Due to the poor lighting condition and limited dynamic range of digital imaging devices, the recorded images are often under-/over-exposed and with low contrast. Most of previous single image contrast enhancement (SICE) methods adjust the tone curve to correct the contrast of an input image. Those methods, however, often fail in revealing image details because of the limited information in a single image. On the other hand, the SICE task can be better accomplished if we can learn extra information from appropriately collected training data. In this paper, we propose to use the convolutional neural network (CNN) to train a SICE enhancer. One key issue is how to construct a training data set of low-contrast and high-contrast image pairs for end-to-end CNN learning. To this end, we build a large-scale multi-exposure image data set, which contains 589 elaborately selected high-resolution multi-exposure sequences with 4,413 images. Thirteen representative multi-exposure image fusion and stack-based high dynamic range imaging algorithms are employed to generate the contrast enhanced images for each sequence, and subjective experiments are conducted to screen the best quality one as the reference image of each scene. With the constructed data set, a CNN can be easily trained as the SICE enhancer to improve the contrast of an under-/over-exposure image. Experimental results demonstrate the advantages of our method over existing SICE methods with a significant margin.

Learning a Deep Single Image Contrast Enhancer from Multi-Exposure Images

Inferring a high dynamic range (HDR) image from a single low dynamic range (LDR) input is an ill-posed problem where we must compensate lost data caused by under-/over-exposure and color quantization. To tackle this, we propose the first deep-learning-based approach for fully automatic inference using convolutional neural networks. Because a naive way of directly inferring a 32-bit HDR image from an 8-bit LDR image is intractable due to the difficulty of training, we take an indirect approach; the key idea of our method is to synthesize LDR images taken with different exposures (i.e., bracketed images) based on supervised learning, and then reconstruct an HDR image by merging them. By learning the relative changes of pixel values due to increased/decreased exposures using 3D deconvolutional networks, our method can reproduce not only natural tones without introducing visible noise but also the colors of saturated pixels. We demonstrate the effectiveness of our method by comparing our results not only with those of conventional methods but also with ground-truth HDR images.

Deep reverse tone mapping

In typical cameras the optical system is designed first; once it is fixed, the parameters in the image processing algorithm are tuned to get good image reproduction. In contrast to this sequential design approach, we consider joint optimization of an optical system (for example, the physical shape of the lens) together with the parameters of the reconstruction algorithm. We build a fully-differentiable simulation model that maps the true source image to the reconstructed one. The model includes diffractive light propagation, depth and wavelength-dependent effects, noise and nonlinearities, and the image post-processing. We jointly optimize the optical parameters and the image processing algorithm parameters so as to minimize the deviation between the true and reconstructed image, over a large set of images. We implement our joint optimization method using autodifferentiation to efficiently compute parameter gradients in a stochastic optimization algorithm. We demonstrate the efficacy of this approach by applying it to achromatic extended depth of field and snapshot super-resolution imaging.

/pdf/end-to-end-optimization-of-optics-and-image-processing-for-3zecvm245u.pdf

End-to-end optimization of optics and image processing for achromatic extended depth of field and super-resolution imaging

This paper proposes the first non-flow-based deep framework for high dynamic range (HDR) imaging of dynamic scenes with large-scale foreground motions. In state-of-the-art deep HDR imaging, input images are first aligned using optical flows before merging, which are still error-prone due to occlusion and large motions. In stark contrast to flow-based methods, we formulate HDR imaging as an image translation problem without optical flows. Moreover, our simple translation network can automatically hallucinate plausible HDR details in the presence of total occlusion, saturation and under-exposure, which are otherwise almost impossible to recover by conventional optimization approaches. Our framework can also be extended for different reference images. We performed extensive qualitative and quantitative comparisons to show that our approach produces excellent results where color artifacts and geometric distortions are significantly reduced compared to existing state-of-the-art methods, and is robust across various inputs, including images without radiometric calibration.

/pdf/deep-high-dynamic-range-imaging-with-large-foreground-obzkepxmc1.pdf

Deep High Dynamic Range Imaging with Large Foreground Motions

Ghosting artifacts caused by moving objects or misalignments is a key challenge in high dynamic range (HDR) imaging for dynamic scenes. Previous methods first register the input low dynamic range (LDR) images using optical flow before merging them, which are error-prone and cause ghosts in results. A very recent work tries to bypass optical flows via a deep network with skip-connections, however, which still suffers from ghosting artifacts for severe movement. To avoid the ghosting from the source, we propose a novel attention-guided end-to-end deep neural network (AHDRNet) to produce high-quality ghost-free HDR images. Unlike previous methods directly stacking the LDR images or features for merging, we use attention modules to guide the merging according to the reference image. The attention modules automatically suppress undesired components caused by misalignments and saturation and enhance desirable fine details in the non-reference images. In addition to the attention model, we use dilated residual dense block (DRDB) to make full use of the hierarchical features and increase the receptive field for hallucinating the missing details. The proposed AHDRNet is a non-flow-based method, which can also avoid the artifacts generated by optical-flow estimation error. Experiments on different datasets show that the proposed AHDRNet can achieve state-of-the-art quantitative and qualitative results.

/pdf/attention-guided-network-for-ghost-free-high-dynamic-range-1pjn90wpam.pdf

Attention-Guided Network for Ghost-Free High Dynamic Range Imaging

Camera sensors can only capture a limited range of luminance simultaneously, and in order to create high dynamic range (HDR) images a set of different exposures are typically combined. In this paper we address the problem of predicting information that have been lost in saturated image areas, in order to enable HDR reconstruction from a single exposure. We show that this problem is well-suited for deep learning algorithms, and propose a deep convolutional neural network (CNN) that is specifically designed taking into account the challenges in predicting HDR values. To train the CNN we gather a large dataset of HDR images, which we augment by simulating sensor saturation for a range of cameras. To further boost robustness, we pre-train the CNN on a simulated HDR dataset created from a subset of the MIT Places database. We demonstrate that our approach can reconstruct high-resolution visually convincing HDR results in a wide range of situations, and that it generalizes well to reconstruction of images captured with arbitrary and low-end cameras that use unknown camera response functions and post-processing. Furthermore, we compare to existing methods for HDR expansion, and show high quality results also for image based lighting. Finally, we evaluate the results in a subjective experiment performed on an HDR display. This shows that the reconstructed HDR images are visually convincing, with large improvements as compared to existing methods.

HDR image reconstruction from a single exposure using deep CNNs

FovVideoVDP is a video difference metric that models the spatial, temporal, and peripheral aspects of perception. While many other metrics are available, our work provides the first practical treatment of these three central aspects of vision simultaneously. The complex interplay between spatial and temporal sensitivity across retinal locations is especially important for displays that cover a large field-of-view, such as Virtual and Augmented Reality displays, and associated methods, such as foveated rendering. Our metric is derived from psychophysical studies of the early visual system, which model spatio-temporal contrast sensitivity, cortical magnification and contrast masking. It accounts for physical specification of the display (luminance, size, resolution) and viewing distance. To validate the metric, we collected a novel foveated rendering dataset which captures quality degradation due to sampling and reconstruction. To demonstrate our algorithm's generality, we test it on 3 independent foveated video datasets, and on a large image quality dataset, achieving the best performance across all datasets when compared to the state-of-the-art.

/pdf/fovvideovdp-a-visible-difference-predictor-for-wide-field-of-xim150z4n4.pdf

FovVideoVDP: a visible difference predictor for wide field-of-view video

Limited GPU performance budgets and transmission bandwidths mean that real-time rendering often has to compromise on the spatial resolution or temporal resolution (refresh rate). A common practice is to keep either the resolution or the refresh rate constant and dynamically control the other variable. But this strategy is non-optimal when the velocity of displayed content varies. To find the best trade-off between the spatial resolution and refresh rate, we propose a perceptual visual model that predicts the quality of motion given an object velocity and predictability of motion. The model considers two motion artifacts to establish an overall quality score: non-smooth (juddery) motion, and blur. Blur is modeled as a combined effect of eye motion, finite refresh rate and display resolution. To fit the free parameters of the proposed visual model, we measured eye movement for predictable and unpredictable motion, and conducted psychophysical experiments to measure the quality of motion from 50 Hz to 165 Hz. We demonstrate the utility of the model with our on-the-fly motion-adaptive rendering algorithm that adjusts the refresh rate of a G-Sync-capable monitor based on a given rendering budget and observed object motion. Our psychophysical validation experiments demonstrate that the proposed algorithm performs better than constant-refresh-rate solutions, showing that motion-adaptive rendering is an attractive technique for driving variable-refresh-rate displays.

A perceptual model of motion quality for rendering with adaptive refresh-rate and resolution

Rendering in virtual reality (VR) requires substantial computational power to generate 90 frames per second at high resolution with good-quality antialiasing. The video data sent to a VR headset requires high bandwidth, achievable only on dedicated links. In this paper we explain how rendering requirements and transmission bandwidth can be reduced using a conceptually simple technique that integrates well with existing rendering pipelines. Every even-numbered frame is rendered at a lower resolution, and every odd-numbered frame is kept at high resolution but is modified in order to compensate for the previous loss of high spatial frequencies. When the frames are seen at a high frame rate, they are fused and perceived as high-resolution and high-frame-rate animation. The technique relies on the limited ability of the visual system to perceive high spatio-temporal frequencies. Despite its conceptual simplicity, correct execution of the technique requires a number of non-trivial steps: display photometric temporal response must be modeled, flicker and motion artifacts must be avoided, and the generated signal must not exceed the dynamic range of the display. Our experiments, performed on a high-frame-rate LCD monitor and OLED-based VR headsets, explore the parameter space of the proposed technique and demonstrate that its perceived quality is indistinguishable from full-resolution rendering. The technique is an attractive alternative to reprojection and resolution reduction of all frames.

/pdf/temporal-resolution-multiplexing-exploiting-the-limitations-1bwaflffjn.pdf

Gyorgy Denes

Papers

HDR image reconstruction from a single exposure using deep CNNs

HDR image reconstruction from a single exposure using deep CNNs

FovVideoVDP: a visible difference predictor for wide field-of-view video

A perceptual model of motion quality for rendering with adaptive refresh-rate and resolution

Temporal Resolution Multiplexing: Exploiting the limitations of spatio-temporal vision for more efficient VR rendering