Loss-Specific Training of Non-Parametric Image Restoration Models: A New State of the Art

Jancsary, Jeremy; Nowozin, Sebastian; Rother, Carsten

doi:10.1007/978-3-642-33786-4_9

Cited by 95 publications

(138 citation statements)

References 16 publications

Supporting

Mentioning

137

Contrasting

Order By: Relevance

“…A proper integration of different priors could further get better the denoising performance. For example, the methods in [10], [11], and [12] integrate image local sparsity prior with nonlocal NSS prior, and they have shown promising denoising results. In the proposed GHP model, the following sparse nonlocal regularization term used in the non -locally centralized sparse representation (NCSR) model [11]:…”

Section: Denoising With Gradient Histogram Preservation 1 the Denoismentioning

confidence: 99%

Nonlocal Hierarchical Dictionary Learning using Wavelets and Gradient Histogram Preservation for Image Denoising: A Review

Prajapati¹,

Gyanchandani²

2015

IJCA

View full text Add to dashboard Cite

Image denoising is an important image processing task, both as a process itself, and as a component in other processes. The main properties of a good image denoising model are that it will remove noise while preserving edges. Traditionally, linear models have been used. One common approach is to use a Gaussian filter, In spite of the great success of many denoising algorithms; they tend to smooth the fine scale image textures when removing noise, degrading the image visual quality. To address this problem we compare two methods in this paper. The Nonlocal Hierarchical Dictionary Learning using Wavelet (NHDLW) and Gradient Histogram Preservation (GHP),which is large success in denoising. Experimental result shows that the NHDLW get significantly better denoising results especially on an image denoising algorithms on higher noise levels.

show abstract

Section: Denoising With Gradient Histogram Preservation 1 the Denoismentioning

confidence: 99%

Nonlocal Hierarchical Dictionary Learning using Wavelets and Gradient Histogram Preservation for Image Denoising: A Review

Prajapati¹,

Gyanchandani²

2015

IJCA

View full text Add to dashboard Cite

show abstract

“…For example, the methods in [9], [8], [10] integrate image local sparsity prior with nonlocal NSS prior and they have shown promising denoising results. In the proposed GHP model, we adopt the following sparse nonlocal regularization term proposed in the non-locally centralized sparse representation (NCSR) model [8]: (8) where is defined as the weighted average of :…”

Section: International Journal Of Computer Applications (0975 -8887) mentioning

confidence: 99%

Improved Texture Enhanced Image Denoising

Rajak¹,

Chugh²

2015

IJCA

View full text Add to dashboard Cite

ABSTRACThere this work is introducing the new technique using the improved texture enhanced framework for image denoising. This technique is fast as compared to the higher order singular value decomposition (HOSVD) as we have in the previous work. The HOSVD technique simply compose in a cluster, alike Patches of noisy image in 3D heap, work out HOSVD factors of this heap, handles these factors by stiff thresholding, and turn upside down the HOSVD transmute to yield the final resultant image. Whereas improved texture enhanced image denoising have proven to be effective and robust in many image denoising tasks. It is experimentally demonstrating approximately 5 percent improved PSNR characteristics of ITEID technique on gray scale images. The ITEID process yields state-of-the-art outcomes on gray images, than HOSVD image data denoising process at moderately great noise stages.

show abstract

“…Learning-based approaches have been restricted to generatively trained models [25], but have found limited adoption due to computational challenges in inference. This is in contrast to image denoising, where discriminative approaches have been used extensively [2,4,14,27], and are often characterized by state-of-the-art restoration performance combined with low computational effort.…”

Section: Introductionmentioning

confidence: 99%

Discriminative Non-blind Deblurring

Schmidt

Rother

Nowozin

et al. 2013

2013 IEEE Conference on Computer Vision and Pattern Recognition

Self Cite

134

View full text Add to dashboard Cite

Non-blind deblurring is an integral component of blind approaches for removing image blur due to camera shake. Even though learning-based deblurring methods exist, they have been limited to the generative case and are computationally expensive. To this date, manually-defined models are thus most widely used, though limiting the attained restoration quality. We address this gap by proposing a discriminative approach for non-blind deblurring. One key challenge is that the blur kernel in use at test time is not known in advance.To address this, we analyze existing approaches that use half-quadratic regularization. From this analysis, we derive a discriminative model cascade for image deblurring. Our cascade model consists of a Gaussian CRF at each stage, based on the recently introduced regression tree fields. We train our model by loss minimization and use synthetically generated blur kernels to generate training data. Our experiments show that the proposed approach is efficient and yields state-of-the-art restoration quality on images corrupted with synthetic and real blur.

show abstract

Loss-Specific Training of Non-Parametric Image Restoration Models: A New State of the Art

Cited by 95 publications

References 16 publications

Nonlocal Hierarchical Dictionary Learning using Wavelets and Gradient Histogram Preservation for Image Denoising: A Review

Nonlocal Hierarchical Dictionary Learning using Wavelets and Gradient Histogram Preservation for Image Denoising: A Review

Improved Texture Enhanced Image Denoising

Discriminative Non-blind Deblurring

Contact Info

Product

Resources

About