Omnidirectional Video Region Segmentation for Zoom Quality
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Omnidirectional videos face challenges in providing high-quality viewing experiences due to lower resolution per unit area and limited bitrate, resulting in poor image quality, especially when trying to focus on regions of interest, which existing technologies have not effectively addressed.
Innovation Solution
A method and device for processing omnidirectional video data by determining and encoding sphere or 2D regions on a projected picture, allowing for direct zooming without switching code streams, enabling continuous zoom operations and improving video quality by storing and decoding video data based on zoom region information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If omnidirectional video provides panoramic viewing experience in horizontal (360 degrees) and vertical (180 degrees) directions, then the viewing coverage is improved, but the video quality per unit area or resolution (number of pixels per degree) deteriorates under the same bitrate condition
Solution Approach 1:
The omnidirectional video is segmented into multiple spherical regions, each further divided into multiple zoom regions. This segmentation allows different regions to be encoded independently with different bitrates, enabling high-quality encoding for regions of interest while maintaining panoramic coverage, thus resolving the contradiction between viewing coverage and video quality per unit area
Solution Approach 2:
Different spherical regions and zoom regions are assigned different bitrates based on their importance. Regions of interest receive higher bitrates for better quality, while less important regions use lower bitrates. This local quality differentiation maintains overall panoramic coverage while ensuring high video quality where needed
2Productivity
If the video bitrate of omnidirectional video is limited by network bandwidth conditions, then the transmission efficiency is improved, but the image quality of region of interest deteriorates
Solution Approach 1:
By segmenting the video into spherical regions and zoom regions with different importance levels, the limited bitrate can be strategically allocated. High-priority regions of interest receive sufficient bitrate for good quality, while lower-priority regions use fewer bits, achieving both transmission efficiency and acceptable image quality in critical areas
Solution Approach 2:
The encoding parameters (bitrate, resolution) are changed dynamically for different regions based on their importance. This parameter differentiation allows the system to optimize transmission efficiency overall while ensuring adequate image quality for regions of interest within the constrained bandwidth
3Ease of operation
If existing technologies are used to zoom into regions of interest, then the focus on specific areas is achieved, but the video stream disruption and quality loss occur due to switching code streams
Solution Approach 1:
All spherical regions and their zoom regions are pre-encoded and stored during the encoding phase. When zooming is needed during playback, the system simply switches to the pre-prepared zoom region data without needing to process or transcode in real-time. This preliminary encoding action ensures smooth zoom transitions without video stream disruption
Solution Approach 2:
Multiple copies of the same video content are created at different zoom levels and resolutions during encoding. These copies (base spherical region + multiple zoom regions) are stored and can be directly switched during playback, enabling seamless zooming without stream disruption while maintaining video quality
Data Source
AI summary
The present application provides a video data processing method and device, a server, a terminal, and a storage medium. The method includes: determining a sphere region of an omnidirectional video, and one or more zoom regions of the sphere region on a projected picture of the omnidirectional video; and decoding and playing video data of the one or more zoom regions of the sphere region on the projected picture of the omnidirectional video, in response to a zoom operation on the sphere region of the omnidirectional video.