Euclidean Distance Transform of 3D Images on L.a.r.p.b.s. Model: an Efficient Parallel Algorithm

EDT of 2-D and 3-D images is one of the useful tools in various image processing algorithms. It is one of the basic operations in image processing and computer vision fields and essentially used in expanding, shrinking, thinning, segmentation, clustering and computing of images, object reconstruction, etc. It converts a binary image consisting of black and white voxels to a representation where each voxel has the Euclidean distance of the nearest black voxel. The objective of this work is to derive, implement and analyze an efficient time-optimal and scalable algorithm for EDT computation of 3-D images, on a recently proposed multiprocessor architecture, known as Linear Array with Reconfigurable Pipe-lined Bus System (LARPBS).