Point Cloud Track Partitioning for Temporal Scalability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods struggle with efficiently processing vast amounts of point cloud data required for services like virtual reality, augmented reality, and self-driving, particularly in handling temporal scalability and sample extraction based on target parameters, and storing G-PCC bitstreams in a way that supports efficient access and signaling.
Innovation Solution
A method and device that identify tracks based on target temporal levels, extract samples efficiently, and partition bitstreams into multiple tracks while providing signaling for G-PCC files, allowing for temporal scalability and efficient storage and access.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If point cloud data is stored in a single track, then storage structure is simple, but access efficiency and temporal scalability are limited
Solution Approach 1:
The patent divides the point cloud data into multiple tracks with different temporal levels (e.g., base layer and enhancement layers). Each track contains samples at specific temporal levels, allowing independent access and processing. This segmentation enables efficient temporal scalability and selective sampling without requiring complex parsing of the entire bitstream, thus improving access efficiency while maintaining manageable structural complexity.
2Productivity
If all samples are extracted for processing, then data completeness is maintained, but processing time and computational load increase
Solution Approach 1:
The patent implements a sample extraction mechanism that selectively extracts only the necessary samples based on target temporal levels. The system identifies tracks containing samples at the required temporal levels and extracts only those samples, discarding unnecessary lower-level samples. This extraction approach maintains data completeness for the target temporal level while significantly reducing processing time and computational load compared to processing all samples.
Solution Approach 2:
The patent performs preliminary organization of point cloud data into tracks with explicit temporal level annotations during encoding. This preliminary structuring enables the decoding system to quickly identify and extract only the necessary samples without needing to parse the entire bitstream sequentially. The preliminary track organization acts as a guide that accelerates the sample extraction process and reduces computational overhead.
3Adaptability or versatility
If temporal scalability is added to G-PCC files, then adaptability to different quality levels is improved, but file format complexity increases
Solution Approach 1:
The patent extends the G-PCC file format to support multiple temporal levels within a unified track structure. The same file format infrastructure is used to store both base layer and enhancement layer samples, with temporal level information encoded as metadata. This universal approach allows the file format to handle temporal scalability without requiring separate encoding schemes or complex parsing logic, thus improving adaptability while minimizing format complexity.
4Measurement precision
If sample extraction is performed without track identification, then processing is simpler, but target parameter accuracy is compromised
Solution Approach 1:
The patent introduces track identification as an intermediary step between the raw bitstream and the sample extraction process. The system first identifies tracks based on their temporal level characteristics and annotations, then uses this identification information to guide the sample extraction process. This intermediary track identification mechanism ensures accurate targeting of samples at the desired temporal levels while maintaining relatively simple extraction logic, thus improving measurement precision without excessive complexity.
Data Source
AI summary
A transmission device of point cloud data, a method performed in the transmission device, a reception device, and a method performed in the reception device are provided. A method performed by a reception device of point cloud data may comprise identifying a track set including one or more tracks from the point cloud data based on a predetermined target temporal level and extracting one or more samples from the identified track set. Each track included in the track set may include a temporal level less than or equal to the target temporal level.