Multiscale models for data processing: an experimental sensitivity analysis

Hierarchical radial basis functions (HRBFs) networks have been recently introduced as a tool for adaptive multiscale image reconstruction from range data. These are based on local operation on the data and are able to give a sparse approximation. In this paper, HRBFs are reframed for the regular sampling case, and they are compared with wavelet decomposition. Results show that HRBFs, thanks to their constructive approach to approximation, are much more tolerant on errors in the parameters when errors occur in the configuration phase.