This chapter concerns the visual saliency and the perceptual quality assessment of 3D meshes. Firstly, the chapter proposes a definition of visual saliency and describes the state-of-the-art methods for its detection on 3D mesh surfaces. A focus is made on a recent model of visual saliency detection for 3D colored and non-colored meshes whose results are compared with a ground-truth saliency as well as with the literature's methods. Since this model is able to estimate the visual saliency on 3D colored meshes, named colorimetric saliency, a description of the construction of a 3D colored mesh database that was used to assess its relevance is presented. The authors also describe three applications of the detailed model that respond to the problems of viewpoint selection, adaptive simplification and adaptive smoothing. Secondly, two perceptual quality assessment metrics for 3D non-colored meshes are described, analyzed, and compared with the state-of-the-art approaches.
TopVisual Saliency Of 3D Meshes: Introduction
In every look thrown at a scene or an object, visual attention is fixed on particular regions distinct from their surrounding zones. These striking areas, essentially prominent in the field of 3D objects, are content dependent. However, they are not dependent of the behavior or the experience relative to the human observer (Itti and Koch, 1998). This faculty of fixation is known as visual attention. The saliency computation would allow detecting these perceptually important regions that guide the visual attention. Visual saliency approaches proposed in the state-of-the-art are inspired from low level features of the Human Visual System (HVS). This permits to replace the geometric attributes used for the saliency computation by perceptual ones, and, as confirmed in Kim et al. (2010), these perceptual models succeed in modeling correctly the eye movements of the human observer.
Many applications in the field of 3D computer vision rely on the detection of visual saliency phenomena such as: 1) Automatic viewpoint selection (Lee et al., 2005) that aims at generating the most informative and attractive viewpoint. 2) Adaptive simplification (Shilane & Funkhouser, 2007) where the goal is to more simplify the non-salient areas for the visual quality preservation. Similarly, other applications such as Surface Matching (Gal and Cohen-or, 2006), automatic resizing (Jia, Zhang, and Zhou, 2014), facial recognition (Lee et al., 2004), etc. benefit from visual saliency.
In the following, a definition of visual saliency is proposed. Then, a description of the physiological elements that are involved in the sensitivity of the HVS to visual saliency is provided. Afterwards, different approaches proposed in the state-of-the-art that predict visual saliency are analyzed. This chapter focuses also on a recent approach proposed by Nouri et al. (2015a) that takes into account two characteristics of the Human Visual System (HVS) which are the sensitivity to strong fluctuations and high contrast. For this, a local vertex descriptor in the form of a local adaptive patch is introduced to characterize the 3D surface mesh. This descriptor is used as a basis for similarity measurement and integrated into a weighted multi-scale saliency computation allowing the enhancement of the quality measure and the robustness to the noise. Qualitative and quantitative comparisons with a pseudo ground truth as well as a comparison with the sate-of-the-art methods are achieved to assess its relevance. Finally, three applications guided by the detailed model of saliency are analyzed.