ABSTRACT
Wavelet transform (WT) is a powerful computational tool that allows to map signals to spaces of lower dimensionality and to extract their main features. But the transform has many interconnected parameters that have a strong influence on the final results. In this paper, different configurations of wavelets are analyzed in order to optimize the classification and retrieval of two-dimensional signals. The study is applied to nuclear fusion plasma images. Results show how the classification varies when using one or another configuration.
