Fields of Experts

By Stefan Roth, Michael J. Black. International Journal of Computer Vision, 2009

876 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/dfcae80f4d34ac09ba8063c5cfb5be954d0bf5f1

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Deep Unfolding Network for Image Super-Resolution (2020, methodology): “We choose the widely-used color BSD68 dataset [40, 46] to quantitatively evaluate different methods.”
Weighted Nuclear Norm Minimization and Its Applications to Low Level Vision (2016, methodology): “…WNNM-MC with NNM-MC and several representative and state-of-the-art inpainting methods, including the TV method (Chan and Shen, 2005), FOE method14 (Roth and Black, 2009), variational nonlocal example-based (VNL) method15 (Arias et al, 2011) and the beta process dictionary learning (BPDL) method16…”
Regularized estimation of image statistics by Score Matching (2010, methodology): “Subsequently, a convolutional PoE model known as Fields-of-Experts [21] (FoE) was estimated using regularized SM: E(x;W,α) = ∑”
On instabilities of deep learning in image reconstruction and the potential costs of AI (2019, methodology): “n, with c= 15. The network structure is inspired by the unfolding of a variational minimization problem including a ﬁdelity term and a regularization term deﬁned according the Fields of Experts model [44]. In particular, each iteration of the corresponding Landweber method [45] corresponds to a laye”
Learning Proximal Operators: Using Denoising Networks for Regularizing Inverse Imaging Problems (2017, neutral): “18 FoE [32] 24.”
Benchmarking Denoising Algorithms with Real Photographs (2017, methodology): “For the task of (non-blind) denoising, we compare the performance of Weighted Nuclear Norm Minimization (WNNM) [15], K-SVD [1], Expected Patch Log Likelihood (EPLL) [39], Field of Experts (FoE) [32] with the filters of [14], Nonlocally Centralized Sparse Representations (NCSR) [8], and BM3D [7].”
Generative Image Modeling Using Spatial LSTMs (2015, methodology): “Generative image models are also useful in more traditional applications such as image reconstruction [33, 35, 49] or compression [47].”
Learning to Maintain Natural Image Statistics (2018, background): “Maintaining natural image statistics has been known for years as a key factor in the generation of natural looking images [1,2,3,4,5].”
Plug-and-Play Image Restoration With Deep Denoiser Prior (2020, methodology): “Table 2 reports the color image denoising results of different methods for noise levels 15, 25 and 50 on CBSD68 [3], [44], [49], Kodak24 [71] and McMaster [72] datasets.”
Patches, Planes and Probabilities: A Non-Local Prior for Volumetric 3D Reconstruction (2016, background): “Even in 2D, high-order spatial priors are expensive to represent and optimize [27, 33, 39].”
Edge-based blur kernel estimation using patch priors (2013, background): “This is also why state-of-theart image restoration methods often involve larger neighborhoods or image patches [2, 15, 18, 20].”
Patch Group Based Nonlocal Self-Similarity Prior Learning for Image Denoising (2015, methodology): “The Fields of Experts (FoE) [17] proposed by Roth and Black models the filtering responses with Student’s t-distribution to learn filters through Markov Random Field (MRF) [18].”
Parameter Learning and Convergent Inference for Dense Random Fields (2013, background): “Models with hierarchical structure and higher-order potentials have been proposed to reduce these artifacts (He et al., 2004; Roth & Black, 2009; Kohli et al., 2009).”
Learning Deep CNN Denoiser Prior for Image Restoration (2017, methodology): “We empirically find using large dataset does not improve the PSNR results of BSD68 dataset [50] but can slightly improve the performance of other testing images.”
Learning Enriched Features for Real Image Restoration and Enhancement (2020, methodology): “Method DnCNN MLP GLIDE TNRD FoE BM3D WNNM NLM KSVD EPLL CBDNet RIDNet VDN MIRNet [118] [11] [95] [19] [84] [20] [37] [10] [2] [126] [38] [5] [117] (Ours)”
Noise-contrastive estimation: A new estimation principle for unnormalized statistical models (2010, methodology): “This is diﬀerent compared to (Roth & Black, 2009), where the ﬁlters had no clear structure.”
Blind Deblurring Using Internal Patch Recurrence (2014, background): “Such priors have led to state-of-the-art results in various inverse problems [16] including nonblind deblurring [3,22,4] (namely, deblurring with a known blur kernel).”
Analysis K-SVD: A Dictionary-Learning Algorithm for the Analysis Sparse Model (2013, methodology): “Instead we focus on the FoE approach and refer the readers to [18] for a comprehensive review.”