PROJECT TITLE :

Parallel Branch-Cut Algorithm Based on Simulated Annealing for Large-Scale Phase Unwrapping

ABSTRACT:

2-dimensional phase unwrapping may be a key step within the phase extraction method, a picture-processing stage that's common to many totally different systems. Many varied approaches have been proposed over the past many decades. However, with the growth of image scale, it poses new challenges in terms of computational and memory requirements to phase unwrapping that need a global approach to get smart results. Due to solely a single method utilized in most previous algorithm implementations, it becomes additional problematic to unwrapping when the required computing resources exceed the aptitude of one laptop. Meanwhile, with the event and application of supercomputer techniques, high-performance computing is rising as a promising platform for scientific applications. During this paper, a completely unique hybrid multiprocessing and multithreading algorithm is proposed in order to overcome the problem of unwrapping large knowledge sets. In this algorithm, we tend to improve on Goldstein's branch-cut algorithm using simulated annealing plan to additional optimize the set of branch cuts in parallel. For giant information sets, the tiling strategy based on the character of parallel computing guarantees the globality of section unwrapping and avoids massive-scale errors introduced. Using real and simulated interferometric knowledge, we tend to demonstrate that our algorithms are highly competitive with other existing algorithms in speed and accuracy. We tend to also demonstrate that the proposed algorithm will be efficiently parallelized and performed across nodes in an exceedingly high-performance computing cluster.