SPATIAL THRESHOLDS, IMAGE-OBJECTS, AND UPSCALING - A MULTISCALE EVALUATION

When examining a remotely sensed signal through various scale changes, what is the most appropriate upscaling technique to represent this si gnal at different scales? And how can this be validated? Solutions to these questions were approached by examining how the 660 nm signal of six forest stands vary through four different scales of same-sensor im agery, four traditional resampling techniques, and a new object-specif ic resampling technique. Analysis of the original and modeled datasets suggests that appropriately upscaled imagery represents a more accura te scene-model than an image obtained at the upscaled resolution. Resu lts further indicate the need for a multiscale approach to feature ext raction and upscaling, as no single spatial resolution of imagery appe ars optimal for detecting or upscaling the varying sized, shaped, and spatially distributed objects within a scene. By employing the human e ye as a model, we describe a novel object-specific approach for addres sing this challenge. Upscaling evaluation is based on visual interpret ation, an understanding of the applied resampling theories, and the ro ot mean square error results of 6000 samples collected from a 10 m CAS I scene, and from 1.5 m, 3 m, and 5 m same site CASI images upscaled t o 10 m. Potential application of this object-specific approach in hier archical ecosystem modeling is also briefly described. (C) Elsevier Sc ience Inc., 1997.