We derive and discuss an algorithm for separating measurement- and nul
l-space components of an object with respect to a given imaging system
. This algorithm requires a discrete-to-discrete approximation of a ty
pically continuous-to-discrete imaging system, and problems associated
with such an approximation are examined. Situations where knowledge o
f the measurement and null spaces is crucial for analyzing imaging sys
tems are discussed. We conclude with two examples demonstrating the us
efulness of this algorithm, even in the discrete-to-discrete approxima
tion. (C) 1997 Optical Society of America.