Sudhakar Prasad (University of New Mexico), Qiang Zhang (Wake Forest University)
Keywords: hyperspectral imaging, material identification, segmentation, shape primitives, moving averages, rotational kine
Abstract:
We present simulation-based studies of a spectral correlation approach to segment and disambiguate materials on the surface of a satellite from its hyperspectral image (HSI) data. The kinematical evolution of the segment boundaries as inferred from a time series of poses of a rotating satellite contains unique signatures and should thus enable the extraction of the rotational parameters and local surface morphology of the object for the purposes of SOI. As a first step in the realization of this goal, we demonstrate the uniqueness of recovery of the geometrical shape, orientation, and linear dimensions of a simple 3D object from the 2D line-of-sight projections of curves on its surface when these projections are given for two poses of the object.
Date of Conference: September 14-17, 2010
Track: Adaptive Optics & Imaging