A Parallel IMAGE Processing Server for Distributed Applications

The use of parallel libraries for image processing is a common practice in the implementation of monolithic applications. Nowadays there is an increasing interest in moving towards distributed and heterogeneous applications also in the image processing community. The computational Grid may represent an adequate middleware support and infrastructure at this purpose. A problem of interest in this context is the interoperability and reuse of available parallel libraries, possibly through a dynamic interaction model that permits to accommodate request arriving from heterogeneous clients in a distributed environment. In this paper we present PIMA(GE), a Parallel IMAGE processing GEnova server obtained wrapping up a legacy code parallel library. The parallel server is implemented using CORBA and may be integrated in Grid architecture. The PIMA(GE) server organization and its interaction with user requests are discussed. Early experimental results are presented.