7 November 2016 Gaussian total variation blind restoration of ground-based space object imagery
Author Affiliations +
Proceedings Volume 10141, Selected Papers of the Chinese Society for Optical Engineering Conferences held July 2016; 1014104 (2016) https://doi.org/10.1117/12.2251735
Event: Selected Proceedings of the Chinese Society for Optical Engineering Conferences held July 2016, 2016, Changchun, China
Abstract
We focus on the restoration of ground-based space object adaptive optics (AO) images distorted by atmospheric turbulence. A total variation (TV) blind AO images restoration method taking advantage of low-order Gaussian derivative operators is presented. Unlike previous definition of the TV regularization term, we propose to define the TV prior by the Gaussian gradient operators instead of the general finite-difference gradient operators. Specifically, in each iterative step of alternating minimization when solving the TV blind deconvolution problem, the first-order Gaussian derivative operator (i.e. gradient magnitude of Gaussian) is used to construct the total variation norm of object image, and the secondorder Gaussian derivative operator (i.e. Laplacian of Gaussian) is used to spatially adjust the regularization parameter. Comparative simulation experiments show that this simple improvement is much practicable for ground-based space object images and can provide more robust performance on both restoration accuracy and convergence property.
© (2016) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Shiping Guo, Shiping Guo, Rongzhi Zhang, Rongzhi Zhang, Rong Xu, Rong Xu, Changhai Liu, Changhai Liu, Jisheng Li, Jisheng Li, } "Gaussian total variation blind restoration of ground-based space object imagery", Proc. SPIE 10141, Selected Papers of the Chinese Society for Optical Engineering Conferences held July 2016, 1014104 (7 November 2016); doi: 10.1117/12.2251735; https://doi.org/10.1117/12.2251735
PROCEEDINGS
9 PAGES


SHARE
Back to Top