TOOLS AND METHODS FOR FUSION OF IMAGES OF DIFFERENT SPATIAL RESOLUTION

Abstract

The usefulness of remote sensing data, in particular of images from Earth observation satellites, largely depends on their spec tral, spatial and temporal resolution. Each system has its specific characteristics providing different types of parameters on the ob served objects. Focussing on operational and most commonly used commercial remote sensing satellite sensors, this paper describes how image fusion techniques can help increase the usefulness of the data acquired. There are plenty of possibilities of combining i mages from different satellite sensors. This paper concentrates on the existing techniques that preserve spectral characteristics, wh ile increasing the spatial resolution. A very common example is the fusion of SPOT XS with PAN data to produce multispectral (3band) imagery with 10 m ground resolution. These techniques are also referred to as image sharpening techniques. A distinction has to be made between the pure visual enhancement (superimposition) and real interpolation of data to achieve higher resolution (e .g. wavelets). In total, the paper describes a number of fusion techniques, such as RGB colour composites, Intensity Hue Saturation (IHS) transformation, arithmetic combinations (e.g. Brovey transform), Principal Component Analysis, wavelets (e.g. ARSIS method) and Regression Variable Substitution (RVS), in terms of concepts, algorithms, processing, achievements and applications . It is mentioned in which way the results of various techniques are influenced by using different pre-processing steps as well as modifications of the involved parameters. All techniques are discussed and illustrated using examples of applications in the va rious fields that are part of ITC’s educational programme and consulting projects.