Geometry-Based Point Cloud Tile Selection for Adaptive Streaming
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding systems lack adequate mechanisms for streaming geometry-based point clouds over a network, particularly in supporting adaptive streaming and efficient data transmission.
Innovation Solution
Adaptive streaming of geometry-based point clouds is enabled by signaling elements, attributes, and metadata associated with point cloud components, allowing clients to identify and select appropriate point cloud streams and components based on client capabilities, using a media presentation descriptor (MPD) to manage bandwidth and quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all point cloud data is streamed, then complete geometry representation is achieved, but network bandwidth consumption increases
Solution Approach 1:
The point cloud data is divided into multiple tiles, where each tile represents a spatial portion of the complete point cloud. Clients can selectively stream only the tiles they need based on their viewing frustum or region of interest, rather than downloading all point cloud data. This segmentation enables efficient bandwidth utilization while maintaining the ability to represent complete geometry when all tiles are combined.
2Adaptability or versatility
If multiple versions of point cloud components are provided, then client adaptability is improved, but system complexity increases
Solution Approach 1:
The system provides multiple versions of point cloud components (such as different compression levels, resolutions, or quality settings) that can be dynamically selected based on client capabilities and network conditions. The MPD (Media Presentation Description) structure enables dynamic adaptation by allowing clients to choose appropriate versions without requiring complex manual configuration, thus achieving adaptability with manageable system complexity.
3Measurement precision
If detailed metadata is signaled for all components, then client selection accuracy is improved, but data transmission overhead increases
Solution Approach 1:
Essential metadata elements are extracted and signaled in the MPD structure to enable client selection and adaptation. Rather than transmitting complete detailed metadata for all point cloud components, the system extracts only the most critical information needed for client decision-making (such as tile boundaries, component types, and quality indicators). This extraction approach maintains client selection accuracy while significantly reducing data transmission overhead.
Data Source
AI summary
Systems, methods, and instrumentalities are disclosed for receiving content related to geometry-based point clouds. A client device may receive a media presentation description (MPD) file from a content server. A set of pre-selection elements may be identified from the MPD file. One or more adaptation sets associated with at least one pre-selection element of the set of pre-selection elements may be identified. The adaptation set(s) may be indicated by an attribute associated with one of the pre-selection elements. A geometry-based point cloud compression (GPCC) tile identifier associated with a viewport may be determined. The GPCC tile identifier may be determined based on a first descriptor received in the MPD file. One or more adaptation sets associated with the GPCC tile identifier may be selected using a second descriptor. A point cloud component associated with the selected one or more adaptation sets may be requested. Point cloud component may be received.