Parallel Watershed Algorithm Based on Sequential Scanning

摘要: The watershed transformation is widely used in morphological image segmentation [1], e.g., industrial, and biomedical applications, where digital pictures of 512 x pixels, 1K 1K, or even larger are not uncommon. Therefore, large amount data and, consequently, complex analysis, entail parallel algorithms. In this paper, a fast SPMD (Single Program Multiple Data) algorithm based on sequential scannings rendered. task performed by the present to compute integrating gradient original image. technique, iterative nature, implemented multitransputer system (with Bruijn interconnection network) repeated scans message passing among processors until computation stabilizes. well suited for SIMD Instruction computers since no ordered queues (see [1,4]) used. Speedup evaluates quality algorithm.