David Gerwe (Boeing), Brad Eucker (Boeing), Carlos Luna (Boeing), Brandoch Calef (Boeing)
Keywords: Imaging
Abstract:
A generalization of diversity algorithms for image restoration and wavefront sensing is presented that allows joint processing of image sets in which the sampling geometry may be different for each frame. The sampling relations are assumed to be well approximated by affine transformations. This embraces translations, rotations, skew, and magnification, which may differ in x- and y-axis directions but not higher order distortions such as pincushion or as resulting from 3D parallax issues. This work extends superresolution processing with affine distortions to blind deconvolution and phase diversity in which the wavefront or PSF is jointly estimated with the object allowing these algorithms to be applied to collection engagement with 3D dynamic motion between the sensor and target. Simulations demonstrate the approach for ground-to-space and air-to-ground applications. A regularization approach is shown to balance image sharpening against noise amplification and handle spatial variations in effective sample density.
Date of Conference: September 12-15, 2007
Track: Imaging