A symbolic approach to the analysis of mixed data types found in the spatial database of a Geographical Information System is developed. Unlike statistical techniques, that are essentially based on the manipulation of numeric quantities, the method here attempts to emulate photo-interpretation by manipulating symbolic representations of photo-interpretive knowledge. This representation consists of a mixture of normal and ‘defeasible’ rules. The latter enables analysis to proceed in the face of uncertainty by making (and possibly revising) assumptions about the contents of the scene being analysed. The analysis procedure is decomposed into three stages. The first deals with the analysis of data from groups of sensors (sources) providing the same type of information, the second combines the inferences of these individual analyses and the third imposes spatial consistency constraints. This decomposition implies a structuring of the knowledge, an issue that appears essential for a knowledge-based analysis to be efficient. Examples are presented which demonstrate the system's ability to handle mixed multi-spectral and radar imagery, and images at different spatial resolutions.

A simple modification to existing probabilistic relaxation procedures is suggested which allows the information contained in initial labels to exert an influence on the direction of relaxation throughout the process. In this manner, the initial labels assume more importance than with conventional algorithms and are used in combination with the outcome of relaxation at each iteration to produce a cooperative estimate of the correct label for a particular object. Pixel labeling examples are presented which show the performance that can be obtained with the modified algorithm. The procedure is readily generalized to allow other data to influence the process.