ACTIVE CONTOUR SEGMENTATION FOR THE IDENTIFICATION OF METALLOGRAPHIC AND MORPHOLOGICAL ELEMENTS OF INTEREST IN DUCTILE CAST IRON
AbstractDuctile cast irons are characterized by a wide range of mechanical properties that depend on graphite elements morphology and microstructure properties. Both chemical composition and manufacturing conditions control matrix microstructure, and ferritic, pearlitic, ferritic-pearlitic, martensitic, bainitic, austenitic and austempered ductile irons can be obtained. Considering crack propagation resistance of ductile cast irons, their peculiar behaviour is due to the graphite elements shape, that is approximately spheroidal. Due to their morphology, graphite elements can act as crack arresters: as a consequence ductile cast irons are characterized by high ductility and toughness values and can be used for loading conditions that could be considered as critical for other cast irons types (e.g. fatigue loading conditions). Up to some years ago, microstructure analysis was mainly performed by means of semi-quantitative procedures applied to metallographically prepared specimens, with the characteristics evaluation that was mainly based on the operator expertise. Only recently there has been an increasing interest in numerical procedures of image analysis for quantitative evaluation of materials. In this work the problem of the estimation of the morphological parameters of elements such as graphite nodules, domains of chemical etching and metallic matrix has been taken up by a variational approach of image segmentation by active contours. Considering ductile irons, images obtained by means of a light optical microscope (LOM) on metallographically prepared specimens show both graphite elements (spheroids, nodules, lamaellas etc.) and microstructure elements (ferrite grains, pearlite lamaellas, etc.) and some artefacts due the preparing procedure that should be distinguished by more interesting elements. An automatic identification procedure is here proposed to distinguish the nodules from the metallic matrix and to evaluate the nodules shape parameters of interest and the composition of the metallic matrix (ferrite/pearlite volume fraction).