Angular signature retrieval and comparison with spectral signatures from AirMISR data

Most image classification techniques are built on the spectral variability of the data, which is the primary information retrieval mechanism for nadir-acquired images. However, with the advent of simultaneous multi-directional observations, like with the Multi-angle Imaging SpectroRadiometer (MISR) instrument on NASA´s Earth Observing System (EOS), to be launched on the Terra platform in July 1999, the directional variability on the data can also be exploited. The additional information content in the directional data, which is called an angular signature, is expected to be particularly useful in the identification and classification of vegetated land surfaces because it derives from the degree of how much shadowing within a 3D canopy may be detected. The shadow fraction of a vegetated surface element as seen from a particular view direction contains unique information about the canopy structure and architecture when observed from multiple directions; this information does not manifest itself in any spectral signature