Confidence-Aware Tile Selection for 6DoF Mixed Reality Streaming

NCU Institutional Repository > 資訊電機學院 > 通訊工程研究所 > 博碩士論文 > Item 987654321/95363

jsp.display-item.identifier=請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/95363

题名:	Confidence-Aware Tile Selection for 6DoF Mixed Reality Streaming
作者:	簡正星;Chien, Cheng-Hsing
贡献者:	通訊工程學系
关键词:	信心度;資源分配;點雲;視野預測;confidence;resource allocation;MR;FoV prediction;volumetric media streaming
日期:	2024-08-13
上传时间:	2024-10-09 16:42:51 (UTC+8)
出版者:	國立中央大學
摘要:	混合實境（MR）的應用越來越受歡迎，因為它允許將虛擬物件疊加到現實世界上，從而促進虛擬宇宙的發展。本研究論文介紹了一種創新的 MR 串流媒體系統，該系統考慮了置信水平，其目標是在存在網路限制的情況下提高體驗品質 (QoE) 並優化資源分配。除了促進使用者在六個自由度上的移動之外，還設計了視野 (FoV) 分類問題來預測 FoV 中心和置信度分數。該方法用於建立新穎的可見性似然圖。據我們所知，這是第一個解決容易出錯的 FoV 預測挑戰的研究，其程度達到所有圖塊不需要基本資源的程度。結果表明，與替代方法相比，效用提高了約 33\%。當系統在 20 Mb/s 的頻寬限制下達到最大容量且丟失瓦片最少時，就會觀察到這種增加。;The application of mixed reality (MR) is gaining popularity as it allows virtual objects to be overlaid onto the real world, contributing to the development of the metaverse. This research paper introduces an innovative MR streaming system that takes into account confidence levels, with the goal of enhancing the quality of experience (QoE) and optimizing resource allocation in the presence of network limitations. Alongside facilitating user movements in six degrees of freedom, a problem of classifying the field of view (FoV) is devised to predict FoV centers and confidence scores. This approach is used to construct a novel visibility likelihood map. To the best of our knowledge, this is the first study to address the challenge of error-prone FoV prediction to such an extent that no basic resources are required for all tiles. The results demonstrate a utility increase of approximately 33\% compared to alternative methods. This increase is observed when the system reaches its maximum capacity at a bandwidth constraint of 20 Mb/s, with minimal missing tiles.
显示于类别:	[Graduate Institute of Communication Engineering] Electronic Thesis & Dissertation

文件中的档案:

档案	描述	大小	格式	浏览次数
index.html		0Kb	HTML	5	检视/开启

在NCUIR中所有的数据项都受到原著作权保护.

社群 sharing

数据加载中.....