High dynamic range image reconstruction from hand-held cameras

Abstract

This paper presents a technique for reconstructing a high-quality high dynamic range (HDR) image from a set of differently exposed and possibly blurred images taken with a hand-held camera. Recovering an HDR image from differently exposed photographs has become very popular. However, it often requires a tripod to keep the camera still when taking photographs of different exposures. To ease the process, it is often preferred to use a hand-held camera. This, however, leads to two problems, misaligned photographs and blurred long-exposed photographs. To overcome these problems, this paper adapts an alignment method and proposes a method for HDR reconstruction from possibly blurred images. We use Bayesian framework to formulate the problem and apply a maximumlikelihood approach to iteratively perform blur kernel estimation, HDR image reconstruction and camera curve recovery. When convergence, we simultaneously obtain an HDR image with rich and clear structures, the camera response curve and blur kernels. To show the effectiveness of our method, we test our method on both synthetic and real photographs. The proposed method compares favorably to two other related methods in the experiments.

Exploiting Light Polarization for Deep HDR Imaging from a Single Exposure

“…Numerous examples can be found in the literature, where researchers reconstructed HDR using multi-exposure images [ 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 ]. One of the earliest efforts in creating HDR images using multiple exposures was made by [ 16 ]; the authors introduced a novel method to recover the CRF as well as an HDR radiance map using multi-exposure images.…”

“…One of the earliest efforts in creating HDR images using multiple exposures was made by [ 16 ]; the authors introduced a novel method to recover the CRF as well as an HDR radiance map using multi-exposure images. Similarly, [ 17 ] proposed an HDR reconstruction method for handheld cameras since images taken by hand are more prone to artifacts, such as blurs. Images are first registered using the MTB algorithm, and then maximum likelihood is used to find the blur kernel and CRF.…”

Yang Chengzong and the University of Science and Technology of China

“…These methods aim to identify and reject moving regions inconsistent with a reference frame's background during synthesis. Various techniques, such as non-parametric estimation [22], local entropy [18], threshold bitmap [32,40], intensity histogram [36], and bi-directional similarity [25,68], detect moving pixels. However, they often lose fine details in dynamic areas and struggle to adapt to different cameras and exposure settings due to fixed threshold reliance [12].…”

“…Optimization-based methods. They jointly perform alignment and reconstruction in the form of an energy function, or only use the framework to perform one of them, for example, using Poisson equation [8,10], graph cut [7], Markov random field [12,19], Bayesian [32], energy-based optic flow [69], Patch-Match [17,47], rank minimization [23,39] and background estimation [13]. Despite progress, they fall short of learning-based methods in deghosting results and tend to be time-consuming.…”

“…Recent advancements in HDR imaging have evolved from early methods using hand-crafted features [17,18,21,32,37,47] Fig. 1: Visual and quantitative comparison.…”