A number of imaging systems are undersampled for a variety of reasons. For example, a user may want a wide field of view, but the imaging sensor used has a small size. Also, a system that has pixels corresponding to a small solid angle in object space may result in low SNR imagery. In order to compensate, a user may want to bin the pixels or use a lower f number optical system in order to obtain a higher SNR image. These systems will result in a phenomena referred to as aliasing. High spatial frequency components of the imagery will be aliased to lower spatial frequencies due to the way the image is sampled.
The Drizzle algorithm was developed (Williams4 et al. 1996; Fruchter & Hook 1998) in order to compensate for undersampling with the first set of CCD detectors used with the Hubble Space Telescope.
I have attached a PDF thatStochastic UpResolution.pdf discusses another method to remove aliasing with undersampled imagery. This is accomplished by processing a large number of images for which there are random image shifts. The processing method is similar to the Drizzle algorithm1. However, the details are different and its ability to eliminate aliasing is proved in the appendix of this paper.
