Showing papers on "Tone mapping published in 2003"

Adaptive Logarithmic Mapping For Displaying High Contrast Scenes

Abstract: We propose a fast, high quality tone mapping technique to display high contrast images on devices with limited dynamic range of luminance values. The method is based on logarithmic compression of luminance values, imitating the human response to light. A bias power function is introduced to adaptively vary logarithmic bases, resulting in good preservation of details and contrast. To improve contrast in dark areas, changes to the gamma correction procedure are proposed. Our adaptive logarithmic mapping technique is capable of producing perceptually tuned images with high dynamic content and works at interactive speed. We demonstrate a successful application of our tone mapping technique with a high dynamic range video player enabling to adjust optimal viewing conditions for any kind of display while taking into account user preference concerning brightness, contrast compression, and detail reproduction.

A multigrid solver for boundary value problems using programmable graphics hardware

Abstract: We present a case study in the application of graphics hardware to general-purpose numeric computing. Specifically, we describe a system, built on programmable graphics hardware, able to solve a variety of partial differential equations with complex boundary conditions. Many areas of graphics, simulation, and computational science require efficient techniques for solving such equations. Our system implements the multigrid method, a fast and popular approach to solving large boundary value problems. We demonstrate the viability of this technique by using it to accelerate three applications: simulation of heat transfer, modeling of fluid mechanics, and tone mapping of high dynamic range images. We analyze the performance of our solver and discuss several issues, including techniques for improving the computational efficiency of iterative grid-based computations for the GPU.

High Dynamic Range from Multiple Images: Which Exposures to Combine?∗

Abstract: Many computer vision algorithms rely on precise estimates of scene radiances obtained from an image. A simple way to acquire a larger dynamic range of scene radiances is by combining several exposures of the scene. The number of exposures and their values have a dramatic impact on the quality of the combined image. At this point, there exists no principled method to determine these values. Given a camera with known response function and dynamic range, we wish to find the exposures that would result in a set of images that when combined would emulate an effective camera with a desired dynamic range and a desired response function. We first prove that simple summation combines all the information in the individual exposures without loss. We select the exposures by minimizing an objective function that is based on the derivative of the response function. Using our algorithm, we demonstrate the emulation of cameras with a variety of response functions, ranging from linear to logarithmic. We verify our method on several real scenes. Our method makes it possible to construct a table of optimal exposure values. This table can be easily incorporated into a digital camera so that a photographer can emulate a wide variety of high dynamic range cameras by selecting from a menu. 1 Capturing a Flexible Dynamic Range Many computer vision algorithms require accurate estimates of scene radiance such as color constancy [9], inverse rendering [13, 1] and shape recovery [17, 8, 18]. It is difficult to capture both the wide range of radiance values real scenes produce and the subtle variations within them using a low cost digital camera. This is because any camera must assign a limited number of brightness values to the entire range of scene radi∗This work was completed with support from a National Science Foundation ITR Award (IIS-00-85864) and a grant from the Human ID Program: Flexible Imaging Over a Wide Range of Distances Award No. N000-14-00-1-0929 (a) Small and large exposures combine to capture a high dynamic range (b) Similar exposures combine to capture suble variations Figure 1: Illustration showing the impact of the choice of exposure values on which scene radiances are captured. (a) When large and small exposures are combined the resulting image has a high dynamic range, but does not capture some scene variations. (b) When similar exposure values are combined, the result includes subtle variations, but within a limited dynamic range. In both cases, a set of exposures taken with a camera results in an “effective camera.” Which exposures must we use to emulate a desired effective camera? ances. The response function of the camera determines the assignment of brightness to radiance. The response therefore determines both the camera’s sensitivity to changes in scene radiance and its dynamic range. A simple method for extending the dynamic range of a camera is to combine multiple images of a scene taken with different exposures [6, 2, 3, 10, 11, 12, 15, 16]. For example, the left of Fig. 1(a) shows a small and a large exposure, each capturing a different range of scene radiances. The illustration on the right of Fig. 1(a) shows that the result of combining the exposures includes the entire dynamic range of the scene. Note that by using these exposures values we fail to capture subtle variations in the scene, such as the shading of the ball. Once these variations are lost they can not be restored by methods that change the brightness of an image, such as the recent work on tone mapping [4, 5, 14]. In Fig. 1(b), two similar exposures combine to produce an image that captures subtle variations, but within a limited dynamic range. As a result, in both Fig. 1(a) and (b), the images on the right can be considered as the outputs of two different “effective cameras.” The number and choice of exposures determines the dynamic range and the response of each effective camera. This relationship has been ignored in the past. In this paper we explore this relationship to address the general problem of determining which exposure values to use in order to emulate an effective camera with a desired response and a desired dynamic range. Solving this problem requires us to answer the following questions: • How can we create a combined image that preserves the information from all the exposures? Previous work suggested heuristics for combining the exposures [3, 11, 12]. We prove that even without linearizing the camera, simple summation preserves all the information contained in the set of individual exposures. • What are the best exposure values to achieve a desired effective response function for the combined image? It is customary to arbitrarily choose the number of exposures and the ratio (say, 2) between consecutive exposure values [3, 10, 11, 12]. For example, when this is done with a linear real camera, the resulting combined image is relatively insensitive to changes in large radiances. This can bias vision algorithms that use derivatives of radiance. Such biases are eliminated using our algorithm, which selects the exposure values to best achieve a desired response. • How can we best achieve a desired dynamic range and effective response function from a limited number of images? It is common to combine images with consecutive exposure ratios of 2 (see [3, 11, 12]). to create a high dynamic range image. With that choice of exposure ratio, is often necessary to use 5 or more exposures to capture the full dynamic range of a scene. This is impractical when the number of exposures that can be captured is limited by the time to acquire the images, changes in the scene, or resources needed to process the images. Our algorithm determines the exposure values needed to best emulate a desired camera with a fixed number of images. Our method allows us to emulate cameras with a wide variety of response functions. For the class of linear real cameras, we present a table of optimal exposure values for emulating high dynamic range cameras with, for example, linear and logarithmic (constant contrast) responses. Such a table can be easily incorporated into a digital camera so that a photographer can select his/her desired dynamic range and camera response from a menu. In other words, a camera with fixed response and dynamic range can be turned into one that has a “flexible” dynamic range. We show several experimental results using images of real scenes that demonstrate the power of this notion of flexible dynamic range. 2 The Effective Camera When we take multiple exposures of the same scene, each exposure adds new information about the radiance values in the scene. In this section, we create an effective camera by constructing a single image which retains all the information from the individual exposures. By information we mean image brightness values which represent measurements of scene radiance. Scene radiance is proportional to image irradiance E [7]. In a digital camera, the camera response function f jumps from one image brightness value B to the next at a list of positive irradiance values (shown below the graph in Fig. 2) which we call the measured irradiance levels. An image brightness value indicates that the corresponding measured irradiance lies in the interval between two of these levels. Hence, without loss of generality, we define B as the index of the first of these two levels, EB , so that f(EB) = B. Hence, the response function is equivalent to the list of measured irradiance levels. Now, consider the measured irradiance levels using unit exposure e1 = 1 with a real non-linear camera having 4 brightness levels. These levels are shown on the bar at the bottom of Fig. 3(a). The irradiance levels for a second exposure scale by 1/e2, as shown in Fig. 3(b). We combine the measured irradiance levels from the first and the second exposures by taking the union of all the 1The value we call exposure accounts for all the attenuations of light by the optics. One can change the exposure by changing a filter on the lens, the aperture size, the integration time, or the gain. 2Note that the slope of the response function determines the density of the levels, as shown by the short line segment in Fig. 2. 3Note that the number of exposures and brightness levels are for illustration only. Our arguments hold in general.

High dynamic range image viewing on low dynamic range displays

Abstract: Techniques and tools for displaying/viewing HDR images are described. In one aspect, a background image constructed from HDR image information is displayed along with portions of the HDR image corresponding to one or more regions of interest. The portions have at least one display parameter (e.g., a tone mapping parameter) that differs from a corresponding display parameter for the background image. Regions of interest and display parameters can be determined by a user (e.g., via a GUI). In another aspect, an intermediate image is determined based on image data corresponding to one or more regions of interest of the HDR image. The intermediate image has a narrower dynamic range than the HDR image. The intermediate image or a derived image is then displayed. The techniques and tools can be used to compare, for example, different tone mappings, compression methods, or color spaces in the background and regions of interest.

Rendering HDR images

Abstract: Color imaging systems are continuously improving, and have now improved to the point of capturing high dynamic range scenes. Unfortunately most commercially available color display devices, such as CRTs and LCDs, are limited in their dynamic range. It is necessary to tone-map, or render, the high dynamic range images in order to display them onto a lower dynamic range device. This paper describes the use of an image appearance model, iCAM, to render high dynamic range images for display. Image appearance models have greater flexibility over dedicated tone-scaling algorithms as they are designed to predict how images perceptually appear, and not designed for the singular purpose of rendering. In this paper we discuss the use of an image appearance framework, and describe specific implementation details for using that framework to render high dynamic range images.

Perceptual evaluation of tone mapping operators

Abstract: High dynamic range (HDR) photography and physically based rendering produce images with full range of luminance data. The problem of mapping real world luminance into the limited luminance range of display devices has been addressed in the last decade through the development of many different tone mapping algorithms (see [Devlin et al. 2002] for a recent survey). Clearly, a sound methodology for comparing existing algorithms is needed to understand their strengths and weaknesses; this work takes a first exploratory step towards achieving this goal. We performed a series of psychophysical experiments in which human subjects assessed their perceptions associated with a set of 24 images, constructed by submitting four different scenes (both synthetic and photographic) to six popular tone mapping operators: Photographic Tone Reproduction (Reinhard et al. 2002), Uniform Scaling Quantization (Schlick 1994), Retinex (Frankle et al. 1983), Visual Adjustment (Ferwerda 1996), Revised Tumblin-Rushmeier (1993), and Histogram Adjustment (Ward et al. 1997; the human contrast sensitivity option was switched off). In this initial attempt, we chose an exploratory rather than confirmatory approach in which subjects first made global judgments regarding how perceptually similar or dissimilar the images were, without specifying the ways in which they might differ from one another. Global dissimilarity judgments were made for all pairwise comparisons of the six tone mapping operators separately for each of the four scenes by each of 11 subjects, and these data were submitted to INdividual DifferencesSCALing (INDSCAL) analysis. The primaryINDSCAL result of interest is a derivedStimulus Spacein which each stimulus is assigned coordinates on the dimensions describing the perceptual differences between the stimuli (see [Borg and Groenen 1997] for background on this analysis method). The spatial configuration of the six tone mappers on the two most salient dimensions is shown in Figure 1.

Method and apparatus for transforming a high dynamic range image into a low dynamic range image

Abstract: A method and an apparatus for transforming a high dynamic range image into a low dynamic range image. The method includes converting first luminance values associated with pixels into a plurality of second luminance values, and utilizing a film transfer function for mapping the second luminance values associated with the pixels into a plurality of third luminance values to generate the low dynamic range image. A second luminance range of the second luminance values is smaller than a first luminance range of the first luminance values, and the film transfer function adds no visual artifact to the low dynamic range image.

Segmentation and adaptive assimilation for detail-preserving display of high-dynamic range images

Abstract: Realistic display of high-dynamic range images is a difficult problem. Previous methods for high-dynamic range image display suffer from halo artifacts or are computationally expensive. We present a novel method for computing local adaptation luminance that can be used with several different visual adaptation-based tone-reproduction operators for displaying visually accurate high-dynamic range images. The method uses fast image segmentation, grouping, and graph operations to generate local adaptation luminance. Results on several images show excellent dynamic range compression, while preserving detail without the presence of halo artifacts. With adaptive assimilation, the method can be configured to bring out a high-dynamic range appearance in the display image. The method is efficient in terms of processor and memory use.

A wide field, high dynamic range, stereographic viewer

Abstract: In this paper we present a High Dynamic Range viewer based on the 120-degree field-of-view LEEP (Large Expanse Extra Perspective) stereo optics used in the original NASA virtual reality systems. By combining these optics with an intense backlighting system (20 Kcd/m2) and layered transparencies, we are able to reproduce the absolute luminance levels and full dynamic range of almost any visual environment. This is important because it allows us to display environments with luminance levels that would not be displayable on a standard monitor. This technology may enable researchers to conduct controlled experiments in visual contrast, chromatic adaptation, and disability and discomfort glare without the usual limitations of dynamic range and field of view imposed by conventional CRT display systems. In this paper, we describe the basic system and techniques used to produce the transparency layers from a high dynamic range rendering or scene capture. We further present a series of psychophysical experiments demonstrating the device's ability to reproduce visual percepts, and compare this result to the real scene and a visibility matching tone reproduction operator presented on a conventional CRT display.

Optimal Tone Mapping

Abstract: Traditional autosegmental accounts of tone mapping invoke three independent factors: morphological category, tone quality, and a phonological directionality parameter. This article argues that the evidence for phonological directionality must be reconsidered. The article introduces a theory of Optimal Tone Mapping, in which attested patterns derive solely from the interaction of morphological directionality with quality-sensitive markedness constraints. The more restrictive theory of tone association that results from eliminating constraints that impose phonological directionality provides a new typology of tone melody languages.

Visualization of high dynamic range images

Abstract: A novel paradigm for information visualization in high dynamic range images is presented in this paper. These images, real or synthetic, have luminance with typical ranges many orders of magnitude higher than that of standard output/viewing devices, thereby requiring some processing for their visualization. In contrast with existent approaches, which compute a single image with reduced range, close in a given sense to the original data, we propose to look for a representative set of images. The goal is then to produce a minimal set of images capturing the information all over the high dynamic range data, while at the same time preserving a natural appearance for each one of the images in the set. A specific algorithm that achieves this goal is presented and tested on natural and synthetic data.

Design of a Tone Mapping Operator for High Dynamic Range Images Based upon Psychophysical Evaluation and Preference Mapping

Abstract: A tone mapping algorithm for displaying high contrast scenes was designed on the basis of the results of experimental tests using human subjects. Systematic perceptual evaluation of several existing tone mapping techniques revealed that the most "natural" appearance was determined by the presence in the output image of detailed scenery features often made visible by limiting contrast and by properly reproducing brightness. Taking these results into account, we developed a system to produce images close to the ideal preference point for high dynamic range input image data. Of the algorithms that we tested, only the Retinex algorithm was capable of retrieving detailed scene features hidden in high luminance areas while still preserving a good contrast level. This paper presents changes made to Retinex algorithm for processing high dynamic range images, and a further integration of the Retinex with specialized tone mapping algorithms that enables the production of images that appear as similar as possible to the viewer's perception of actual scenes.

Delivering interactivity to complex tone mapping operators

Abstract: The accurate display of high dynamic range images requires the application of complex tone mapping operators. These operators are computationally costly, which prevents their usage in interactive applications. We propose a general framework that delivers interactive performance to an important subclass of tone mapping operators, namely global tone mapping operators. The proposed framework consists of four steps: sampling the input image, applying the tone mapping operator, fitting the point-sampled tone mapping curve, and reconstructing the tone mapping curve for all pixels of the input image. We show how to make use of recent graphics hardware while keeping the advantage of generality by performing tone mapping in software. We demonstrate the capabilities of our method by accelerating several common global tone mapping operators and integrating the operators in a real-time rendering application.

High Dynamic Range Image Display Using Level Set Framework.

Abstract: This paper presents a novel solution to the High Dynamic Range (HDR) image compression problem using level set framework. Using level set framework, this method separates the HDR image into detail and profile. Then it uses a global tone mapping function to compress the profile. Finally it produces a low dynamic range version of the image for display by adding the details to the compressed profile. This method is capable of compressing the dynamic range while retaining the detail in the final image of various HDR images in a short time.

Robust Parameter Estimation for Tone Mapping Operator

Abstract: Tone Mapping Operators are used to compress a large range of pixel luminances into a smaller range that is suitable for display on devices with limited dynamic range. This work presents effective and easy to use Tone Mapping Operator based on last ideas in this direction. The estimation process uses sampling method. Essential attention was devoted to providing robustness of algorithm parameter estimation, especially critical for animation applications. The resulting operator produces good images and practically does not require manual parameter tuning.

Reproducing 3D Prints on Monitor by Relative-Glossiness Matching Technique.

Abstract: For over 3 decades, computer graphics technology has been developed to simulate physically accurate image of real scene. Meanwhile, useful tone mapping methods have been developed to map luminance range of the simulated image into that of usual monitor without perceptual distortion. However, it is difficult to reproduce accurate glossiness of real objects within the limited monitor luminance range. This also causes inaccurate glossiness sequence among several reproduced images. In this paper, relative-glossiness matching technique is proposed for reliable business to business (B to B) e-commerce system on 3D prints such as beverage cans, PET bottles, snack packages, and so on. The relative-glossiness-matching technique is introduced to preserve perceptual ratio of glossiness of the real 3D print objects in reproduced images. We also propose two operations to control surface-texture gloss and contrast gloss of reproduced images in Hunter’s classification for physical gloss; adding Gaussian noise to surface normal in rendering process and scaling specular reflection in tone mapping. Procedure of subjective evaluation to determine the standard deviation of Gaussian noise and scaling factor for specular reflection is described. An experiment for the relative-glossiness matched images is performed using four types of real papers shaped into 3D cylinders. It was visually confirmed that the reproduced images preserved the glossiness sequence of the real 3D cylinder.

High dynamic range image texture mapping based on VRML

Abstract: In this paper, we describe a technique for representing and displaying high dynamic range image (HDRI) as a texture map in VRML structure. To do this, we designed and implemented extended texture nodes for supporting HDRI in VRML and developed HDRI-based mapping tool that can be used for authoring objects with HDR image as a texture data. Even though the authoring result of our tool is not identical with standard VRML format exactly, it is good enough to be used in virtual world. To verify this, we developed an extended VRML browser that supports a parsing and viewing of our new texture nodes related to HDRI and allows the exposure to be adjusted continuously and arbitrarily at the time of navigation.

Faithful presentation of luminance contrast: evaluation of photographic and computational display methods

Abstract: Representations necessarily lose some of the visual information available in corresponding real-world scenes. This paper will discuss evaluations of the extent to which luminance contrast and visibility is preserved with three different methods for representing real-world scenes. Method one involves using psychophysical data from contrast charts to select the best print from among a density-varied series of photographic prints. The second and third methods involve extending the dynamic range of the representation by using High Dynamic Range Image (HDRI) techniques. HDRI's can be created by combining multiple overlapping exposures of a scene, or via computer simulation. In method two, algorithms are used to compress the luminance information in the HDRI into the luminance range available in the display, while preserving visible contrast as much as possible. The third method uses a wide-field, high-dynamic-range viewer to present an image with a much wider dynamic range than is available in a photographic print or a CRT display. Each method represents an improvement over simple photographic representation. In conjunction with appropriate instructions on how to interpret the images and the extent to which the images can be regarded as faithful, methods such as these can support practical decisions in visual design and reconstruction.

Reducing inaccuracies caused by tone mapping operations in dark areas of an image

Abstract: The invention provides method and apparatus for reducing inaccuracies in tone mapping of image pixel colour components in dark areas of an image. Pixel colour component mapping is performed only if the pixel is determined not to be in a dark area of the image, this being determined on the basis of comparison of the colour component with a threshold. Pixel values of pixels having colour components less than the threshold are preserved. Another embodiment determines a range around a threshold value, with interpolation being utilised to modify the colour component within the range. There are broader claims not requiring the image data to be colour component data, but the specification is directed towards processing of colour components.