We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Blind Image Deblurring via Regularization and Split Bregman.
- Authors
Su Xiao; Fangzhen Ge; Ying Zhou
- Abstract
Degradation caused by blurring is ubiquitous in digital images, and blind image deblurring (BID) has been proposed to solve this issue. Over the past several decades, various techniques and tools have been developed for BID problems, and continual efforts have been made to improve the speed and accuracy of sharp image estimation. This study proposes a new BID method, which incorporates a regularization technique, sparsity-inducing priors, and the split Bregman method. In the first phase, the proposed method equates blur estimation with a constrained minimization problem in which sparsity-inducing priors are employed to regularize the gradient image and blur the kernel. The split Bregman method is then applied to divide and conquer the minimization problem to optimize the blur estimation. To enhance the accuracy of the outputs, a coarse-to-fine updating procedure is integrated into the Bregman iterations. The resulting subproblems are efficiently addressed during the alternating iteration by employing methods such as the fast Fourier transform (FFT) and hard shrinkage. In the second phase, the total variation (TV) deconvolution model is applied to sharp image reconstruction, and a classic half-quadratic approach is applied to handle the model with high efficiency. In our experiments, the proposed method and three similar methods are employed to deal with synthetic blurry images and real-world blurry images from open image databases. The deblurring results are presented in the form of recovered images and peak signal-to-noise ratio (PSNR) values. To compare speed performances, the computation times for image deblurring are computed and reported. The proposed method can be applied to efficiently handle various types of blurry images and produce satisfactory outputs. Experimental outputs indicated that the proposed method provides superior restoration quality and computing speed compared with alternatives.
- Subjects
IMAGE processing; IMAGE reconstruction; SPLIT Bregman method; FINITE element method; ARTIFICIAL intelligence; NUMERICAL analysis
- Publication
IAENG International Journal of Computer Science, 2018, Vol 45, Issue 4, p123
- ISSN
1819-656X
- Publication type
Article