Image hallucination with primal sketch priors

By Nanning Zheng, Hai Tao, Harry Shum. 2003 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2003. Proceedings., 2003

435 citations recorded by the source corpus.

We propose a Bayesian approach to image hallucination. Given a generic low resolution image, we hallucinate a high resolution image using a set of training images. Our work is inspired by recent progress on natural image statistics that the priors of image primitives can be well represented by examples. Specifically, primal sketch priors (e.g., edges, ridges and corners) are constructed and used to enhance the quality of the hallucinated high resolution image. Moreover, a contour smoothness constraint enforces consistency of primitives in the hallucinated image by a Markov-chain based inference algorithm. A reconstruction constraint is also applied to further improve the quality of the hallucinated image. Experiments demonstrate that our approach can hallucinate high quality super-resolution images.

Subfields: Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/e922ad9daa9c6561ad754a18bf5edd069105dee6

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

Super-resolution: a comprehensive survey (2014, methodology): “– [13], [138], [151], [180], [246], [286], [322], [367], [386],”
Fusing multi-scale information in convolution network for MR image super-resolution reconstruction (2018, neutral): “According to previous research on SR [38], the extraction of multi-scale information improves the reconstruction results.”
Accurate Blur Models vs. Image Priors in Single Image Super-resolution (2013, background): “Developing sophisticated image priors has been the focus of much single image SR research in the past decade, with many significant successes [7, 20, 23, 8, 6, 9, 21, 27, 25, 5, 19, 22, 1, 8, 13].”
Single-Image Super-Resolution: A Benchmark (2014, methodology): “Various learning methods of the mapping functions have been proposed such as weighted average [31,2], kernel regression [17], support vector regression [23], Gaussian process regression [13], sparse dictionary representation [46,7,5,24,45,39,47,19,14].”
Image super-resolution as sparse representation of raw image patches (2008, methodology): “Since there are no explicit edges in most part of the image, methods proposed in [12], [23], and [6] would have tremendous difficulty here.”
Gradient Profile Prior and Its Applications in Image Super-Resolution and Enhancement (2011, methodology): “The learning based approach [31], [32], [5], [33], [4], [28], [34]–[36], [17] “hallucinates” high frequency details from a training set of HR/LR image pairs.”
Learning to Restore Low-Light Images via Decomposition-and-Enhancement (2020, background): “Such low dimensionality implies that a small number of image examples are sufficient to represent the image primitives well [41].”
Perceptual Losses for Real-Time Style Transfer and Super-Resolution (2016, methodology): “They group super-resolution techniques into prediction-based methods (bilinear, bicubic, Lanczos, [24]), edge-based methods [25,26], statistical methods [27,28,29], patch-based methods [25,30,31,32,33,34,35,36], and sparse dictionary methods [37,38].”
High-resolution hyperspectral imaging via matrix factorization (2011, methodology): “This approach is often used in image super-resolution methods [29].”
A Convolutional Neural Network Approach for Post-Processing in HEVC Intra Coding (2016, background): “Note that residue learning is also a common strategy in super-resolution with or without CNN [8,14].”
A Comprehensive Survey to Face Hallucination (2013, background): “2000, 2002), including image hallucination (Sun et al. 2003; Xiong et al. 2009), image analogy (Hertzmann et al.”
Super-resolution through neighbor embedding (2004, methodology): “Some methods make use of a training set of images [1, 9, 10, 11, 12, 14, 19], while others do not require a training set but reThis is the Pre-Published Version”