Publisher's Synopsis
Describing a computer vision environment which supports course granularity parallelism on transputer-based systems, this study emphasizes a system which treats the organization of the image data and its representation with equal measure to the visual processes operating in parallel. The system, called Vis a Vis, provides the dynamic creation of images and attendant data structures, each of which is linked to provide an explicit representation of the inter-dependence of the visual cues. As a vision system, Vis a Vis offers the required functionality of 3-D image understanding, from image acquisition, through detection and analysis of intensity discontinuities, to 3-D model construction and object recognition.;Throughout the text, authors emphasize system integration of information and computer architectures, rather than purely theoretical and algorithmic issues.
