A parallel any-time control algorithm for image understanding

This paper presents a problem-independent control algorithm for image understanding based on a semantic network formalism for knowledge representation. The algorithm focusses on the achievement of any-time-characteristics, which are important for real world applications. For that purpose, knowledge based image understanding is treated as an optimization problem and solved by iterative combinatorial optimization procedures, like e.g. genetic algorithms. Parallel processing of knowledge and the parallelization of optimization algorithms are shown to provide an efficient approach for an improved any-time-behaviour