Generation of chain-coded contours and contours inclusion relationship under multiprocessor environment

A new algorithm which operates in a distributed processing environment for extracting 8-direction chain-coded contours and determining contours inclusion relationship from a binary-valued image is described. The image is divided into regions, with any two regions having either one row or no pixel in common. Regions are concurrently processed by WORKER processes. Partial results from these WORKERs are then fed to a COMBINER process to generate the overall results. The inclusion relationship among contours can be correctly determined by sending partial results in a specific order. Simulation on a transputer system indicates a speedup of five times for six WORKERs and six times for eight WORKERs