Image smoothing via L0 gradient minimization

By Li Xu, Cewu Lu, Yi Xu, Jiaya Jia. ACM Transactions on Graphics, 2011

1,497 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/000b9a90bbea62e4222704c616c0c2ee65609aa7

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

Image melding (2012, methodology): “Others [Tappen et al. 2005; Xu et al. 2011] observed that since gradients are sparse in natural images, one should use robust norms (Lp with p = 1 or lower) for optimization terms involving image gradients.”
Underwater Image Enhancement by Dehazing With Minimum Information Loss and Histogram Distribution Prior (2016, neutral): “(a) Raw texture image, (b) the result of TV method [32], (c) the result of L0 smoothing method [33], and (d) the result of the RTV method.”
Differentiable surface splatting for point-based geometry processing (2019, methodology): “We experimented with two types of image filters, L0 smoothing [Xu et al. 2011] and superpixel segmentation [Achanta et al. 2012].”
Pretraining is All You Need for Image-to-Image Translation (2022, methodology): “In particular, we also introduce L 0 ﬁlter [55] to mimic the oversmoothed effect.”
Global Texture Enhancement for Fake Face Detection in the Wild (2020, background): “• L0-filtered (skin): Small textures of the skin regions are filtered with L0 filter [31].”
Unnatural L0 Sparse Representation for Natural Image Deblurring (2013, methodology): “It can be optimized using the half-quadratic L0 minimization solver introduced in [27].”
Deep Joint Image Filtering (2016, result): “(a) Input (b) RGF [9] (c) L0 [34] (d) Xu [8] (e) Ours (f) Kopf [33]”
Rolling Guidance Filter (2014, methodology): “Optimization-based methods include total variation (TV) [23], weighted least squares (WLS) [9], and L0 gradient minimization [29].”
Blind Image Deblurring Using Dark Channel Prior (2016, methodology): “To tackle the L0-regularized term, we use the half-quadratic splitting L0 minimization approach [35].”
Local Laplacian filters: edge-aware image processing with a Laplacian pyramid (2011, neutral): “…L 0 ( x 0 , y 0 ) apply remapping function: ˜ R 0 ← r g 0 , σ r ( R 0 ) compute sub-pyramid { L 0 [ ˜ R 0 ] } update output pyramid: L 0 [ I (cid:48) ]( x 0 , y 0 ) ← L 0 [ ˜ R 0 ]( x 0 , y 0 ) end for collapse output pyramid: I (cid:48) ← collapse ( { L (cid:96) [ I (cid:48) ] } ) cient at a time.”
Spectral–Spatial Hyperspectral Image Classification With Edge-Preserving Filtering (2014, methodology): “, the bilateral filter [34], the weighted least squares filter [37], the guided filter [35], the smoothed local histogram filter [46], the domain transform filter [47], the local linear Stein’s unbiased risk estimate (LL-SURE) filter [48], and the L0 gradient filter [49] have been proposed.”
StructureFlow: Image Inpainting via Structure-Aware Appearance Flow (2019, background): “The edge-preserved smooth methods [30, 31] aim to remove high-frequency textures while retaining the sharp edges and low-frequency structures.”
Image Deblurring via Extreme Channels Prior (2017, methodology): “(10) directly, we instead use an alternating minimization algorithm based on the half-quadratic splitting algorithm [24].”
Box-supervised Instance Segmentation with Level Set Evolution (2022, background): “Chan and Vese [7] later simplified the Mumford-Shah functional as a variational level set, which has been explored aplenty [31,34,46,52].”
CartoonGAN: Generative Adversarial Networks for Photo Cartoonization (2018, methodology): “Such methods use either image filtering [33] or formulations in optimization problems [35].”
Sparse Gradient Regularized Deep Retinex Network for Robust Low-Light Image Enhancement (2021, methodology): “Compared to [38], we provide a learnable solution, which can be integrated into the whole Retinex-Net for an end-to-end training.”