Solving the Region Growing Problem on the Connection Machine

Real time VLSI implementation of a fast split and merge segmentation algorithm

“…al. showed that random selection is more efficient than the selection with smaller ID [5]. Xuexing et.al.…”

“…have been chosen for splitting as well as merging the regions. For comparison of our result with other relevant algorithms [4], [5], [7] we have chosen the difference between the minimum and maximum values of pixels of adjacent regions. Thus a region is represented by minimum and maximum intensity values within that region.…”

“…For merging; a region selects its neighbor for merging with minimum weight between their edges. For multiple edges with minimum weight, the tie is broken either by selecting the neighbor with smaller (or larger) ID or random selection [4] [5]. Two regions actually merge if both of them select each other for merging.…”

“…Initially it was proposed to select the neighbor with lowest ID to be merged among all the eligible neighbors [4]. Copty et al proposes a random neighbor selection to break the tie between multiple neighbors satisfying Best Merge Criterion (BMC) [5]. Most of the time efficient approach of recursive classical SM algorithm deals with data structure arrangement in software domain [6].…”