On improving dissimilarity-based classifications using a statistical similarity measure

Abstract

The aim of this paper is to present a dissimilarity measure strategy by which a new philosophy for pattern classification pertaining to dissimilarity-based classifications (DBCs) can be efficiently implemented. In DBCs, classifiers are not based on the feature measurements of individual patterns, but rather on a suitable dissimilarity measure among the patterns. In image classification tasks, such as face recognition, one of the most intractable problems is the distortion and lack of information caused by the differences in illumination and insufficient data. To overcome the above problem, in this paper, we study a new way of measuring the dissimilarity distance between two images of an object using a statistical similarity metric, which is measured based on intra-class statistics of data and does not suffer from the insufficient number of the data. Our experimental results, obtained with well-known benchmark databases, demonstrate that when the dimensionality of the dissimilarity representation has been appropriately chosen, DBCs can be improved in terms of classification accuracies. © 2010 Springer-Verlag.