Abstract: Method of extraction of information of interest to multi-dimensional, multi-parametric and/or multi-temporal datasets related to a same object under observation through data fusion, in which a plurality of different data sets are provided concerning a single object, with the data related to various parameters and/or different time acquisition instants of said parameters. The data set are subjected to a first processing step by principal component analysis generated by an identical number of datasets with transformed data; and each of the datasets is combined in non-linearly with the corresponding transformed data set to obtain a certain predetermined number of combinations of parameters by weighing using parameters determined empirically using training datasets which determine the values of the non-linear weighting parameters that maximize the value of the new features associated with the data of interest, as compared to those of other data.

Abstract: Method of extraction of information of interest to multi-dimensional, multi-parametric and/or multi-temporal datasets related to a same object under observation through data fusion, in which a plurality of different data sets are provided concerning a single object, with the data related to various parameters and/or different time acquisition instants of said parameters. The data set are subjected to a first processing step by principal component analysis generated by an identical number of datasets with transformed data; and each of the datasets is combined in non-linearly with the corresponding transformed data set to obtain a certain predetermined number of combinations of parameters by weighing using parameters determined empirically using training datasets which determine the values of the non-linear weighting parameters that maximize the value of the new features associated with the data of interest, as compared to those of other data.