Spectral classification of WorldView-2 multi-angle sequence

Abstract

In this study, we investigate the ability of the spectral data from a multi-angle WorldView-2 image sequence to improve classification accuracy of an urban scene. Specifically, we investigate the multi-angle reflectance, as well as two data extraction methods applied to the reflectance data, developed from thirteen images collected over downtown Atlanta, GA in Dec. 2009. These images were collected sequentially by WorldView-2 within two minutes and range from approximately 30 degrees off-nadir southward to 30 degrees off-nadir northward.