High-dynamic-range imaging

About: High-dynamic-range imaging is a research topic. Over the lifetime, 766 publications have been published within this topic receiving 22577 citations.

Fidelity evaluation of metallic luster in six-band high-dynamic-range imaging

Abstract: This paper presents subjective evaluation results for images reproducing metallic luster. The images were produced by a six-band high-dynamic-range imaging technique, which provides high color fidelity and the capability of various tone mapping operations. The preferences for images produced by four kinds of tone mapping operations were evaluated by means of a paired comparison method under the conditions that observers could or could not refer to real scenes. The evaluation results indicated that preferences for images depend on tone mapping operations, especially for objects that have the luster changing over a large area with moderately high intensity. In addition, when referring to a real scene, the spatially varying tone compression, which reproduces the color and complex characteristics of metallic luster, was significantly preferable to the tone mappings by other approaches. Without a real scene, the preference was also strong for images produced by simply saturating pixel values at highlights.

Fast exposure time decision in multi-exposure HDR imaging

Abstract: Currently available imaging and display system exists the problem of insufficient dynamic range, and the system cannot restore all the information for an high dynamic range (HDR) scene. The number of low dynamic range(LDR) image samples and fastness of exposure time decision impacts the real-time performance of the system dramatically. In order to realize a real-time HDR video acquisition system, this paper proposed a fast and robust method for exposure time selection in under and over exposure area which is based on system response function. The method utilized the monotony of the imaging system. According to this characteristic the exposure time is adjusted to an initial value to make the median value of the image equals to the middle value of the system output range; then adjust the exposure time to make the pixel value on two sides of histogram be the middle value of the system output range. Thus three low dynamic range images are acquired. Experiments show that the proposed method for adjusting the initial exposure time can converge in two iterations which is more fast and stable than average gray control method. As to the exposure time adjusting in under and over exposed area, the proposed method can use the dynamic range of the system more efficiently than fixed exposure time method.

Learning Regularized Multi-Scale Feature Flow for High Dynamic Range Imaging

Abstract: Reconstructing ghosting-free high dynamic range (HDR) images of dynamic scenes from a set of multi-exposure images is a challenging task, especially with large object motion and occlusions, leading to visible artifacts using existing methods. To address this problem, we propose a deep network that tries to learn multi-scale feature flow guided by the regularized loss. It first extracts multi-scale features and then aligns features from non-reference images. After alignment, we use residual channel attention blocks to merge the features from different images. Extensive qualitative and quantitative comparisons show that our approach achieves state-of-the-art performance and produces excellent results where color artifacts and geometric distortions are significantly reduced.

Detecting and composing near-identical HDR images without exposure information

Abstract: In high dynamic range (HDR) imaging, two essential problems are to compose HDR image from conventional image set without any prior information about their exposures, and to access the synthesis result. To solve these problems, we first develop an exposure ratio estimation algorithm based on intensity mapping function (IMF). Then, we introduce an HDR image comparison method to verify whether two HDR images are from the same scene by using their log histogram similarity. Even though the images carrying the same information, their similarity cannot be detected by pixel-wise comparisons. We name such a pair of HDR images as near-identical images. According to experiments, our detection method is able to identify near-identical HDR images effectively, and our synthesis algorithm is able to recover the correct exposure ratios and compose near-identical HDR images.

High Dynamic Range imaging: is the game worth the candle?

Abstract: High dynamic range imaging (HDRi) is a technology concerned with representing a range of luminances larger than state of the art displays and closer to luminance ranges occuring in natural scenes. We investigate whether the approach of evaluating a dynamic range by the number of just noticeable differences contained within does make sense in a HDRi workflow. We found that disturbing effects of neighboring luminances were hardly perceivable on a standard dynamic range display if the background luminance exceeds 5 cd/m2. Introduction Representation in imaging media is limited by the spatial resolution, the temporal resolution and the sets of available colors, named gamut. Since the human visual system has a higher ability to distinguish luminance differences than chromatic differences on a spatial scale [1, 2, 3], the luminance information transfers the structure of an image representing a scene [4, 5]. The luminance range of an image is defined by the difference of the highest to the lowest luminance contained in it, while luminance contrast is defined by ratios of them. Contemporary standard displays achieve peak luminances between 80 and 400 cd/m2. In print, the reflectance of the paper can be reduced to 0.5% to 10% , depending on the paper type. Within this dynamic range, a number of luminance steps can be made out by the human visual system. The number of these luminance steps is the dynamic resolution of a visual system. The number of representable levels in an imaging system should be higher than the dynamic resolution, levels are often represented in the binary system, e.g. 256 representable levels are expressed as log2(256) = 8bit (Figure 1). Distinction of gray levels is of particular interest in medical applications, since many medical imaging methods (tomography, mammography) deliver achromatic pictures only [6]. Taking the data on which the DICOM standard is based and assuming a peak luminance of 4000 cd/m2, approximately 210 = 1024 gray levels can be simultaneously differentiated under optimal conditions [7]. However, in images, neighboring high intensity regions can deteriorate the contrast detection performance. This is due to glare effects in the eye [8, 9]. Reflections of ambient light at a display surface can significantly influence contrasts in low intensity regions [10]. Since contrasts in low intensity regions of an image tend to be of importance for the diagnostic purpose, the studies of medical displays therefore often use very low luminance levels of background intensity and room illumination [9]. Glare effects have been quantitatively investigated in the first half of the 20th century with the equivalent background technique [11, pg. 578]. The Stiles-Holladay equation states that the influence of a point glare source E at eccentricity θ on an increment contrast threshold for a background luminance L leads to an equivalent increment contrast threshold as a background luminance β : β = L+10 E θ 2 , θ > 0.5deg . (1) Assuming independent contributions, glare sources are additive. The Stiles–Holladay equation has later been extended for age effects, influence of ocular pigmentation and eccentricity angles larger than 30◦ [8, 12]. We were not interested in adding to this theory, but to test whether glare effects can be reproduced on displays when background luminance exceeds 5 cd/m2. For this, we simulated a standard dynamic range display with a peak luminance below 200 cd/m2 on a high dynamic range display with a peak luminance exceeding 2000 cd/m2. Figure 1. A particular spatial arrangement of the 256 luminance steps available in 8bit sRGB. While one should hardly see any horizontal borders between adjacent squares represented in print or on standard displays, almost each border is clearly visible on a display with a high luminance range when the maximal luminance is chosen as a white point. 21st Color and Imaging Conference Final Program and Proceedings 231