A non-local algorithm for image denoising

By A. Buades, Bartomeu Coll, Jean-Michel Morel. 2005

7,791 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology

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

In Media Technology

Citations in the corpus

Axial-DeepLab: Stand-Alone Axial-Attention for Panoptic Segmentation (2020, background): “Additionally, CNNs augmented with non-local means [10] are shown to be more robust to adversarial attacks [89].”
Non-local Neural Networks (2017, methodology): “Non-local means [4] is a classical filtering algorithm that computes a weighted mean of all pixels in an image.”
Early Convolutions Help Transformers See Better (2021, background): “[46] show that (single-headed) self-attention is a form of non-local means [2] and that integrating it into a ResNet [19] improves several tasks.”
Image-to-Image Translation with Conditional Adversarial Networks (2016, background): “, [14, 23, 18, 8, 10, 50, 30, 36, 16, 55, 58]), despite the fact that the setting is always the same: predict pixels from pixels.”
Feature Denoising for Improving Adversarial Robustness (2018, background): “Non-local means [2] compute a denoised feature map y of an input feature map x by taking a weighted mean of features in all spatial locations L:”
Countering Adversarial Images using Input Transformations (2018, background): “, 2011) are stronger defenses than deterministic denoising procedures such as bit-depth reduction, JPEG compression, or non-local means (Buades, 2005).”
Deep Image Prior (2017, methodology): “For reference, the scores for the two popular approaches CMB3D [6] and Non-local means [4] that do not require pretraining are 31.”
Going deeper with Image Transformers (2021, background): “denoising [10]), and more recently in conjunction with convolutions [68].”
Bottleneck Transformers for Visual Recognition (2021, background): “Connection to Non-Local Neural Nets: Non-Local (NL) Nets [67] make a connection between the Transformer and the Non-Local-Means algorithm [4].”
Scaling Local Self-Attention for Parameter Efficient Visual Backbones (2021, methodology): “Self-attention may also be regarded as an adaptive nonlinearity paralleling a long history of techniques in computer vision, such as bilateral filtering [36] and non-local means [3].”
Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising (2016, background): “Over the past few decades, various models have been exploited for modeling image priors, including nonlocal self-similarity (NSS) models [1], [2], [3], [4], sparse models [4], [5], [6], gradient models [7], [8], [9] and Markov random ﬁeld (MRF) models [10], [11], [12].”
Non-local sparse models for image restoration (2009, methodology): “…an explanatory variable for y[i] leads to the non-local means formulation, where the denoised pixel x[i] is obtained by a weighted average (the corresponding Nadaraya-Watson estimator [3]): x[i] = n j=1 K h (y i − y j ) n l=1 K h (y i − y l ) y[j], (1) and K h is a Gaussian kernel of bandwidth h.”
Single Image Super-Resolution With Non-Local Means and Steering Kernel Regression (2012, background): “Representative works are non-local means (NLM) for image denoising [32], [33], inpainting [34], and image SR reconstruction [10]–[12].”
Multi-Atlas Segmentation of Biomedical Images: A Survey (2014, methodology): “An interesting and more recent technique is inspired by the non-local means method (Buades et al., 2005) and utilizes a patch-based search strategy to identify correspondences with the atlases.”
Asymmetric Non-Local Neural Networks for Semantic Segmentation (2019, background): “Different from these works, our network uniquely incorporates pyramid sampling strategies with non-local blocks to capture the semantic statistics of different scales with only a minor budget of computation, while maintaining the excellent performance as the original non-local modules.”