A hierarchical approach to image and blue identification

Summary form only given. In image restoration the point-spread function (PSF) of the degrading system, as well as the variance of the observation noise and a model of the original image, usually has to be estimated from the noisy blurred images, g₀, themselves. When the blur identification problem is formulated as a maximum-likelihood estimation problem, the optimization of the resulting likelihood function is a highly complicated and nonlinear problem. Various strategies that indirectly solve this optimization problem, such as recursive and gradient-based algorithms or methods based on the expectation-maximization (EM) algorithm, have been proposed. All suffer from the fact that the computation involved is considerable and the identification algorithm may converge to a suboptimal solution, which is typically useless in image restoration. A hierarchical identification approach is proposed to tackle these problems