MuSIP-multisensor image processor (a system for the automated fusion and analysis of multisource images)

As the volume of data collected by an increasing number of satellites escalates, it has become of paramount importance that methods of coping with and making best use of this information are developed. Current trends are moving towards automatic and knowledge-based analysis systems particularly for monitoring applications where the same kinds of interpretation task must be repeated many times. The MuSIP project, part funded by the European ESPRIT programme for research information, aims to develop a software demonstrator for the automatic fusion and analysis of multisource data. The project recognises this need across all computer vision applications and so will test the demonstrator on two dissimilar applications, one in the field of remote sensing and one in the field of medical image analysis. The target remote sensing application is the monitoring of forestry using a dataset of radar and optical imagery, map and ground data. This paper introduces the modules that comprise the demonstrator that is: a supervisor with its associated knowledge bases, algorithms and the databases. Attention is focussed in particular on the role of knowledge bases and the methods of data fusion.<
>