Subjective Visual Quality Assessment of Immersive 3D Media Compressed by Open-Source Static 3D Mesh Codecs

Abstract

While studies for objective and subjective evaluation of the visual quality of compressed 3D meshes has been discussed in the literature, those studies were covering the evaluation of 3D-meshes that were created either by 3D artists or generated by a computationally expensive 3D reconstruction process applied on high quality 3D scans. With the advent of RGB-D sensors that operate at high frame-rates and the utilization of fast 3D reconstruction algorithms, humans can be captured and 3D reconstructed into a 3D mesh representation in real-time, enabling new (tele-)immersive experiences. The way of producing the respective 3D mesh content is dramatically different between the two cases, leading to apparent structural difference between the output meshes. On one hand, the first type of content is nearly perfect and clean, while on the other hand, the second type is much more irregular and noisy. Thus, evaluating compression artifacts on this new type of immersive 3D media, constitutes a yet unexplored scientific area. In this paper, we aim to subjectively assess the compression artifacts introduced by three open-source static 3D mesh codecs, when compressing 3D meshes generated for immersive experiences. The subjective evaluation of the content is conducted in a Virtual Reality setting, using the forced-choice pairwise comparison methodology with existing reference. The results of this study is a mapping of the compared conditions to a continuous ranking scale that can be used in order to optimize codec choice and compression parameters to achieve optimum balance between bandwidth and perceived quality in tele-immersive platforms.

Visual Quality of 3D Meshes With Diffuse Colors in Virtual Reality: Subjective and Objective Evaluation

“…Many authors have conducted subjective quality assessment tests involving 3D meshes [2], [5], [8], [9], [10], [11], [12], [13], [14], [15], [16] or 3D point clouds [17], [18], [19], [20], [21]. They considered a variety of methods: Absolute Category Rating (ACR) [11], [13], [22], Double-Stimulus Impairment Scale (DSIS) [2], [8], [9], [18], [19], [21] and Pairwise Comparison (PC) [5], [14], [15], [23]. Very recently, a study [16] attempted to compare these subjective methods and showed that, for the particular case of 3D graphical content, the DSIS method tends to produce more accurate results than ACR (i.e., MOS with smaller confidence intervals).…”

“…Very recently, a study [16] attempted to compare these subjective methods and showed that, for the particular case of 3D graphical content, the DSIS method tends to produce more accurate results than ACR (i.e., MOS with smaller confidence intervals). Existing subjective experiments also considered different ways of presenting 3D content: static images [8], animated content without interaction (usually low-speed rotation) [5], [9], [10], [16], [19], [20], [21], [22], [23] or interactive content [2], [11], [13], [14], [15], [17], [18]. We denote that only in [15], [16], the experiments were conducted in a VR environment.…”

“…Our dataset allows us to provide an initial investigation on the influence of movement and viewpoint on the quality evaluation of 3D content. As in [15], [16] we considered a VR context and we used a DSIS method with 24 observers per stimulus, as recommended in [16].…”

Textured Mesh Quality Assessment: Large-Scale Dataset and Deep Learning-based Quality Metric

“…Subjective quality tests involving 3D models were initially introduced on meshes, more precisely on geometry-only models, to assess the artifacts induced by the simplification, smoothing, watermarking and compression [Christaki et al 2018;Corsini et al 2007;Lavoué 2009;Lavoué et al 2006;Rogowitz and Rushmeier 2001;Silva et al 2009;Torkhani et al 2015;Vanhoey et al 2017;Váša and Rus 2012;Váša and Skala 2011;Watson 2001]. Little work considered meshes with color attributes (vertex color or texture) [Doumanoglou et al 2018;Guo et al 2016;Gutiérrez et al 2020;Nehmé et al 2020;Nehmé et al 2021b;Pan et al 2005;Zerman et al 2020].…”

“…The experimental methodologies used were inspired by existing image/video subjective methodologies. They are mainly derived from single stimulus methods, in which participants see only one stimulus and rate its quality [Corsini et al 2007;Gutiérrez et al 2020;Lavoué et al 2006;Subramanyam et al 2020;Torkhani et al 2015;Viola et al 2022;Zerman et al 2020], double stimulus methods in which participants rate the visual degradation after seeing the reference and distorted stimuli [da Silva Cruz et al 2019;Javaheri et al 2019;Lavoué 2009;Liu et al 2021;Nehmé et al 2021b;Perry et al 2020;Silva et al 2009;Su et al 2019;Torlig et al 2018;Watson 2001], and pairwise comparison methods in which participants choose the better quality stimulus from two stimuli presented to them Christaki et al 2018;Guo et al 2016;Vanhoey et al 2017;Váša and Rus 2012]. Recently, a couple of comprehensive/comparative studies Nehmé et al 2020] evaluated the impact of the subjective methodologies on the obtained quality scores.…”