A morphometric and densitometric approach to image interpretation

In this article, an approach is proposed for the definition of a model suitable for digital image interpretation. This model considers the densitometric, morphometric, and statistic characteristics peculiar to the images. The morphometric features are related to geometric and topologic properties and euclidean relations (shapes, distances, measures, convexity, ...); the densitometric properties are dependent on texture; statistics are mainly used to compare unknown images with paradigmatic data. All the characteristics are then integrated in a descriptive general reference model. A knowledge base which contains both data strictly referred to the examined images and data belonging to non-image domains is also considered for the classification and final image diagnosis.