A case study of the influence of shadows and shading on multispectral airborne imaging data.

Abstract

Non-uniform illumination due to three-dimensional shape of land surface and land coverage can cause seriously falsified results when remote sensing data are analyzed automatically, whereas shadow borders and shading patterns are highly informative image features for photogrammetric applications. In this study, the influence of shading and shadows on the radiance measured by a Compact Airborne Spectrographic Images (casi) is investigated. the image data were recorded in the visible and near infrared wavelength range and have a high spectral, spatial and radiometric resolution. We examine the feasibilty of an information extraction from strongly shaded areas based on histrogram properties of shaded image regions and by estimating the noise performance of the imager. We show, that the numerical information content of shaded image areas is only reduced by a small factor compared with well illuminated regions. Analyzing shaded surfaces can be performed with acceptable signal to noise rations using high radiometric resolution sensors. We discuss our findings with respect to the problem of detecting illumination variations and present results of a detection algorithm based on thresholding and a multispectral gradient classification procedure.