Parallel iterative image restoration algorithms

Three parallel iterative image restoration algorithms with and without preconditioning are proposed and analyzed. The first algorithm corresponds to a coarse-grained general-purpose multiprocessor computer, the second to a massively parallel computer with synchronization, and the third to a massively parallel computer without synchronization. It is shown that the second and third algorithms can give a speed-up proportional to the number of processors when proper assumptions are satisfied, while the first one performs the best when simulated by a uniprocessor computer.<>