DocumentCode :
1383731
Title :
Parallelizable Bayesian tomography algorithms with rapid, guaranteed convergence
Author :
Zheng, Jun ; Saquib, Suhail S. ; Sauer, Ken ; Bouman, Charles A.
Author_Institution :
Delphi Delco Electron. Syst., Kokomo, IN, USA
Volume :
9
Issue :
10
fYear :
2000
fDate :
10/1/2000 12:00:00 AM
Firstpage :
1745
Lastpage :
1759
Abstract :
Bayesian tomographic reconstruction algorithms generally require the efficient optimization of a functional of many variables. In this setting, as well as in many other optimization tasks, functional substitution (FS) has been widely applied to simplify each step of the iterative process. The function to be minimized is replaced locally by an approximation having a more easily manipulated form, e.g., quadratic, but which maintains sufficient similarity to descend the true functional while computing only the substitute. We provide two new applications of FS methods in iterative coordinate descent for Bayesian tomography. The first is a modification of our coordinate descent algorithm with one-dimensional (1-D) Newton-Raphson approximations to an alternative quadratic which allows convergence to be proven easily. In simulations, we find essentially no difference in convergence speed between the two techniques. We also present a new algorithm which exploits the FS method to allow parallel updates of arbitrary sets of pixels using computations similar to iterative coordinate descent. The theoretical potential speed up of parallel implementations is nearly linear with the number of processors if communication costs are neglected
Keywords :
Bayes methods; Newton-Raphson method; approximation theory; computerised tomography; convergence of numerical methods; image reconstruction; iterative methods; parallel algorithms; 1D Newton-Raphson approximations; Bayesian tomographic reconstruction algorithms; communication costs; convergence speed; coordinate descent algorithm; functional optimization; functional substitution; iterative coordinate descent; iterative process; parallel updates; parallelizable Bayesian tomography algorithms; pixels; rapid guaranteed convergence; simulations; Bayesian methods; Computational modeling; Concurrent computing; Convergence; Image reconstruction; Iterative algorithms; Iterative methods; Reconstruction algorithms; Subspace constraints; Tomography;
fLanguage :
English
Journal_Title :
Image Processing, IEEE Transactions on
Publisher :
ieee
ISSN :
1057-7149
Type :
jour
DOI :
10.1109/83.869186
Filename :
869186
Link To Document :
بازگشت