numero26

A. De Santis et alii, Frattura ed Integrità Strutturale, 26 (2013) 12-21 ; DOI: 10.3221/IGF-ESIS.26.02 14 The importance of the different damaging micromechanisms is influenced by the matrix microstructure, but it is worth to note that, considering ferritic-pearlitic ductile irons, matrix – graphite nodule debonding is not the most important damaging micromechanism. Focusing the role played by graphite elements in a completely different cast iron (a fully pearlitic flake cast iron), the flaky graphite tends to open in the middle and a void appears inside, considering both graphite elements orientated perpendicular and parallel to the loading direction [12]. The load increase implies an increase of the cast iron internal damage, with a graphite elements - pearlitic matrix debonding. This mechanism is more evident with the graphite elements oriented perpendicular to the loading direction. Furthermore, a microplasticity at graphite tips is observed in the pearlitic matrix (Fig. 4). Figure 4 : Damaging micromechanims in a fully pearlitic flake cast iron [12]. Considering the results in Fig.-1-4 and in Videos 1-3, it is evident the strong influence of the graphite nodules morphology on the damaging micromechanisms: the visual qualitative approach followed in the EN ISO standard [13] does not seem to be sufficient to fully characterize the graphite elements in a cast iron and therefore a quantitative approach seems to be necessary. Images obtained by means of LOM, despite a good visual appearance, are represented by a quite irregular signal due to various kind of degradations stemming from the acquisition process: additive noise, albedos due to dust and specimen oxidation, artifacts coming from scratches occurring during the specimen preparation. High performance image analysis procedure to distinguish the nodules from the background can be obtained within the framework of image segmentation: the original image is partitioned into disjoint domains where the signal has homogeneous characteristics, and passing from one domain to another these characteristics vary significantly [14]. For the cast iron (and ductile iron) classification it suffices to choose the class of piecewise constant functions to approximate the gray level distribution of the LOM images, so that in the segmented image the nodules are sharply enhanced over a uniform background. Then any standard software can quantify the graphite elements morphological parameters of interest; in particular, the Image Processing Toolbox of MatLab© provided a good performance. The segmentation problem can be solved by various techniques [15-16] with different pros and cons. The method of active contours [17] was preferred since can deal with the complex topology of the graphite elements without compromising the numerical complexity: as a consequence the cast iron metallographies can be reliably segmented and evaluated in real time. In this work, different graphite elements morphologies have been considered, ranging from lamellae to nodules, and image segmentation by the active contours method has been optimized in order to perform a quantitative analysis and characterization. A complete automatization of this approach allows to perform statistical analysis of many morphological parameters (e.g., graphite elements density, distribution, shape), allowing to fully characterize the investigated cast iron. I MAGE SEGMENTATION BY THE ACTIVE CONTOURS METHOD espite a visual inspection can distinguish the objects from the background on the available real data, the signal is quite irregular and does not feature a clear cut between the background and the different objects of interest in the picture. In Fig. 5-8, some examples of specimens are presented; as it can be noted, in each image there are some scratches and dust and, obviously, graphite elements like exploded graphite and flakes. It is important also to distinguish among the different kind of objects. D