Context: a new paradigm to control distributed perceptual systems

This paper deals with problems of representation and handling of concurrent processes in multiprocessor machines or in distributed and co-operating systems oriented to image analysis. For this purpose a new synchronization mechanism, named context is presented. Contexts are introduced as object variables in pictorial languages to represent distributed computation on spatial data. In particular, details of its implementation on the PIctorial C Language (PICL) are given.