Texture synthesis by non-parametric sampling

By Alexei A. Efros, Thomas K. Leung. 1999

Subfields: Computer Vision and Pattern Recognition

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

Context Encoders: Feature Learning by Inpainting (2016, background): “Inpainting and hole-ﬁlling It is important to point out that our hole-ﬁlling task cannot be handled by classical in-painting [4, 32] or texture synthesis [2, 11] approaches, since the missing region is too large for local non-semantic methods to work well.”
Large Scale Image Completion via Co-Modulated Generative Adversarial Networks (2021, methodology): “Traditional methods (Ballester et al., 2001; Barnes et al., 2009; Darabi et al., 2012; Efros & Freeman, 2001; Efros & Leung, 1999) utilize only low-level features and fail to generate semantically consistent contents.”
Free-Form Image Inpainting With Gated Convolution (2018, methodology): “The former approach [3, 8, 9] can synthesize plausible stationary textures, but usually makes critical failures in non-stationary cases like complicated scenes, faces and objects.”
Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks (2015, background): “The former copy patches from training images to perform, for example, texture synthesis [7] or super-resolution [9].”
A Learned Representation For Artistic Style (2016, positive): “On the computer vision side, Efros & Leung (1999) and Wei & Levoy (2000) attempt to “grow” textures one pixel at a time using non-parametric sampling of pixels in an examplar image.”
Arbitrary Style Transfer in Real-Time with Adaptive Instance Normalization (2017, background): “The problem of style transfer has its origin from non-photo-realistic rendering [28], and is closely re-lated to texture synthesis and transfer [13, 12, 14].”
Generative Image Modeling Using Style and Structure Adversarial Networks (2016, methodology): “Non-parametric approaches perform matching of an image or patch with the database for tasks such as texture synthesis [20] or super-resolution [21].”
Toward Multimodal Image-to-Image Translation (2017, background): “A different approach is autoregressive models [11, 32, 33], which are effective at modeling natural”
Combining Markov Random Fields and Convolutional Neural Networks for Image Synthesis (2016, methodology): “Usually, this is done using a simple nearest-neighbor scheme [7], and inference is performed by approximate MRF inference [6, 15, 16] or greedy approximations [3, 7, 13, 26].”
View Synthesis by Appearance Flow (2016, neutral): “In texture synthesis [28,29], the synthesized image is pieced together by combining samples of the input texture image in a visually consistent way, whereas for texture transfer [30,31], an additional constraint aims to make the overall result also mimic a secondary “source” image.”
Generative Image Inpainting with Contextual Attention (2018, background): “Early works [3, 14] attempted to solve the problem using ideas similar to texture synthesis [10, 11], i.e. by matching and copying background patches into holes starting from low-resolution to high-resolution or propagating from hole boundaries.”
Exploring the structure of a real-time, arbitrary neural artistic stylization network (2017, background): “Such work focused on building non-parametric techniques for “growing” visual textures one pixel [5, 27] or one patch [4, 17] at a time.”
Scribbler: Controlling Deep Image Synthesis with Sketch and Color (2016, neutral): “Previously, the most successful methods tended to be non-parametric approaches which found clever ways to reuse existing image fragments [29, 9, 15, 6, 3].”
High-Resolution Image Inpainting Using Multi-scale Neural Patch Synthesis (2016, methodology): “The ﬁrst group of approaches [14, 13, 27, 26] rely on texture synthesis techniques, which ﬁll in the hole by extending textures from surrounding regions [6, 12, 40, 41, 23, 24, 2].”
Image inpainting (2000, methodology): “Moreover, since the inpainting algorithm here presented can be used not just to restore damaged photographs but also to remove undesired objects and writings on the image, the regions to be inpainted must be marked by the user, since they depend on his/her subjective selection.”