Hierarchical Clustering of Moving Images Based on Statistical Modeling of 3-D DFT Coefficients

Abstract

This paper statistically models 3-D DFT coefficients of a moving image with our previously presented 2-component spherically-symmetric Gaussian mixture distribution, defines dissimilarity between moving images on the basis of geometric metrics in the model space, constructs hierarchical clustering with the dissimilarity, and applies the clustering to ITE standard moving images.