Image mining utilising multitemporal image fusion based on map algebras and stereology to derive risk estimates and rates of change

This article extends the notion of image mining with map algebras and stereology. It considers natural or man-made spatial processes observable at the earth surface. Image mining addresses the chain from object identification from remotesensing images through modelling, tracking on a series of images in time, prediction, towards communication to stakeholders. Image mining may benefit from image fusion during the object identification step and the tracking step. A map algebra when applied as a quantitative combination of different images, e.g. created at different times, may give cumulative information in space and time. Stereology allows us to assess basic properties from a series of images in a statistical way. Stereology is considered at the object level where the size of an increasing and decreasing object in space and time is addressed. In this article both are treated and applied from the decision-making perspective and they are illustrated with data on flooding. We conclude that map algebras and stereology are complementary in their role within an image-mining perspective.