Burst photography for high dynamic range and low-light imaging on mobile cameras

By Samuel W. Hasinoff, Dillon Sharlet, Ryan Geiss, Andrew Adams, Jonathan T. Barron, Florian Kainz, Jiawen Chen, Marc Levoy. ACM Transactions on Graphics, 2016

557 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/2d6723d19df684b21c9b40f11eb03b0a9e3d36b0

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

Learning to See in the Dark (2018, methodology): “[14] propose to denoise a burst of images from the same scene.”
NeRF in the Dark: High Dynamic Range View Synthesis from Noisy Raw Images (2021, methodology): “We then capture a stabilized burst of 50-100 longer exposures on a tripod and robustly merge them using HDR+ [21] to create a clean ground truth frame.”
Noise2Noise: Learning Image Restoration without Clean Data (2018, background): “Other effects, e.g., dark current and quantization, are dominated by Poisson noise, can be made zero-mean (Hasinoff et al., 2016), and hence pose no problems for training with noisy targets.”
Deep high dynamic range imaging of dynamic scenes (2017, background): “It is worth noting that dynamic range can also be increased by combining multiple images captured with the same exposure time [Zhang et al. 2010; Hasinoff et al. 2016].”
Unprocessing Images for Learned Raw Denoising (2018, methodology): “An image from the HDR+ dataset [21], where we present (a) the noisy input image and (b) the output of our model, in the same format as Figure 1.”
Burst Denoising with Kernel Prediction Networks (2017, methodology): “This compresses the histogram of intensities to more closely match our real data, which is underexposed to avoid highlight clipping as in the HDR+ burst imaging pipeline [8].”
Single image portrait relighting (2019, methodology): “Modern digital cameras carefully consider and modify the brightness, exposure, and tone of their output images, all of which can be thought of as the coarse manipulation of scene lighting [Hasinoff et al. 2016].”
Seeing Motion in the Dark (2019, methodology): “Some existing methods rely on pre-warping [28, 18, 15] or learning to align [35] to reduce noise temporally.”
Generating Training Data for Denoising Real RGB Images via Camera Pipeline Simulation (2019, background): “For example, most works do not consider demosaicking artifacts, and the ones that do typically use bilinear demosaicking [5, 13, 25], which is rarely used in real consumer cameras [15, 14].”
Learning Image-Adaptive 3D Lookup Tables for High Performance Photo Enhancement in Real-Time (2020, methodology): “The HDR+ dataset [45] is a burst photography dataset collected by Google camera group for research of high dynamic range (HDR) and low-light imaging on mobile cameras.”
Perceptual Quality Assessment of Smartphone Photography (2020, background): “Arguably it is often challenging for professional photographers to acquire high-quality pictures consistently across a variety of natural scenes, especially in low-light and highdynamic-range (HDR) scenarios [7].”
End-to-end optimization of optics and image processing for achromatic extended depth of field and super-resolution imaging (2018, background): “Many of these approaches use either optical coding, multiplexing, burst photography [Hasinoff et al. 2016], or multi-shot approaches to capture high-dimensional visual data [Wetzstein et al.”