Small targets detection with hyper-dimensional data

This paper presents the results on the individuation of illegal waste sites, and relatively small objects in Northern Italy using airborne MIVIS data (Multispectral Infrared and Visible Imaging Spectrometer; Deadalus AA5000). Information acquired by the sensors of the MIVIS hyperspectral system, developed by the National Research Council of Italy - LARA (Airborne Laboratory for Environmental Research) project, presents three main characteristics: local scale study, possibility to plan the proper period based on the objectives of the study, high number of spectral bands with high spectral and geometrical resolution. Unlike to common mono and multispectral airborne and satellite remote sensing system, MIVIS hyperspectral scanners provide a such amount of data with high geometrical and spectral resolution that now it's no more necessary to develop complex classification methodology in order to extract maximum information available. The problem is shifted to the selection of a relative small number of bands, less correlated each other, that emphasize and characterize even small targets. The flow of an operative program for the characterization, identification and classification of defined categories of objects is described. Requirements in data collection by airborne MIVIS data oriented to this target are defined. Flight planning, based on the targets like small or unusual features detection (e.g. roof skylight), is important to improve the usefulness of these sensors. The results obtained are encouraging This instrument supported by correct analysis techniques, may offer new interesting prospects in territorial analysis, small target detection and, consequently, environmental monitoring and management.