Paper
24 November 2023 Hybrid MPI-CUDA accelerated multi-frame blind deconvolution algorithm
Author Affiliations +
Proceedings Volume 12935, Fourteenth International Conference on Information Optics and Photonics (CIOP 2023); 129352B (2023) https://doi.org/10.1117/12.3007580
Event: Fourteenth International Conference on Information Optics and Photonics (CIOP 2023), 2023, Xi’an, China
Abstract
Multi-Frame Blind Deconvolution (MFBD) algorithm is the mainstream post-processing method for atmospheric turbulence degraded images. The restoration of a single object from multiple images requires several minutes or even tens of minutes, which severely limits MFBD’s practical application. To achieve the goal of fast restoration, this paper proposes hybrid MPI-CUDA (Message Passing Interface-Compute Unified Device Architecture) accelerated MFBD algorithm. In this hybrid programming model, MPI is responsible for assigning computing tasks, and CUDA is responsible for parallel computing. Hybrid MPI-CUDA accelerated MFBD algorithm has been tested on simulated images and achieved satisfactory results. The execution speed for restoring a single high-resolution image from multiple blurred images has been increased by over 50-fold.
(2023) Published by SPIE. Downloading of the abstract is permitted for personal use only.
Zongxu Chen, Xenwen Liu, and Yonghui Liang "Hybrid MPI-CUDA accelerated multi-frame blind deconvolution algorithm", Proc. SPIE 12935, Fourteenth International Conference on Information Optics and Photonics (CIOP 2023), 129352B (24 November 2023); https://doi.org/10.1117/12.3007580
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Adaptive optics

Point spread functions

Blind deconvolution

Mathematical optimization

Parallel computing

Back to Top