HEVC Tile Streaming with Spatial Segments for Parallel Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing HEVC-tiled video streaming methods face challenges in reducing network traffic and processor load, with current techniques either compromising compression efficiency or increasing the number of HTTP requests, while also limiting the ability to efficiently decode and process subsets of video data.
Innovation Solution
The method involves creating a spatial manifest file that allows clients to request and decode HEVC-tiled video data as independent spatial segments, reducing the number of HTTP requests and processor resources by enabling parallel decoding of video data within defined spatial segments, which are subsets of HEVC tiles, without spatial and temporal decoding dependencies across segment boundaries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If HEVC tiles are independently encoded to enable parallel decoding and subset selection, then decoding flexibility and parallelism are improved, but compression efficiency deteriorates due to absence of spatial and temporal predictions between tiles
Solution Approach 1:
The video stream is divided into spatial segments containing one or more HEVC tiles that are independently encoded. This segmentation allows the decoder to process only the requested spatial segment without decoding the entire video stream, enabling parallel processing and improving decoding flexibility while maintaining reasonable compression efficiency within each segment.
Solution Approach 2:
Spatial segments are designed with independent encoding characteristics suitable for their specific function (e.g., ROI vs. non-ROI areas). The encoding parameters, tile configurations, and prediction constraints are locally optimized for each spatial segment, allowing high compression efficiency within segments while maintaining the ability to decode segments independently for parallel processing.
2Adaptability or versatility
If the full HEVC-tiled video stream is transmitted to the client, then all video data is available for any region of interest, but network traffic increases and client processing load increases
Solution Approach 1:
The server extracts and transmits only the requested spatial segment containing the region of interest from the full HEVC-tiled video stream. The client receives a reduced data set that contains only the necessary video tiles for the current ROI, significantly reducing network traffic and client processing load while maintaining the ability to request other segments as the ROI changes.
Solution Approach 2:
The system dynamically adapts the transmitted video data based on the client's current region of interest. As the user interacts with the video (panning, zooming), the server dynamically selects and transmits different spatial segments, ensuring that the network traffic and processing load are optimized for the current viewing requirements rather than transmitting the entire video stream.
3Productivity
If multiple independent HEVC tile streams are requested from the network, then selective decoding of subsets is enabled, but the number of HTTP requests and client synchronization complexity increases
Solution Approach 1:
Multiple HEVC tiles that form a coherent spatial segment are merged into a single encoded stream with independent decodability. The server packages related tiles together in a spatial segment that can be requested and decoded as one unit, reducing the number of HTTP requests from multiple individual tile requests to a single segment request, while maintaining the ability to selectively decode different segments.
Solution Approach 2:
The spatial segment structure acts as an intermediary between the full HEVC-tiled video stream and the individual tiles. It provides a middle-level organization that allows the server to manage tile groups efficiently and the client to request and process manageable segments without dealing with the complexity of individual tile synchronization and stitching.
Data Source
AI summary
A method is described of streaming HEVC-tiled video data to a client device comprising: providing said client device with a spatial manifest file comprising one or more spatial segments identifiers for locating one or more delivery nodes configured for delivering video data associated with a spatial segment identified by at least one of said one or more spatial segments identifiers, preferably (part of) one or more URLs, to said client device, a spatial segment being associated with HEVC-tiled video data comprising a subset of HEVC tiles of a HEVC-tiled (panorama) video stream; and, selecting a spatial segment identifier in said spatial manifest file for requesting a delivery node to deliver at least part of the video data of a spatial segment as a HEVC-tiled video stream to the client device.