A GNC ALGORITHM FOR CONSTRAINED IMAGE-RECONSTRUCTION WITH CONTINUOUS-VALUED LINE PROCESSES

Image reconstruction is formulated as the problem of minimizing a non- convex functional F(f) in which the smoothness stabilizer implicitly r efers to a continuous-valued line process. Typical functionals propose d in the literature are considered. The minimum of F(f) is computed us ing a GNC algorithm that employs a sequence F(p)(f) of approximating f unctionals for F(f), to be minimized in turn by gradient descent techn iques. The results of a simulation evidence that GNC algorithms are co mputationally more efficient than simulated annealing algorithms, even when the latter are implemented in a simplified form. A comparison be tween the performance of these functionals and that of a functional th at refers to an implicit binary line process is also carried out; this shows that assuming a continuous-valued line process gives a better r econstruction of the smooth, planar or quadratic regions of the image, even with first-order models.