Current browse context:
eess.IV
Change to browse by:
References & Citations
Electrical Engineering and Systems Science > Image and Video Processing
Title: An Analysis and Implementation of the HDR+ Burst Denoising Method
(Submitted on 18 Oct 2021)
Abstract: HDR+ is an image processing pipeline presented by Google in 2016. At its core lies a denoising algorithm that uses a burst of raw images to produce a single higher quality image. Since it is designed as a versatile solution for smartphone cameras, it does not necessarily aim for the maximization of standard denoising metrics, but rather for the production of natural, visually pleasing images. In this article, we specifically discuss and analyze the HDR+ burst denoising algorithm architecture and the impact of its various parameters. With this publication, we provide an open source Python implementation of the algorithm, along with an interactive demo.
Link back to: arXiv, form interface, contact.