1 Introduction

All astronomical images observed from the ground are subject to degradations of various kinds, instrumental or atmospheric, that cancel out part of the information. In recent years many methods of image restoration have been elaborated. However, such methods are very expensive in computing time and allow only a partial recovery of the information. Therefore it is convenient, before any numerical treatment of a series of images, to select the best ones (Bonet 1999). In solar astronomy a good indicator of image quality is granular contrast, given by the standard deviation of the image normalized to the average (Ricort et al. 1981; Collados & Vázquez 1987). Such an indicator allows rapid estimates, but it is not very sensitive to the presence of structures of small dimensions (a result known as Parseval's theorem), while we are often interested in the study and analysis of flux-tubes or other small magnetic structures (del Toro et al. 1990).

In this letter I propose a new method for the analysis of image quality that is sensitive to the dimensions of the structures present. The method is applicable to images of solar granulation but could, in principle, be applied in other astronomical contexts. One of the principal characteristics of this method is that it allows us to select the most meaningful zones of the image to provide a quality map: i.e. it allows us to distinguish, within the same image, zones that are more or less degraded. This might allow us, when searching for details in the image, to restrict the use of the segmentation algorithms that trace the contours of these structures only to good-quality zones. Besides, when we are interested in the reconstruction of the image, the quality map helps us to find the most suitable restoring strategy for each image zone given its quality.