METHOD AND RIKDEDIN SOFTWARE PACKAGE FOR INTERPRETATION OF REMOTE SENSING DATA

The developed structure-statistical method and RIKDEDIN software package have the main authoring innovations, such as the use of the probability (but not certain) learning samples for algorithmic learning-recognition, the method of Fisher information matrix calculation with the further estimation of classification quality criterion on its basis; the compression of dimension of the initial object feature space in multi- and hyperspectral (multichannel) sensing of environment, the taking into account the changes over time of features and characteristics of objects. Moreover, the RIKDEDIN software package prepares simulation examples for self-test and gives to user the new unique visualization tools for the interactive analysis of multi- dimensional data.