A parallel algorithm for image information restoration

An image information restoration algorithm based on Long-Range correlation (LR-IIRA) can be efficiently applied in image interpretation, restoration, and error concealment. But this algorithm requires excessive computing time, which restricts its application on large-scale problems and real time problems. We propose a parallel LR-IIRA algorithm (PLR-IIRA) to parallelize the restoration processing on a Dawning 2000 cluster system. Meanwhile, a new distributed processing method (Scatter-Gather) is proposed for our PLA-IIRA. To reduce communication, a key position technique is used in our PLA-IIRA. From theory analysis and experimental results, it shows that the computing time of LR-IIRA can be reduced greatly on the cluster system, and PLA-IIRA has high speedup and efficiency. The scalability of PLA-IIRA is good. PLA-IIRA provides a new efficient way for real time application of LR-IIRA.