Parallelizing a multi-frame blind deconvolution algorithm on clusters of multicore processors

Linderman, Richard; Spetka, Scott; Emeny, Susan; Fitzgerald, Dennis

doi:10.1109/aero.2009.4839545

Cited by 2 publications

(2 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Related research in parallelization of MFBD so far has focused on parallelization of MFBD algorithm on large cluster architectures, such as the work on physically constrained iterative deconvolution (PCID) algorithm [16,17], implemented on a 1280-node cluster. However, there has been little focus on the implementation of the algorithm on massively parallel architectures (MPA) of graphical processing units (GPU), where a small form-factor and lowcost implementation is possible.…”

Section: Introductionmentioning

confidence: 99%

“…Previous parallelization techniques employ multiple CPUs to parallelize processing of multiple frames or tiles. Processing multiple frames separately has been implemented on a cluster network architecture [16]. The approach to process multiple tiles relies on segmenting the images into a number of smaller images, each of which can be processed in parallel on multiple cores or nodes on a cluster.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Parallel hybrid bispectrum-multi-frame blind deconvolution image reconstruction technique

Hajmohammadi

Nooshabadi

Archer

et al. 2016

J Real-Time Image Proc

View full text Add to dashboard Cite

This paper presents B-MFBD, a parallel hybrid of bispectrum speckle imaging (SI) and multi-frame blind deconvolution (MFBD) image reconstruction techniques for anisoplanatic, long horizontal path imaging. Our aimis to recover an enhanced version of a turbulence-corrupted image by massive parallelization of an SI and MFBD algorithms. The bispectrum SI technique is used in place of the multi-frame ensemble averaging to initialize the iterative parallel MFBD algorithm. B-MFBD technique, through massive parallelization, provides significantly large improvement in execution speed to both the bispectrum SI and MFBD parts of the hybrid algorithm. We report 85 % improvement in processing time with respect to the sequential implementation of the same algorithm for a 256 Â 256, gray-scale image, with comparable improvement in image quality.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%