In thermal infrared astronomy a differential technique is used for extracti
ng the weak astronomical image from the large background due to the environ
ment. The resulting images, the so-called chopped and nodded images, presen
t large negative values and provide reliable representations uniquely in th
e case of isolated sources. In this paper we discuss the mathematical prope
rties of the problem of restoring the original image from its chopped and n
odded version and we investigate an iterative regularization method, the so
-called projected Landweber method, for approximating non-negative solution
s of the problem. This method provides satisfactory results in some cases b
ut, in general, the restored images are still affected by several artifacts
. We suggest an observation strategy which, combined with our proposed rest
oration method, can significantly reduce the effect of these artifacts.