Decision level fusion by clustering algorithms for person authentication

In this paper, the use of clustering algorithms for decision level data fusion is proposed. Person authentication results coming from several modalities (e.g. still image, speech), are combined by using the fuzzy k-means (FKM) and the fuzzy vector quantization (FVQ) algorithms, two modification of them that use fuzzy data FKMfd and FVQfd, and a median radial basis function (MRBF) network. The modifications of the FKM and FVQ algorithms are based on a novel fuzzy vector distance definition, and they utilize the quality measure of the results, that is provided by the authentication methods. Simulations show that the proposed algorithms have better performance compared to classical clustering algorithms and other known fusion algorithms.