A Viewpoint-Centric Approach to Transition and Quality Delays in Multi-View Video Streaming

Multi-view video streaming systems encounter challenges in managing delays during view-switching, which are related to switching response and smooth quality transition. Both factors prevent multi-view streaming to achieve seamless viewpoint transition, reducing the overall quality of experience. To tackle viewpoint delay, we employ a viewpoint-based group segment request method that assigns specific user perceptual quality targets for different viewpoint priorities. While this method effectively addresses switching response delay, it struggles to maintain smooth quality transition, which results from a quality drop during view-switching. The quality drop occurs because the bandwidth limitation hinders the transmission of high-quality video content from all viewpoints. This paper investigates viewpoint delays in multi-view video streaming based on Dynamic Adaptive Streaming over HTTP service. Through a detailed analysis, we identify buffer management as a pivotal factor in mitigating viewpoint delays. Thus, we propose dynamic quality target adjustments to regulate buffer levels. Additionally, we introduce the concept of segment redownloading to enhance low-quality segments after view-switching. The implementation results demonstrate the effectiveness of adjusting quality targets to minimize rebuffering, while segment redownloading mitigates quality delay issues.