Input/output self-tuning algorithms applied to image processing

The paper discusses two dimensional self-tuning signal processing algorithms. The methods are, in essence, extensions of self-tuning methods found in control engineering applications and have the distinctive feature that they use two-dimensional (2-D) transfer function representations. They thus constitute an input-output approach to prediction and filtering which complements the state-space algorithms already available. The aim of this paper is to summarise previously published results on 2-D self-tuning predictors and show how similar ideas can be applied to give self-tuning de-blurring methods based on 2-D transfer function results.