This paper assesses the merits of three diflerent approaches t o pixel-level h u m a n skin detection. T h e basisfor the 3 approaches has been reported recently in the literature. T h e first two approaches [1, 21 use simple ratios and colour space transforms respectively, whereas the third is a numerically eficient approach based o n a 3-D RGB probability map, first implemented by Rehg [3]. T h e Bayesian probabilities are made possible t o compute only with the availability of a large appropriately labeled database. Over 12,000 images f r o m the Compaq skin and nonskin databases [4] are used t o quantitatively assess the three approaches. Thresholds are determined empiricall y t o detect 95% of all skin-associated pixels and assessment is then made in terms of the percentage of non-skin pixels incorrectly accepted. T h e lowest of these false acceptance rates is found t o be about 20% given by the 3-D probability map.

This paper describes the development and quantitative assessment of an approach to face detection (FD), with the application of image classification in mind. The approach adopted is a direct extension of an earlier approach by Huang [Pattern Recognition 19941. Huang's intensity based approach is found to be susceptible t o variations in lighting conditions and complex backgrounds. It is hypothesised that by integrating colour information into Huang's approach, the number of false faces can be reduced. A skin probability map (SPM) is generated from a large quantity of labeled data (530 images containing faces and 714 images that do not) and is used to pre-process colour test images. The SPM allows image regions to be ranked in terms of their skin content, thus removing improbable face regions. The performance improvements are shown in terms of false acceptance (FA) and false rejection (FR) scores. As a front-end to Huang's approach, the benefits of skin segmentation can be seen by a reduction in the FA score from 79% to 15% with a negligible impact on FR.

This paper evaluates lip features for person recognition, and compares performance with that of the acoustic signal. Recognition accuracy is found to be equivalent in the 2 domains, agreeing with the findings of Chibeluushi. The optimum dynamic window length for both acoustic and visual modalities is found to be about 100ms. Recognition performance of the upper lip is considerably better than the lower lip, achieving 15% and 35% identification error rates respectively, using a single digit test and training token.