Wavelet histograms for gesture classification

This paper evaluates the efficacy of directional information of wavelet multi-resolution decomposition for histogram-based classification of human gestures represented by spatio-temporal templates. This template collapses temporal component into gesture representation in a way that no explicit sequence matching or temporal analysis is needed, and characterizes the motion from a very high dimensional space to a low dimensional space. These templates are modified to be invariant to translation, rotation and scale. Two dimensional, 3 level dyadic wavelet transform applied on these templates results in one low pass subimage and nine highpass directional subimages. Histograms of wavelet coefficients at different scales are compared to establish significance of available information for classification. The preliminary experiments show that while the statistical properties of the template provide high level of classification accuracy, the available information in high pass or low pass decompositions by itself is not sufficient to provide significant efficiency of accuracy.