Image Analogies

By Aaron Hertzmann, Charles E. Jacobs, Nuria Oliver, Brian Curless, David Salesin. 2001

1,936 citations recorded by the source corpus.

Subfields: Computer Graphics and Computer-Aided Design, 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 Graphics and Computer-Aided Design

In Computer Vision and Pattern Recognition

Citations in the corpus

Colorful Image Colorization (2016, methodology): “Then, following the Image Analogies framework [17], color is transferred onto the input image from analogous regions of the reference image(s) [18,19,20,21].”
Semantic Image Synthesis With Spatially-Adaptive Normalization (2019, background): “An other widely-used form is image-to-image translation [16, 20,22,23,24,28,32,54,56,57], where both input and out put are images.”
Image-to-Image Translation with Conditional Adversarial Networks (2016, background): “This connection was previously explored in [35], and is also the common assumption in models of texture [15, 19] and style [14, 23, 20, 34].”
Unpaired Image-to-Image Translation Using Cycle-Consistent Adversarial Networks (2017, background): “’s Image Analogies [17], who employ a nonparametric texture model [8] on a single input-output training image pair.”
Rewriting a Deep Generative Model (2020, background): “Common operations include color transfer [65,53], image deformation [67,75], object cloning [62,11], and patch-based image synthesis [20,5,29].”
A Learned Representation For Artistic Style (2016, positive): “Hertzmann et al. (2001) introduces the concept of “image analogies”: given a pair of “unﬁltered” and “ﬁltered” versions of an examplar image, a target image is processed to create an analogous “ﬁltered” result.”
Precomputed Real-Time Texture Synthesis with Markovian Generative Adversarial Networks (2016, background): “Consequently, the ﬁdelity of details is good, but additional guidance is required if non-trivial global structure should be reproduced [6,12,1,8,21].”
Multimodal Unsupervised Image-to-Image Translation (2018, background): “→ Note that this is similar to classical style transfer algorithms [5, 51–56] that transfer the style of one image to another.”
Shape Completion Using 3D-Encoder-Predictor CNNs and Shape Synthesis (2016, positive): “Similar to Hertzmann et al. [11], our volumetric synthesis searches for similar volumetric patches in the set of k nearest shape neighbors to reﬁne the voxel predictions from the 3D-EPN.”
Combining Markov Random Fields and Convolutional Neural Networks for Image Synthesis (2016, methodology): “Concretely, we implement this by splitting up the input into a “style” image and a “content” image [8, 12].”
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.”
Exploring the structure of a real-time, arbitrary neural artistic stylization network (2017, methodology): “(2001) [11] demonstrated a non-parametric technique for imbuing an arbitrary filter to an image based on pairs of unfiltered and filtered images.”
Semantic Style Transfer and Turning Two-Bit Doodles into Fine Artworks (2016, methodology): “The image analogy algorithm (Hertzmann et al. 2001) is able to transfer artistic style using pixel features and their local neighborhoods.”
Fast Patch-based Style Transfer of Arbitrary Style (2016, background): “Early works deﬁne style using similarity measures or local statistics based on the pixel values [7, 8, 13, 14, 20, 24].”
Universal Style Transfer via Feature Transforms (2017, methodology): “Existing style transfer methods are mostly example-based [13, 25, 24, 7, 21].”
Deep Photo Style Transfer (2017, background): “Local style transfer algorithms based on spatial color mappings are more expressive and can handle a broad class of applications such as time-of-day hallucination [4, 15], transfer of artistic edits [1, 14, 17], weather and season change [4, 8], and painterly stylization [5, 6, 10, 13].”
Real-time user-guided image colorization with learned deep priors (2017, background): “…photo in one of two ways: either by matching it to an exemplar color image in a database and non-parametrically “stealing” colors from that photo, an idea going back to Image Analogies (Hertzmann et al. 2001), or by learning parametric mappings from grayscale to color from large-scale image data.”
Single-Image Super-Resolution Using Sparse Regression and Natural Image Prior (2010, background): “This information is available either in the explicit form of a distribution or energy functional defined on the image class [1], [2], [3], [4], and/or in the implicit form of example images which leads to examplebased super-resolution [5], [6], [7], [8], [9], [10], [11], [12], [13].”
Image upsampling via imposed edge statistics (2007, methodology): “Resolution enhancement was also suggested as an application of the Image Analogies paradigm [Hertzmann et al. 2001], and was later modiﬁed in [Freeman et al. 2002], to allow broader applicability.”