Y. Park et M. Soumekh, IMAGE-RECONSTRUCTION SYNTHESIS FROM NONUNIFORM DATA AND ZERO THRESHOLD CROSSINGS, Optical engineering, 33(10), 1994, pp. 3290-3301
We address the problem of reconstructing functions from their nonunifo
rm data and zero/threshold crossings. We introduce a deterministic pro
cess via the Gram-Schmidt orthonormalization procedure to reconstruct
functions from their nonuniform data and zero/threshold crossings. Thi
s is achieved by first introducing the nonorthogonal basis functions i
n a chosen 2-D domain (e.g., for a band-limited signal, a possible cho
ice is the 2-D Fourier domain of the image) that span the signal subsp
ace of the nonuniform data. We then use the Gram-Schmidt procedure to
construct a set of orthogonal basis functions that span the linear sig
nal subspace defined by the nonorthogonal basis functions. Next, we pr
oject the N-dimensional measurement vector (N is the number of nonunif
orm data or threshold crossings) onto the newly constructed orthogonal
basis functions. Finally, the function at any point can be reconstruc
ted by projecting the representation with respect to the newly constru
cted orthonormal basis functions onto the reconstruction basis functio
ns that span the signal subspace of the evenly spaced sampled data. Th
e reconstructed signal gives the minimum mean square error estimate of
the original signal. This procedure gives error-free reconstruction p
rovided that the nonorthogonal basis functions that span the signal su
bspace of the nonuniform data form a complete set in the signal subspa
ce of the original band-limited signal. We apply this algorithm to rec
onstruct functions from their unevenly spaced sampled data and zero cr
ossings and also apply it to solve the problem of synthesis of a 2-D b
and-limited function with the prescribed level crossings.