제안 방법

• In this paper, we analyze the performance on view synthesis from a set of 360 videos in omnidirectional 6DoF in various ways with different distances and number of input views by using Reference View Synthesizer (RVS) [3], [4]. In the experiments, the dataset of ClassroomImage [5] was used, and the objective quality was evaluated by EquiRectangular Projection (ERP) Weighted Spherical PSNR (WS-PSNR) software [6].
• In this experiment, we synthesize the view with different number of input views to investigate the relationship between synthesis performance and processing time depending on the number of input views. Figure 6 shows the various way to synthesize the virtual view by using different multiple input views with ClassroomImage sequence.
• In this experiment, we synthesize the virtual views by using two input views at the same distances of different cases (60, 120, 240, 480mm) from the input view. Figure 3 and Table 1 show the input viewpoints at the various distances from each synthesized view with the ClassroomImage sequence.
• In this paper, we analyze the performance on view synthesis from a set of 360 videos in omnidirectional 6DoF in various ways with different distances and number of input views by using Reference View Synthesizer (RVS) [3], [4]. In the experiments, the dataset of ClassroomImage [5] was used, and the objective quality was evaluated by EquiRectangular Projection (ERP) Weighted Spherical PSNR (WS-PSNR) software [6].
• The goal of the experiments in this paper is to check the performance of view synthesis using RVS3.1, according to various synthesis conditions such the distance between the synthesized view and the input view and the number of input views. For this, we performed two experiments as follows:
• This paper presents experimental results on virtual view synthesis of omnidirectional 6DoF environment using RVS 3.1 in various way with the distance between synthesized view and input view, and the number of input views.