Visual attention in LDR and HDR images

Recent advances in high dynamic range (HDR) capturing and display technologies attracted a lot of interest to HDR imaging. Many issues that are considered as being resolved for conventional low dynamic range (LDR) images pose new challenges in HDR context. One of such issues is human visual attention, which has important applications in image and video compression, camera and displays manufacturing, artistic content creation, and advertisement. However, the impact of HDR imaging on visual attention and on the performance of saliency models is not well understood. Therefore, in this paper, we address this problem by creating a publicly available dataset of 46 HDR and corresponding LDR images with varying regions of interests, scenes, and dynamic range. We conducted eye tracking experiments and obtained fixation density maps, which demonstrate a significant difference in the way HDR and LDR capture attention of the observers.

Tone mapping for high dynamic range images

Laurence Meylan

Tone mapping is an essential step for the reproduction of "nice looking" images. It provides the mapping between the luminances of the original scene to the output device's display values. When the dynamic range of the captured scene is smaller or larger than that of the display device, tone mapping expands or compresses the luminance ratios. We address the problem of tone mapping high dynamic range (HDR) images to standard displays (CRT, LCD) and to HDR displays. With standard displays, the dynamic range of the captured HDR scene must be compressed significantly, which can induce a loss of contrast resulting in a loss of detail visibility. Local tone mapping operators can be used in addition to the global compression to increase the local contrast and thus improve detail visibility, but this tends to create artifacts. We developed a local tone mapping method that solves the problems generally encountered by local tone mapping algorithms. Namely, it does not create halo artifacts, nor graying-out of low contrast areas, and provides good color rendition. We then investigated specifically the rendition of color and confirmed that local tone mapping algorithms must be applied to the luminance channel only. We showed that the correlation between luminance and chrominance plays a role in the appearance of the final image but a perfect decorrelation is not necessary. Recently developed HDR monitors enable the display of HDR images with hardly any compression of their dynamic range. The arrival of these displays on the market create the need for new tone mapping algorithms. In particular, legacy images that were mapped to SDR displays must be re-rendered to HDR displays, taking best advantage of the increase in dynamic range. This operation can be seen as the reverse of the tone mapping to SDR. We propose a piecewise linear tone scale function that enhances the brightness of specular highlights so that the sensation of naturalness is improved. Our tone scale algorithm is based on the segmentation of the image into its diffuse and specular components as well as on the range of display luminance that is allocated to the specular component and the diffuse component, respectively. We performed a psychovisual experiment to validate the benefit of our tone scale. The results showed that, with HDR displays, allocating more luminance range to the specular component than what was allocated in the image rendered to SDR displays provides more natural looking images.

EPFL2006

Impact of Ultra High Definition on Visual Attention

Touradj Ebrahimi, Philippe Hanhart, Hiromi Nemoto

Ultra high definition (UHD) TV is rapidly replacing high definition (HD) TV but little is known of its effects on human visual attention. However, a clear understanding of this effect is important, since accurate models, evaluation methodologies, and metrics for visual attention are essential in many areas, including image and video compression, camera and displays manufacturing, artistic content creation, and advertisement. In this paper, we address this problem by creating a dataset of UHD resolution images with corresponding eye-tracking data, and we show that there is a statistically significant difference between viewing strategies when watching UHD and HD contents. Furthermore, by evaluating five representative computational models of visual saliency, we demonstrate the decrease in models' accuracies on UHD contents when compared to HD contents. Therefore, to improve the accuracy of computational models for higher resolutions, we propose a segmentation-based resolution-adaptive weighting scheme. Our approach demonstrates that taking into account information about resolution of the images improves the performance of computational models.

2014

Tone mapping for high dynamic range images

Laurence Meylan

EPFL2006