A Tutorial on Multi-frame Computational Super-resolution using Statistical Methods

The term super-resolution is used to describe meth-ods aimed at recovering detail information of an imaged scenethat would otherwise be lost during the normal imaging process.Here we discuss only digital cameras that process light reflectedby the objects constituting a scene. A digital camera, consistingof optics a digital imaging sensor and signal processing hardwareis able to capture a digital image of a scene. For digital cameras,resolution has to do with both the number of imaging elementsper unit sensor area – which translates into the number of pixelsper unit image area – as well as the information content of thedigital image. Intuitively speaking, increasing the informationcontent is directly related to increasing the number of pixelsper unit area in the digital image such as to properly renderthe enlarged information content according to the Shannonsampling theorem. This tutorial is concerned with increasingthe level of detail (i.e., information content) in a digital imageby means of statistical processing algorithms starting from aset of several (usually) alias-afflicted images of the same scene,acquired from slightly different positions. Such techniques fallin the category of multi-frame computational super-resolution.We start be describing the maximum likelihood (ML) solutionto this problem and then show how a maximum a posteriori(MAP) approach can improve upon the ML solution. We willdiscuss several solution strategies relying on such principles andpoint to their advantages and disadvantages. We conclude witha short overview of alternative super-resolution approaches.