Title :
s-way parallel finite time Gibbs classification
Author :
Greenshields, I.R. ; Yang, Zhihong
Author_Institution :
Dept. of Comput. Sci. & Eng., Connecticut Univ., Storrs, CT, USA
Abstract :
Gibbs classification is usually performed by annealing relative to the Metropolis scheme. When the number of compute cycles is bounded, Azencott demonstrated that parallel annealing is a viable approach. Here we describe an s-way finite time schedule with a post-annealing merge strategy
Keywords :
image classification; medical image processing; simulated annealing; Metropolis scheme; Visible Human dataset; annealing; compute cycles; parallel annealing; post-annealing merge strategy; s-way parallel finite time Gibbs classification; Annealing; Computed tomography; Computer science; Concurrent computing; Context modeling; Image converters; Image processing; Iterative algorithms; Processor scheduling; Scheduling algorithm;
Conference_Titel :
Bioengineering Conference, 2001. Proceedings of the IEEE 27th Annual Northeast
Conference_Location :
Storrs, CT
Print_ISBN :
0-7803-6717-0
DOI :
10.1109/NEBC.2001.924726
