Saliency Maps of High Dynamic Range Images

Abstract

A number of computational models of visual attention have been proposed based on the concept of saliency map, most of them validated using oculometric data. They are widely used for Computer Graphics applications with Low Dynamic Range images, mainly for image rendering, in order to avoid spending too much computing time on non salient areas. However, these algorithms were not used so far with High Dynamic Range (HDR) inputs. In this paper, we show that in the case of HDR images, the predictions using algorithms based on [Itti and Koch 2000] are less accurate than with 8-bit images. To improve the saliency computation for HDR inputs, we propose a new algorithm derived from [Itti and Koch 2000]. From an eye tracking experiment with a HDR scene, we show that this algorithm leads to good results for the saliency map computation, with a better fit between the saliency map and the ocular fixation map than Itti's algorithm.

Automated parameter tuning for tone mapping using visual saliency

“…If they are applied to HDR images directly, the high dynamic range will be scaled, which may cause the distortion of HDR content that in turn makes salient regions appear not salient or vice versa. Brémond et al [8] propose an improved saliency analysis model for HDR images by normalizing intensity and orientation features. Using virtual photograph techniques, Gao et al [18] present a general framework for applying visual attention models to HDR images.…”

Attention-based high dynamic range imaging

“…We adopt the HDR saliency map [2] as our transient attention model. HDR saliency map is developed based on the saliency map [14], in which Itti et al construct Gaussian pyramids I (σ ) with nine spatial scales σ ∈ [0..8] to simulate the bottom-up process of transient attention in the human visual system.…”

“…To find the attentive and nonattentative regions in an HDR image, we adopt the HDR saliency map proposed by Bré-mond et al [2] since it can better predict the attentive regions Fig. 2 (a) Plot of Eq.…”

“…We adopt the HDR saliency map proposed by Brémond et al [2] to obtain an attention map, which can predict the attentive and nonattentative regions of an HDRI. Once the attentive and nonattentative regions are detected, we then locally adjust the tone mapping function according to the contrast response model found by Pestilli et al [24].…”

Virtual photograph based saliency analysis of high dynamic range images

“…Although existing visual attention systems can lead to satisfying results for LDR images, they do not always perform well when applied to HDR images [Brémond et al 2012]. Compared with LDR images, HDR images allow a much higher dynamic range to represent luminance in the real world.…”