ABSTRACT
Many image reconstruction techniques have been remarkably successful in applications such as medical imaging, remote sensing, nondestructive testing, and radar. The focus here is on a general problem that arises in many such applications: the reconstruction of a compactly supported function f from a limited number of measurements such as its image values. In reconstructing a function f from its finitely-many linear functional values, noisy or not, it is not possible to specify f uniquely. Estimates based on finite data may succeed in recovering broad features of f , but may fail to resolve important detail. While there has been considerable effort in the development of inverse or estimation algorithms, this problem has generally proved difficult.
