Learning how to inpaint from global image statistics

By Anat Levin, Assaf Zomet, Yair Weiss. Proceedings Ninth IEEE International Conference on Computer Vision, 2003

391 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/81410e41f92a63a4aa9618ce0032c9031203a1e7

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

Progressive Image Inpainting with Full-Resolution Residual Network (2019, methodology): “he effectiveness of residual architecture in DNNs. 2.1 Traditional Methods When it comes to traditional inpainting methods, we can further divide them into two sub-categories: diffusion-based methods [1, 3, 4, 7, 15, 17] and patch-based methods [2, 5, 6, 10, 24, 30]. Diffusionbased methods usually f”
Diversity-Generated Image Inpainting with Style Extraction (2019, methodology): “[5] used the variational method, the histogram statistical method based on local features [6], and the fast marching method based on the level set application proposed by Telea [7].”
Distributed message passing for large scale graphical models (2011, background): “Here we focus on message passing algorithms which have proven very effective to solve vision tasks such as super-resolution [3], stereo reconstruction [26], segmentation [15], denoising [13], and inpainting [16].”
Image Inpainting by Patch Propagation Using Patch Sparsity (2010, background): “painting [6]–[8], deblurring [26], [27], and so on.”
Transformation guided image completion (2013, background): “[17] incorporated global image statistics such as gradient magnitude and relative gradient angle for inpainting.”
Semantic Image Inpainting with Progressive Generative Networks (2018, background): “One popular category of approaches is diffusion-based [5, 6, 17, 27].”
Image super-resolution using gradient profile prior (2008, neutral): “This kind of “sparseness prior” has been successfully applied to super-resolution[28], denoising [23] [24], inpainting [17], transparency separation [16] and deblurring [8, 15].”
Blind Motion Deblurring Using Image Statistics (2006, background): “Recent work in computer vision applied natural images priors for a variety of applications like denoising [25, 24], super resolution [27], video matting [2], inpainting [16] and reflections decomposition [15].”
Image Inpainting by End-to-End Cascaded Refinement With Mask Awareness (2021, background): “In the literature, statistics based solutions are also extensively studied [28]–[30].”
Statistics of Patch Offsets for Image Completion (2012, positive): “Gradient-domain statistics have been applied in denoising [29], deconvolution [30], and inpainting [3], whereas recent patch-domain statistics have been shown very successful in denoising [29] and super-resolution [31].”
Fields of Experts (2009, background): “Levin et al. (2003) have a similar motivation in that they exploit learned models of image statistics for inpainting .”
Image Inpainting : Overview and Recent Advances (2014, background): “The total variation (TV) image model, first proposed for denoising and deblurring applications in [12], has been applied to image inpainting in [13]–[15].”
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].”
UCTGAN: Diverse Image Inpainting Based on Unsupervised Cross-Space Translation (2020, methodology): “Among them, patch-based methods [5] and diffusion-based methods [11] are the most typical.”
Gradient Profile Prior and Its Applications in Image Super-Resolution and Enhancement (2011, background): “I. INTRODUCTION T HERE HAS been much attention in modeling naturalimage prior in recent years [1]–[3].”
Coherent Semantic Attention for Image Inpainting (2019, background): “Non-learning approaches such as [11, 1, 3, 5, 6, 9, 15, 2, 32, 18, 34, 28, 12, 29] fill in missing regions by propagating neighboring information or copying information from similar patch of the background.”
Learning to Incorporate Structure Knowledge for Image Inpainting (2020, methodology): “They either propagate low-level features from surroundings to the missing regions following a diffusive process (Bertalmio et al. 2000; Levin, Zomet, and Weiss 2003; Roth and Black 2005) or ﬁll holes by searching and fusing similar patches from the same image or external image databases (Hays and…”
Pluralistic Image Completion (2019, methodology): “Intra-Image Completion Traditional intra-image completion, such as diffusion-based methods [4, 1, 22] and patchbased methods [5, 7, 8, 3], assume image holes share similar content to visible regions; thus they would directly match, copy and realign the background patches to complete the holes.”