This paper describes the performance of the I4U speaker recognition system in the NIST 2008 Speaker Recognition Evaluation. The system consists of seven subsystems, each with different cepstral features and classifiers. We describe the I4U Primary system and report on its core test results as they were submitted, which were among the best-performing submissions. The I4U effort was led by the Institute for Infocomm Research, Singapore (IIR), with contributions from the University of Science and Technology of China (USTC), the University of New South Wales, Australia (UNSW), Nanyang Technological University, Singapore (NTU) and Carnegie Mellon University, USA (CMU).

In this paper we present a study of 25 structural micro features of handwriting extracted automatically from images of grapheme 'th'. The study was carried out on a total of 3823 samples collected from 165 different writers, between 15 and 27 samples per writer. The methods of feature extraction are presented along with the results. We present a direct analysis of the feature usefulness. A measure of usefulness specific to handwriting examination was designed and the usefulness of the features was estimated. It was shown that most of the extracted micro features of 'th' do possess discriminative power. Ranking of features allows simple comparison of their discriminative power.

A method for extraction of strokes from handwriting characters and graphemes is presented. The method allows the modelling of the original pen tip trajectory close to that perceived by humans, thus allowing its use in writer identification and verification tasks. The method is also capable of identifying retraced strokes and recovering hidden loops. Strokes are represented as cubic splines. The method extracts strokes in three stages: vectorisation, merging of skeletal branches and loop recovery, and final adjustment of near-junction and loop pieces. The evaluation of the method is performed by using its results for structural feature extraction and writer classification based on the features.

In this paper a study of some structural features of hand-written letter a' is presented. The features under consideration are structural, style and formation features document examiners use when studying a questioned document to determine its authorship or authenticity.Feature extraction algorithms for the feature set were developed to perform the extraction procedure automatically.The estimate of feature consistency is introduced and measured for each extracted feature. The extracted feature set is analyzed for linear independency and ability to be used for separation of author classes (i.e. authorship decision). The results on a set of 18 authors with approximately 40 letter 'a's per author indicates a high level of consistency and discrimination ability for most of the features analysed.