Savvas Argyropoulos

Research areas

Distributed source coding for biometric recognition

A novel biometric authentication scheme based on distributed source coding principles is proposed. Biometric recognition is formulated as a coding problem with noisy side information at the decoder and error correcting codes are employed for user verification. The effective exploitation of the noise channel statistics in the decoding process improves performance. It is also shown that the proposed method increases the security of the stored biometric templates. As a case study, a novel gait recognition system is developed based on the extraction of depth data from human silhouettes and a set of discriminative features.

Multimodal Signal Processing and Fusion

Automatic understanding of human languages remains an interesting problem to be solved. In this work, a multimodal fusion framework is developed for the automatic recognition of Cued Speech language. The robust feature extraction from the audio, lip shape and gesture modalities is examined in detail. Moreover, the inherent correlation among the signals is modelled using a modified Coupled Hidden Markov Model. The contribution of each modality to the fused decision is modified by assessing its reliability and assigning a weighting factor to improve the accuracy of the recognized words.

Video coding for error-resilient transmission

A novel method is proposed for robust video transmission over packet erasure networks based on error resilient coding and forward error correction. Macroblock classification into unequally important slice groups is considered using Flexible Macroblock Ordering and end-to-end distortion estimation. A rate-distortion framework is introduced to compute the contribution of macroblocks to the overall picture quality. The employment of Reed-Solomon codes is examined for the effective protection of the resulting streams for transmission over error-prone channels.