Point Cloud Random Access via ISOBMFF Metadata Indexing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current solutions for random access in point cloud data bit-streams, such as G-PCC, require parsing of multiple samples to retrieve parameter sets, leading to inefficiencies and increased processing requirements, especially when accessing non-sync samples that depend on parameter sets from other samples.

Innovation Solution

A method for encapsulating point cloud data in an ISOBMFF-based media file that includes metadata to identify and locate required parameter sets, allowing for improved random access by reducing the number of parameter sets that need to be extracted and parsed, enabling decoding of non-sync samples without accessing other samples when possible.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If random access is performed in G-PCC bit-stream without additional metadata, then the bit-stream structure remains simple, but multiple samples must be parsed to retrieve parameter sets, increasing processing time and computational load

Engineering Contradiction:
Improvedecoding speedVSAvoidbit-stream structure
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by pre-computing and storing metadata during the encapsulation phase that identifies which parameter sets are required for each sample. This metadata is generated in advance and stored in the media file, so that during random access operations, the decoder can immediately retrieve the necessary parameter sets without having to parse multiple samples to discover what is needed. The SampleGroupDescriptionBox and SampleToGroupBox structures are created beforehand to map samples to their required parameter sets, enabling efficient random access decoding.

Inventive Principle:
Principle #10Preliminary action

2Quantity of substance

If parameter sets are distributed across multiple samples, then the bit-stream can be more compact, but accessing a single sample requires parsing other samples to find required parameter sets, increasing processing requirements

Engineering Contradiction:
Improvebit-stream sizeVSAvoidrandom access operation
Core Design Contradiction:
Quantity of substanceVSEase of operation

Solution Approach 1:

The patent introduces an intermediary metadata layer (SampleGroupDescriptionBox containing SampleToGroupBox entries) that acts as a map or index between samples and their required parameter sets. This intermediary structure does not increase the actual parameter set data but provides a lookup mechanism that tells the decoder exactly where to find required parameter sets without having to search through multiple samples. The metadata serves as a mediator that simplifies the relationship between samples and parameter sets, enabling efficient random access while maintaining compact bit-stream structure.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Ease of operation

If all parameter sets are included in every sample to enable independent decoding, then random access becomes simpler, but the bit-stream size increases significantly

Engineering Contradiction:
Improverandom access capabilityVSAvoidbit-stream size
Core Design Contradiction:
Ease of operationVSQuantity of substance

Solution Approach 1:

The patent applies segmentation by dividing the bit-stream into samples that reference shared parameter sets stored elsewhere in the media file. Instead of duplicating parameter sets in every sample, the bit-stream is segmented into sample data and parameter set data, with metadata linking them. This segmentation allows the parameter sets to be stored once and referenced by multiple samples, reducing redundancy while enabling efficient random access through the metadata lookup mechanism.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS20240346698A1Method, device, and computer program for improving random access in point cloud data bit-stream
Publication Date: 2024.10.17 CANON KK
  • US20240346698A1 patent drawing
  • US20240346698A1 patent drawing
  • US20240346698A1 patent drawing

AI summary

According to some embodiments of the disclosure, it is provided a method for encapsulating point cloud data into an ISOBMFF-based media file, the point cloud data being organized in point cloud frames. After having generated a track describing a sequence of samples, each sample comprising at least one point cloud frame, the sequence of samples comprising at least one sync sample and a plurality of non-sync samples, at least one non-sync sample of the plurality of non-sync samples comprising a parameter set, and after having generated metadata describing whether a non-sync sample of the plurality of non-sync samples requires, to be decoded, a parameter set from another non-sync sample of the non-sync samples, the track and the metadata are encapsulated in the media file.