurement data of GIS, in terms of digital elevation data and satellite image data. A Nona Tree Space Partitions (NTSP) algorithm is proposed for dealing with very large data processing and visualisation. A new geometric active contours model is used to automatically segment interesting image areas such as water or flooded regions, forest region and residential region. A primitive shape matching method is proposed to detect the residential objects, such as buildings and houses. The experimental results demonstrate that our approach is a promising one, which is able to deal with large environment reconstruction effectively.

Virtual Rehearsal Over Networks

Traditional rehearsals require all the necessary participants, actors, director and some technical crew to be present in a single location at the same time. With the advent of new technology, the rehearsal process can be more flexible. The system described in this chapter acts as a complement to existing rehearsal methods. More often than not, rehearsals take place without a constructed set. In a virtual rehearsal environment, the set can exist from the start of the process. With the ability to see each other and by interacting in a virtual space, the cast gets an idea of what the final performance will look like much sooner than before. This offers significant benefits for the characterization and blocking process. As virtual reality technologies expand over the Internet, so participants in theater can use the new opportunities to create virtual interactive performance spaces. The cast and crew of a piece can “meet” in a computer-based model regardless of their geographic location and experience the proposed environment throughout the whole process, making any necessary changes as they progress. Users not only “exist” within the space, but also have the ability to manipulate it. With additional third-party software, participants can see and communicate with other avatars.