A Study On Tone Mapping For High Dynamic Range Images And Its Application To Low Dynamic Range Displays And Backlight Scaling

Tone mapping(TM) is an essential step for the reproduction of “nice looking”images. It can map the luminances of the original scene to the output device’s displayable values. When the dynamic range of the captured scene is larger or smaller than that of the display device, tone mapping compresses or expends the luminance ratios. High dynamic range(HDR) is a very attractive way of capturing real word appearance, since it permits the preservation of complete information on luminance(radiance) values in the scene for each pixel. In a scene,the luminance ratio between the brightest and the darkest pixel value is called dynamic range. In the real world, the range of light is extremely large, which could be approximately ten orders of absolute range from star-lit scenes to sun-lit snow, and from shadow to highlight in a single scene, the range could reach over four orders of dynamic range. However, the range of light we can reproduce on our printer and screen display devices spans at best about two orders of absolute dynamic range. The tone reproduction problem: is derived from this discrepancy: how should we map scene luminance to display luminance with preserving the contrasts and detail in the dark and bright regions as much as possible. With standard displays, the dynamic range of the captured HDR scene must be compressed significantly, which can include a loss of contrast resulting in a loss of detail visibility. Our thesis has launched deep study on tone mapping algorithms and has mainly undertaken our tasks over following areas:1. The key thought of tone mapping is discussed, several outstanding global tone mapping and local tone mapping methods are described, and their advantages and disadvantages are analyzed.2. Detail-preserving tone mapping for low dynamic range displays with adaptive gamma correction method are proposed. The global mapping method compresses the luminance of each pixel using a fixed curve, which includes logarithmic transformation, gamma correction, histogram equalization, and linear tone mapping. Logarithmic transformation can preserve the image appearance well, close to human visual system. However, this global method preserves overall image quality at the cost of localinformation. In our method, firstly we used the logarithmic transformation to compress the original dynamic range, then applied the adaptive gamma correction, whose gamma parameters were determined by the local contrasts, to increase local contrast and thus improve detail visibility. Through experiments, we have proved that our method has preserved overall image quality as well as enhanced image detail well.3. Hybrid tone mapping method for high dynamic range images with adaptive dodging & burning is put forward. Generally speaking, global methods reproduce overall quality image attributes well, and they are fast to compute and easy to implement, but may wash away important details. However, local approaches excel in reproduction of local contrast(details), but they are computationally intensive and may reproduce overall images attributes poorly. Inspired by the disadvantages and advantages of global and local TM methods, we proposed the hybrid high dynamic range compression with adaptive dodging & burning technique for low dynamic range displays. First, we use the image key-based linear mapping to preserve overall image quality, then apply adaptive dodging & burning to enhance image naturalness, final do local TM to improve local contrasts. The experiments have proved that the proposed method did not cause artifacts, but provided more natural image appearance.4. Backlight dimming based on the tone mapping is put forward. Global dynamic backlight dimming consists of two main steps: reduce backlight brightness and increase image gray levels. The pixel saturation rate is inversely proportional to the image quality, which is caused by the gray levels increasing. In the paper, we do the backlight dimming of liquid crystal display based on the tone mapping compression method. General speaking, the global backlight dimming is the tone mapping for low dynamic range images. Backlight dimming based on tone mapping method determines the maximum pixel value of increased image by image composition and image key-based compression. Various experiments have demonstrated that the proposed method preserved the overall contrast and local details while reducing power consumption as much as possible.