A. Sanfeliu et al. (Eds.): CIARP 2004, LNCS 3287, pp. 574–581, 2004. © Springer-Verlag Berlin Heidelberg 2004 Grey Level Image Components for Multi-scale Representation Giuliana Ramella and Gabriella Sanniti di Baja Istituto di Cibernetica E. Caianiello, CNR, Via Campi Flegrei 34, 80078, Pozzuoli (Naples), Italy (g.ramella,g.sannitidibaja)@cib.na.cnr.it Abstract. A method to identify grey level image components, suitable for multi-scale analysis, is presented. Generally, a single threshold is not sufficient to separate components, perceived as individual entities. Our process is based on iterated identification and removal of pixels, with different grey level values, causing merging of grey level components at the highest resolution level. A growing process is also performed to restore pixels far from the fusion area, so as to preserve as much as possible shape and size of the components. In this way, grey level components can be kept as separated also when lower resolu- tion representations are built, by means of a decimation process. Moreover, the information contents of the image, in terms of shape and relative size of the components, is preserved through lower resolution representations, compatibly with the resolution. 1 Introduction Grey level images are of large use in image analysis tasks. One of the main problems that have to be faced is image segmentation, necessary to distinguish foreground components from the background. The method to be used depends on problem do- main. The easiest way, unfortunately seldom effective, is to fix a threshold and to assign to the foreground all pixels with grey level larger than the threshold, and to the background all remaining pixels, e.g., [1]. The result of this process is, generally, an image with a number of components different from the expected one. Actually, the threshold should assume different values in different parts of the image, to allow cor- rect identification of foreground components. These methods are generally referred to as multi-threshold methods, e.g., [2,3]. An unwanted side effect of thresholding is that the size of foreground components is likely to be significantly reduced with respect to the perceived size. In fact, the same grey level distribution can characterise both the fusion area among foreground components perceived as individual entities, and (pe- ripheral) parts of foreground components, located far from the fusion area. More so- phisticated techniques, especially based on watershed transformation, generally pro- duce better result [4-6]. These techniques, however, need a complex preliminary phase to identify the sources for watershed segmentation, as well as an equally com- plex post-processing phase necessary to reduce the unavoidable over-segmentation. The method that we propose in this paper is a small addition to the literature on image segmentation. It is based on non-topological erosion and topological expan- sion, is computationally non expensive and in the experiments we have done per- formed quite well. Our method can be classified as a multi-threshold method. In fact, once the minimum and the maximum possible values of the threshold have been