Hierarchical multiscale modeling of wavelet-based correlations

This paper presents a multiscale-based analysis of the statistical dependencies between the wavelet coefficients of random fields. In particular, in contrast to common decorrelated-coefficient models, we find that the correlation between wavelet scales can be surprisingly substantial, even across several scales. In this paper we investigate eight possible choices of statistical-interaction models, from trivial models to wavelet-based hierarchical Markov stochastic processes. Finally, the importance of our statistical approach is examined in the context of Bayesian estimation.