Enhancing 360 Video Streaming through Salient Content in Head-Mounted Displays

Area of Science:

• Computer Science
• Human-Computer Interaction
• Multimedia Systems

Background:

• Streaming 360 videos in head-mounted displays (HMDs) is challenging due to bandwidth limitations.
• Accurately predicting user head movements is crucial for optimizing video streaming and enhancing user experience.
• Existing methods struggle with the rapid and unpredictable nature of head movements in immersive environments.

Purpose of the Study:

• To develop a novel approach for predicting user head movements in HMDs by analyzing unique visual saliency in 360 videos.
• To improve the efficiency of 360 video streaming systems by leveraging accurate head movement predictions.
• To enhance the Quality of Experience (QoE) for users viewing 360 videos in HMDs.

Main Methods:

• Extraction of salient visual cues specific to 360 video content.
• Development of a head-movement prediction algorithm based on identified saliency features.
• Implementation of a 360 video streaming framework incorporating the predictive algorithm.

Main Results:

• The proposed saliency-based system reduced video stall duration by 65%.
• The system decreased the number of video stalls by 46%.
• Bandwidth savings of 31% were achieved compared to state-of-the-art methods.

Conclusions:

• The developed saliency-based approach effectively predicts user head movements in HMDs.
• The enhanced streaming framework significantly improves video delivery efficiency and user experience.
• This method offers a promising solution for bandwidth-constrained 360 video streaming.