Learning a Deep Convolutional Network for Light-Field Image Super-Resolution

By Youngjin Yoon, Hae-Gon Jeon, Donggeun Yoo, Joon-Young Lee, In-So Kweon. 2015 IEEE International Conference on Computer Vision Workshop (ICCVW), 2015

378 citations recorded by the source corpus.

Commercial Light-Field cameras provide spatial and angular information, but its limited resolution becomes an important problem in practical use. In this paper, we present a novel method for Light-Field image super-resolution (SR) via a deep convolutional neural network. Rather than the conventional optimization framework, we adopt a datadriven learning method to simultaneously up-sample the angular resolution as well as the spatial resolution of a Light-Field image. We first augment the spatial resolution of each sub-aperture image to enhance details by a spatial SR network. Then, novel views between the sub-aperture images are generated by an angular super-resolution network. These networks are trained independently but finally finetuned via end-to-end training. The proposed method shows the state-of-the-art performance on HCI synthetic dataset, and is further evaluated by challenging real-world applications including refocusing and depth map estimation.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/2c5f3738f959dc42560a2a0d3a131abc10a119f1

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

Light Field Super-Resolution with Zero-Shot Learning (2021, methodology): “[45] proposed the first light field SR network LFCNN, by reusing the SRCNN architecture [8] with multiple channels.”
Light Field Image Processing: An Overview (2017, methodology): “The effectiveness can be more significant when the light field is super-resolved using a depth-free framework, such as learning-based methods [71], [72].”
A survey on deep learning techniques for image and video semantic segmentation (2018, background): “Some of those applications include autonomous driving [1][2][3], human-machine interaction [4], computational photography [5], image search engines [6], and augmented reality to name a few.”
Learning-based view synthesis for light field cameras (2016, neutral): “…color estimation components which we model using two sequential CNNs. Note that although CNNs have been recently used for light ﬁeld super-resolution [Yoon et al. 2015] and depth estimation [Heber and Pock 2016], these methods are not able to directly synthesize novel views at arbitrary locations.”
A Review on Deep Learning Techniques Applied to Semantic Segmentation (2017, background): “Some of those applications include autonomous driving [1] [2] [3], human-machine interaction [4], computational photography [5], image search engines [6], and augmented reality to name a few.”
Light field video capture using a learning-based hybrid imaging system (2017, neutral): “Yoon et al. (2015) apply convolutional neural networks (CNN) to perform spatial and angular super-resolution.”