2000
DOI: 10.1051/cocv:2000117
|View full text |Cite
|
Sign up to set email alerts
|

Image deblurring, spectrum interpolation and application to satellite imaging

Abstract: Abstract. This paper deals with two complementary methods in noisy image deblurring: a nonlinear shrinkage of wavelet-packets coefficients called FCNR and Rudin-Osher-Fatemi's variational method. The FCNR has for objective to obtain a restored image with a white noise. It will prove to be very efficient to restore an image after an invertible blur but limited in the opposite situation. Whereas the Total Variation based method, with its ability to reconstruct the lost frequencies by interpolation, is very well … Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

1
33
0

Year Published

2001
2001
2013
2013

Publication Types

Select...
6
2

Relationship

1
7

Authors

Journals

citations
Cited by 31 publications
(34 citation statements)
references
References 20 publications
1
33
0
Order By: Relevance
“…For convex energies (but not strictly convex, such as the model L 1 + T V ), we have obtained different minimizers for each of the three algorithms (sequential, divide-and-conquer, Ishikawa's algorithms), but with the same energy, as predicted by the theory. in [13,21,22]. We compare our results with the ones obtained by the duality-based algorithm of Chambolle which is presented in [9].…”
Section: Experiments and Discussionmentioning
confidence: 95%
“…For convex energies (but not strictly convex, such as the model L 1 + T V ), we have obtained different minimizers for each of the three algorithms (sequential, divide-and-conquer, Ishikawa's algorithms), but with the same energy, as predicted by the theory. in [13,21,22]. We compare our results with the ones obtained by the duality-based algorithm of Chambolle which is presented in [9].…”
Section: Experiments and Discussionmentioning
confidence: 95%
“…Finallly, a concrete application to satellite images is considered. While we had no access to images taken from satellites, the same procedure described and implemented in [8] was followed in order to simulate the typical degradation of satellite images, in order to appropriately compare degraded and restored images. First, the same convolution kernels were used as the ones described in [8] corresponding to two different satellites: SPOT 5 and a specific mode of SPOT 1.…”
Section: Psnr Time(s) Proposed Modelmentioning
confidence: 99%
“…These reference images are obtained by applying a prolate blur function to oversampled sharp images followed by down-sampling. The reader is referred to [8] for more details about the experiment. In this section, a comparison is offered between the proposed method and the simple ROF model for image restoration.…”
Section: Psnr Time(s) Proposed Modelmentioning
confidence: 99%
“…Since we rely on noise estimates that have a certain accuracy, exceeding this accuracy in the data fitting is useless (as we will see later in this Section). Moreover, as it has been observed in all numerical experiments [14,16,17,18,19,20,24,29,10,31,44,49], using total variation as regularizer in denoising or restoration generally carries some loss of texture and it is not desirable to compute the solution that (absolutely) minimizes the TV but to keep a solution with a slightly higher TV value in order to avoid the loss of textures.…”
Section: Convergence Of the Fixed Point Algorithm For Simplicity Givenmentioning
confidence: 99%
“…Let us finally mention that many numerical algorithms have been proposed to minimize total variation (or similar models) subject to a global constraint as in (1.5) [44,29,14,49,17,18,19,20,24,16,45]. Imposing local constraints in a partition of the image was proposed in [43] and further developed in [13,5,4].…”
Section: Introductionmentioning
confidence: 99%