Fast Image Labeling using Local Operators on Mesh-Connected Computers

Mesh-connected computers are well suited to local computations that exploit the nearest-neighbor interconnections of the mesh. In this correspondence, a new parallel algorithm is proposed for fast image labeling using local operators on image pixels. The algorithm can be implemented on an n x n mesh-connected computer such that, for any integer k in the range [1, log(2/i)], the algorithm requires Q(kn^llk)bits of local memory per processor and takes 0(foi) time. Bit-serial processors and communication links can be used without affecting the asymptotic time complexity of the algorithm. The time complexity of the algorithm has very small leading constant factors, which makes it superior to previous mesh computer labeling algorithms for most practical image sizes (e.g., up to 4096 x 4096 images). Furthermore, our algorithm is based on using stacks that can be realized using very fast shift registers within each PE.