2004
DOI: 10.1109/tip.2004.834669
|View full text |Cite
|
Sign up to set email alerts
|

Fast and Robust Multiframe Super Resolution

Abstract: Super-resolution reconstruction produces one or a set of high-resolution images from a set of low-resolution images. In the last two decades, a variety of super-resolution methods have been proposed. These methods are usually very sensitive to their assumed model of data and noise, which limits their utility. This paper reviews some of these methods and addresses their short-comings. We propose an alternate approach using L1 norm minimization and robust regularization based on a bilateral prior to deal with di… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

5
1,355
3
14

Year Published

2007
2007
2018
2018

Publication Types

Select...
4
4
2

Relationship

0
10

Authors

Journals

citations
Cited by 1,830 publications
(1,378 citation statements)
references
References 23 publications
5
1,355
3
14
Order By: Relevance
“…First, the raw SD OCT images were denoised using an iterative maximum a posteriori-based algorithm, exploiting L2 norm penalty and a variation of the Tikhonov prior as the likelihood and regularization terms, respectively. 12 In this article, we refer to these individually denoised images as enhanced B scans. Then, the contrast of the SVP image was enhanced by comparing and normalizing the intensity of each projected B-scan (horizontal lines on the SVP) with respect to the global intensity of the SVP.…”
Section: After Informed Consent Was Obtained the Studymentioning
confidence: 99%
“…First, the raw SD OCT images were denoised using an iterative maximum a posteriori-based algorithm, exploiting L2 norm penalty and a variation of the Tikhonov prior as the likelihood and regularization terms, respectively. 12 In this article, we refer to these individually denoised images as enhanced B scans. Then, the contrast of the SVP image was enhanced by comparing and normalizing the intensity of each projected B-scan (horizontal lines on the SVP) with respect to the global intensity of the SVP.…”
Section: After Informed Consent Was Obtained the Studymentioning
confidence: 99%
“…which shows that, after shifting and zero filling, D T k B T k W T k copies values from the lowresolution to the high-resolution images, and W k B k D k reverses the operation [11]. Pixel values are unaffected by these complimentary operations, implying that each entry in J 1 is impacted by entries from all low-resolution images.…”
Section: Image Degradation Modelmentioning
confidence: 94%
“…A depth map or a three-dimensional shape of subjects is estimated based on a passive multibaseline stereo method [13]. Then, super-resolution processing [14] unifies all the elemental images in the compound-eye image to a single image with the PSFs that reflect their slight dependency on the depth. Consequently, a three dimensional model of the subjects is built on a computer.…”
Section: Methodsmentioning
confidence: 99%