This paper presents a new technique for measuring the location of point images in an image-sampling system. Primary attention is given to the situation in which the image is under sampled. Such under sampling could be the result of data-storage or bandwidth limitations, and would introduce errors into standard methods of image location. The new technique is valid for under-sampled as well as adequately sampled images, and consists of a linear-filtering procedure of the same complexity required by standard location-measurement techniques. The problem is formulated as one of statistical-estimation theory, the optimum-location procedures being derived for an additive-gaussian-noise environment, and for conditions of known and unknown image irradiance.
S. J. Kishner and T. W. Barnard, "Point-source position estimation in under-sampled optical systems," J. Opt. Soc. Am. 63, 1578-1583 (1973)