ABSTRACT
We will again restrict the presentation of this chapter to the popular linear image degradation model:
g = Hfˆ + n (7.1)
where g, f and n are the lexicographically ordered degraded image, original image and additive white Gaussian noise (AWGN), respectively [21, 145]. H is the linear distortion operator determined by the point spread function (PSF), h. Blind image deconvolution is an inverse problem of rendering the best estimates, fˆ and hˆ, to the original image and the blur based on the degradation model. It is a difficult ill-posed problem as the uniqueness and stability of the solution is not guaranteed [24]. Classical restorations require complete knowledge of the blur to be known prior to restoration [21, 23, 145] as discussed in the previous chapters. However, it is often too costly, cumbersome or, in some cases, impossible to determine the exact blur a priori. These could be due to various practical constraints such as the difficulty of characterizing air turbulence in aerial imaging, or the potential health hazard of employing a stronger incident beam to improve the image quality in X-ray imaging. In these circumstances, blind image deconvolutions are essential in recovering visual clarity from the degraded images.
