Scalable Lifting Decomposition for Point Cloud Attribute Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current graph-based point cloud compression (G-PCC) technologies face inefficiencies in compression efficiency when there are not enough neighboring attribute samples available for prediction, particularly in scenarios requiring scalable reconstruction from lossy to lossless fidelity.
Innovation Solution
The method extends the existing G-PCC lifting design to enable scalable coding of lifting coefficients, utilizing a lifting decomposition that transforms point cloud data for improved attribute coding efficiency, allowing for better prediction and reconstruction even with limited neighboring samples.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If distance-based weighted average prediction is used for attribute coding in G-PCC, then compression efficiency is improved when sufficient neighboring samples are available, but compression efficiency deteriorates when neighboring attribute samples are insufficient
Solution Approach 1:
The patent introduces lifting decomposition as an intermediary transformation that converts original attribute values into lifting coefficients. These coefficients can be predicted using neighboring samples even when direct attribute prediction fails due to insufficient neighbors. The lifting transform acts as a mediator that creates a representation where prediction is more effective.
Solution Approach 2:
The patent transforms the attribute coding from direct attribute value prediction to lifting coefficient prediction. By changing the parameter representation from original attributes to lifting coefficients through lifting decomposition, the system enables effective prediction even with limited neighboring samples, thus resolving the contradiction between compression efficiency and prediction accuracy.
2Adaptability or versatility
If scalable reconstruction from lossy to lossless fidelity is required, then coding flexibility is improved, but device complexity increases
Solution Approach 1:
The patent applies lifting decomposition preliminarily to transform attribute values into lifting coefficients before encoding. This preliminary transformation enables scalable reconstruction by organizing the data in a form that supports progressive refinement from lossy to lossless fidelity, while the transform itself remains computationally efficient.
Solution Approach 2:
The lifting decomposition segments the attribute coding into different components (lifting coefficients) that can be independently encoded at different quality levels. This segmentation enables scalable reconstruction where lower quality levels use coarse coding and higher quality levels add refined details, managing complexity through hierarchical organization.
Data Source
AI summary
A method, computer system, and computer-readable medium are provided for point cloud attribute coding by at least one processor. Data associated with a point cloud is received. The received data is transformed through a lifting decomposition based on enabling a scalable coding of attributes associated with the lifting decomposition. The point cloud is reconstructed based on the transformed data.