Abstract

A technique to uniformly decimate samples in a multidimensional low-pass bandlimited signal without information loss is presented. In many cases, the overall sampling density can be reduced below the Nyquist density¹ and ultimately to the hyper-area of the support of the function's spectrum. Two methods to regain the deleted samples are discussed. Both regain samples in forms of linear interpolations. The specific case of sampling images obtained through systems with circular pupils is addressed. An analysis of the noise sensitivity of the restoration is also presented.

© 1986 Optical Society of America