Adaptive regularization in image restoration using evolutionary programming

Image restoration is in general a difficult problem due to the ill-conditioned nature of the associated inverse filtering operation. Adaptive regularization techniques are usually employed to alleviate this situation. The authors employ evolutionary programming to solve this adaptive regularization problem by generating a population of potential regularization strategies and allow them to compete under the k-pdf error measure, which is a novel characterization of image quality. The population-based approach of EP provides an efficient search method for potential optimizers of this highly irregular and non-differentiable error measure. Most significantly, the very adoption of EP has allowed them to broaden the range of possible cost functions for image processing, so that one can choose the most relevant function rather than the most tractable one for a particular application.