Low-Resolution Image Formation Models

Abstract

As already explained in the previous chapter, our aim in this monograph is to follow a Bayesian formulation in providing solutions to the SR problem. In this chapter we describe the models used in the literature for obtaining the observed LR images from theHR-targeted source image. That is, we look into the details of the building blocks of the system in Fig. 1.2. The analysis will result in the determination of P(o|f_k, d) required by the Bayesian formulation of the SR problem (see Eq. (2.3)). We include both cases of recovering an HR static image and an HR image frame from a sequence of images capturing a dynamic scene. As we have already mentioned in the previous chapters, the LR sequence may be compressed. We therefore first describe the case of no compression and then extend the formation models to include compression.