A technique that automatically locates specific facial points on an unknown face image is presented. This tech-nique uses log Gabor wavelets to represent each facial point by embedding local surrounding features. Using a sample log Gabor response of a facial point, locations on the un-known face image that appear similar to the sample facial point are detected. We identify facial points by employing geometric relationships between the facial points that are invariant to 2D rotations and scaling. An experiment is re-ported that uses three sample face images to determine the log Gabor representation of each facial point, and the tech-nique is tested on an unknown face image. It demonstrates that the log Gabor representation is effective in finding sim-ilar features by successfully detecting and identifying all of the seven specified facial points. 1

Australian sign language, Auslan, uses a combination of hand shapes, positions and movements, as well as fa-cial expressions to communicate. Our ongoing research is to develop an automatic translator that graphically trans-lates English to Auslan. Currently, we have implemented object-oriented C++ graphics libraries to build a whole up-per body model. The model is a kinematic tree that allows physiologically possible movements that are necessary for Auslan sign display. We have experimented with two angle representations: Euler and quaternion. Using interpolation algorithms, we determined that the quaternion angle repre-sentation is more reliable than the Euler angle representa-tion. This paper presents the design and implementation of the Auslan Jam system and our research into angle repre-sentation techniques. It also discusses future development of the model and the sign translator interface, and identifies possible improvements. 1

Lip reading is an important tool used by hearing im-paired people to increase their understanding of spoken lan-guage. Linguistics experiments show that the features of speech that are degraded by noise in audio systems are the features of speech that are most distinct visually. An auto-matic visual speech recognition system should therefore be a beneficial complement to an audio speech recognition sys-tem when the audio system is used in a noisy environment. An automatic lipreading system requires a robust method to track lips from the image sequence, regardless of the vari-ations of lip shapes, colours and movement amongst speak-ers. One of the most common lip tracking methods is the use of active contour models or snakes to detect outer lip con-tours. This paper proposes a robust and adaptable lip track-ing method that uses a combination of snakes and a 2D tem-plate matching technique. The snake, an energy minimising spline, is driven by 2D template matching techniques to find the expected lip contour of a specific speaker. Our experi-ments show that the technique can track the unadorned lips in various colours and shapes of speakers, including the lips of a bearded speaker. 1.