Blind identification of degradations using ascendant hierarchical classification

Filtering and restoration aim to improve the quality of an image to facilitate its interpretation. Most of these algorithms require a priori information about the degradation affecting the image. In the context of blind image processing, when no a priori information is available, the nature of the degradation has to be identified from the observed image. This operation is essential to select the algorithm adapted to the image. Noise, blur and combinations of both are the sources of degradation considered here. We present a blind identification procedure which involves the computation of local statistics and the execution of a non parametric ascendant hierarchical classification algorithm.