Chris Jones: Change Detection

Change Detection with Geographical Data

Funding Source: NERC grant GR3/10569: "Automated Environmental Change Detection Using Spatially Referenced Digital Datasets"

Summary of Research

This research was concerned with developing automated techniques to assist in monitoring change in the natural and man-made environment. The research addressed the specific issue of identifying significant differences between digital map datasets, that can be attributed to change in the real-world phenomena, as opposed to differences that may arise due to error and uncertainty in the methods used to acquire and digitise the data. Techniques have been developed for conflating a pair of time-differing area-class polygon maps in a way that retains differences expected to be due to real-world change while merging nearby features that differ only due to error in the respective source datasets. Bayesian multivariate statistical methods are used to attach probabilities of equivalence to matched polygon boundary segments and probabilities of change to regions for which the feature attributes differ between the source maps. The maps used for experimentation are based on land cover interpretation of aerial photography of Scotland.

Publications on change detection:

Jones C.B, J.M. Ware and D.R. Miller (2000) "Bayesian Probabilistic Methods for Change Detection with Area-Class Maps", Proceedings Accuracy2000, Amsterdam, Proceedings Accuracy2000, Amsterdam, Delft University Press, 329-336.

Jones C.B. and J.M. Ware (1999) "A Probabilistic Approach to Environmental Change Detection with Area-Class Map Data", Integrated Spatial Databases: Digital Images and GIS, ISD'99, Lecture Notes in Computer Science 1737, Springer Verlag, pp 122-136; ISSN 0302-9743. [PDF]

Ware, J.M. and Jones, C.B. (1998), "Matching and Aligning Features in Overlayed Coverages", Proc. of 6th International Symposium on Advances in Geographical Information Systems (ACM-GIS'98), pp 28-33. [PDF]