Image analysis of multiphased ceramics

In this study, we focus on the recognition and characterization of the different phases present in a (Zr,Sn)TiO4 ceramic (with some amounts of La2O3 and NiO as sintering aids). Our work deals with the study of segmentation techniques that automatically extract each of the phases (ZST, TiO2, La2/3TiO3,...) from a scanning electronic microscope image of a polished section of the ceramic. The main difficulty encountered at this stage of the work is that their properties in terms of gray levels are very close together, thus, imposing the combination of simple image processing operators (such as thresholding) with advanced methods based on region growing processes. (C) 2001 Elsevier Science Ltd. All rights reserved.