Distributed Image Processing

Abstract

Image processing is usually approached as the problem of applying numerical algorithms to digitized gray-scale images and may be very expensive in terms of memory space and computation time. Therefore, it is worthwhile to look for more economical solutions. An analysis of typical image-processing tasks reveals that the usual costly approach can often be avoided or, in certain cases, may not even be appropriate. These considerations led us to the design and development of a special “Picture Processing Display Terminal” (PPDT) which has been devised for performing most of the required pre-processing and feature-extraction steps at the periphery of the computer at a small fraction of the costs that arise when the same processing is carried out in the computer. Hence the input to the computer already consists of binary representations of significant features, thus considerably reducing the amount of data to be further processed. Furthermore, each preprocessing step, which may be performed either with the man in the loop or automatically, is made visible in order to provide the feedback and the opportunity to evaluate the performance of the system and correct mistakes. The PPDT is part of a system for “distributed image processing”. The considerations leading to the design of such a system are discussed in this paper, as well as its possibilities and limitations.