Comparative Analysis of Spectral and Spatial Features for Classification of Graphite Grains in Cast Iron

The morphology and the distribution of graphite grains are the decisive factors in judging the properties of the material cast iron. There are six classes of graphite grain morphology defined by ISO-945 through reference drawings for cast iron graphite grain classification. These reference drawings are universally accepted for classification of graphite grains. Many shape representations and retrieval methods exist. Among them, methods based on Fourier descriptors (FD) achieve acceptable results in classification compared to other methods. Different shape signatures have been exploited to derive FDs, however, FDs derived from different signatures can have significant effect on the result of classification [17]. In this paper, a performance analysis of classification of graphite grains using spectral and spatial features is performed. The neural network classifier based on radial basis function has been employed for classification. The experimentation is carried out using the metallographic images from the well known microstructures library [6]. For training and testing the network, the grain shapes identified in ISO-945 reference drawings and the grain classification by the experts are used. The FDs derived from centroid distance function and neural network classifier with radial basis function yield better classification results for graphite grains.