Opto-Electronic Engineering, Volume. 38, Issue 7, 86(2011)
Parallel Connected Component Detection Algorithm for Multi-core Based on BFS
The speed of general connected component detection algorithms was slow, and most of these algorithms needed more than one scanning. A connected component detection algorithm was presented based on Breadth First Search(BFS) with the First In First Out (FIFO) queue. The algorithm was a non-recursive algorithm, the connected component could be detected by one scanning, and the storage space and running time could be reduced. We used a particular scanning template to process each pixel, avoiding more than one scanning. The multi-core parallel processing was used to accelerate the algorithm, and it realized the truly parallel computing. With the connected regions ordered from top-left to bottom-right, a reverse merging method was proposed to simplify the complexity of region merging. The experimental results show that the detection rate has been greatly improved.
Get Citation
Copy Citation Text
ZHOU Lian-ling, YE Yu-tang, LIU Lin, ZHANG Jing, XIE Yu, SUN Qiang, YAO Jiao. Parallel Connected Component Detection Algorithm for Multi-core Based on BFS[J]. Opto-Electronic Engineering, 2011, 38(7): 86
Category:
Received: Mar. 7, 2011
Accepted: --
Published Online: Aug. 10, 2011
The Author Email: Lian-ling ZHOU (zlianling@gmail.com)
CSTR:32186.14.