Inter-prediction Trisoup Coding for Point Cloud Compression

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current point cloud compression techniques, such as V-PCC and G-PCC, face limitations in efficiently compressing dense and dynamic point clouds, particularly in real-time transmission for applications like AR/VR, where high bitrates and latency constraints are critical, and existing methods like TriSoup coding struggle with temporal redundancy and visual quality.

Innovation Solution

The proposed method introduces an inter-prediction based Trisoup coding scheme that reduces temporal redundancy by using reference Trisoup information from previous frames to encode and decode point cloud sequences, improving compression efficiency and maintaining visual quality by encoding Trisoup information residuals and predicting temporal Trisoup information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If intra-prediction based Trisoup coding is used for point cloud compression, then encoding complexity is reduced and processing speed is improved, but temporal redundancy is not reduced leading to lower compression efficiency

Engineering Contradiction:
Improveprocessing speedVSAvoidtemporal redundancy
Core Design Contradiction:
ProductivityVSLoss of information

Solution Approach 1:

The patent applies preliminary action by constructing a Trisoup model from reference point cloud data before encoding the current frame. The Trisoup information is predicted in advance using temporal prediction from previously decoded frames, allowing the encoder to prepare prediction data before the actual encoding process, thus improving processing speed while maintaining compression efficiency through reduced temporal redundancy.

Inventive Principle:
Principle #10Preliminary action

2Loss of information

If inter-prediction based Trisoup coding is used, then temporal redundancy is reduced and compression efficiency is improved, but encoding complexity and computational load increase

Engineering Contradiction:
Improvetemporal redundancyVSAvoidencoding complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The patent applies segmentation by dividing the point cloud data into multiple frames and processing each frame independently using inter-prediction. The Trisoup information is segmented into reference frame data and current frame residuals, allowing the encoder to focus computational resources on encoding only the differences (residuals) rather than the entire point cloud, thus reducing encoding complexity while maintaining compression efficiency.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent uses copying by replicating Trisoup information from reference frames to predict current frame data. Instead of encoding entire point cloud frames, the system copies relevant Trisoup information from previously decoded frames and only encodes the residuals (differences), significantly reducing encoding complexity and temporal redundancy simultaneously.

Inventive Principle:
Principle #26Copying

3Quantity of substance

If high compression ratios are achieved through existing methods, then bitrate is reduced, but visual quality deteriorates due to lossy compression artifacts

Engineering Contradiction:
ImprovebitrateVSAvoidvisual quality
Core Design Contradiction:
Quantity of substanceVSManufacturing precision

Solution Approach 1:

The patent applies parameter changes by transforming point cloud data into Trisoup representation and encoding residuals rather than raw coordinates. This parameter transformation allows for more efficient compression at lower bitrates while maintaining visual quality, as the Trisoup model captures the essential geometric information with fewer bits compared to traditional coordinate-based methods.

Inventive Principle:
Principle #35Parameter changes

4Manufacturing precision

If lossless compression is used to maintain visual quality, then compression ratio decreases and bitrate increases

Engineering Contradiction:
Improvevisual qualityVSAvoidbitrate
Core Design Contradiction:
Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The patent replaces traditional mechanical coordinate-based compression methods with a mathematical Trisoup modeling approach. By substituting the direct encoding of x, y, z coordinates with Trisoup information encoding, the system achieves lossless or near-lossless compression at lower bitrates, maintaining visual quality while reducing the quantity of data required.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentEP4365843A1Encoding/decoding point cloud geometry data
Publication Date: 2024.05.08 BEIJING XIAOMI MOBILE SOFTWARE CO LTD
  • EP4365843A1 patent drawingFigure 1
  • EP4365843A1 patent drawingFigure 2~3
  • EP4365843A1 patent drawingFigure 4

AI summary

There is provided methods and apparatus of encoding/decoding into/from a bitstream, geometry of point cloud frames of a point cloud sequence. Methods use an inter Trisoup coding/decoding scheme, i.e. a scheme that encodes/decodes Trisoup information of a current point cloud frame of the point cloud sequence by making use of Trisoup information of already encoded/decoded point cloud frames of said point cloud sequence.