MATCHING THE RESOLUTION LEVEL TO SALIENT IMAGE FEATURES

This paper illustrates how strategies for structural description of im ages-based on the detection of salient features in the image-can be mo re efficiently realised by integrating them with the multiresolution a pproach to reduce details, up to the level at which such features are still recognisable. Since different structures of interest are visible in a restricted range of resolution levels, the combination of struct ural description with multi-resolution allows the automatic recognitio n of features at the minimum sufficient level, thus saving a costly de scription which would store irrelevant details. (C) 1997 Pattern Recog nition Society. Published by Elsevier Science Ltd.