Title :
Simultaneous perturbation optimization for efficient image restoration
Author :
Maryak, J.L. ; Spall, James C.
Author_Institution :
Appl. Phys. Lab., Johns Hopkins Univ., Laurel, MD, USA
Abstract :
The Gibbs sampler of Geman and Geman (G&G) (1984) is a standard technique for image restoration. We implemented a similar estimation technique based on a new optimization method (simultaneous perturbation stochastic approximation (SPSA)). Preliminary tests of our results show comparable performance to the Gibbs sampler but with faster running times. This study suggests that SPSA is a strong candidate for use in image restoration.
Keywords :
estimation theory; image restoration; optimisation; stochastic processes; Gibbs sampler; image restoration; simultaneous perturbation optimization; simultaneous perturbation stochastic approximation; Annealing; Bayesian methods; Degradation; Image processing; Image restoration; Image sampling; Optimization methods; Pixel; Shape control; Stochastic processes;
Journal_Title :
Aerospace and Electronic Systems, IEEE Transactions on
DOI :
10.1109/TAES.2005.1413767
