Region based image content descriptors and representation

This paper addresses some of the difficulties involved in the automatic generation of image content descriptors within a specific domain of airborne reconnaissance. The initial stages of the work concentrated on the development/collation of a set of feature detectors covering points (corners), lines (edges, ridges) and regions which operate on a variety of image sensors (e.g. SPOT, IRLS-Infra-red Linescan, dd5-Russian satellite data). We also consider the questions of combining several feature detectors to improve their performance and reliability on a wider variety of images. The descriptors once generated and represented in some suitable way could be used for the indexing, matching and searching as well as retrieval of data in an image database. These operations on the image database depend on a fundamentally reliable detection stage. For this reason the initial objective of our work was not to depend purely upon just one algorithm but to have several algorithms, the output of which may then be combined in a novel manner into a composite estimate of the feature(s) that are required.