Octree Point Cloud Coding with Planar and Angular Context Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for encoding and decoding point clouds do not effectively utilize information from neighboring nodes, leading to inefficient data compression and reconstruction.
Innovation Solution
The method involves determining coding modes (planar and angular) and using context information from neighboring nodes in an octree-based structure to enhance entropy encoding, considering occupancy patterns and plane positions for improved compression.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional entropy coding without neighboring node information is used, then the encoding process is simple, but compression efficiency is insufficient
Solution Approach 1:
The patent performs preliminary actions by pre-defining multiple context models (first context model for planar nodes, second context model for angular nodes) that incorporate neighboring node occupancy information before the actual encoding process. This allows the encoder to directly use pre-computed contextual probabilities during encoding, improving compression efficiency without significantly increasing real-time computational complexity.
Solution Approach 2:
The patent introduces context models as intermediary structures that mediate between the raw occupancy data and the entropy coding process. These context models aggregate and process neighboring node information to generate contextual probabilities, which then guide the entropy coding decisions. This intermediary layer enables efficient compression by pre-processing spatial relationships without burdening the main encoding loop.
2Manufacturing precision
If planar coding mode is used for all nodes, then the coding process is uniform and simple, but it cannot accurately represent angular structures
Solution Approach 1:
The patent applies local quality by differentiating the coding approach based on the local geometric characteristics of each node. Planar coding mode is used for nodes with planar surface characteristics, while angular coding mode is used for nodes with angular or concave characteristics. This localized adaptation allows each node to be encoded using the most suitable method, improving overall reconstruction accuracy without requiring complex processing for all nodes uniformly.
Solution Approach 2:
The patent changes the coding parameters (context model selection, probability distribution assumptions) based on the node type. For planar nodes, a context model assuming planar surface continuity is used, while for angular nodes, a different context model accounting for angular transitions is applied. This parameter adaptation enables accurate representation of diverse geometric structures using a unified framework.
3Measurement precision
If neighboring node occupancy information is considered, then prediction accuracy improves, but the context information processing becomes more complex
Solution Approach 1:
The patent segments the context information processing into distinct components: identifying neighboring nodes, determining node types (planar/angular), selecting appropriate context models, and computing contextual probabilities. This segmentation allows each step to be optimized independently and makes the overall complex process more manageable and implementable.
Solution Approach 2:
The patent uses context models that copy and reuse probability distributions from similar neighboring nodes. Instead of computing everything from scratch, the system copies successful probability assignments from neighboring nodes with similar geometric characteristics, adjusting them as needed for the current node. This copying mechanism significantly reduces processing complexity while maintaining high prediction accuracy.
Data Source
AI summary
A method for encoding a point cloud to generate a bitstream of compressed point cloud data is provided. The point cloud's geometry is represented by an octree-based structure with a plurality of nodes having parent-child relationships by recursively splitting a volumetric space containing the point cloud into sub-volumes each associated with a node of the octree-based structure. The method includes: determining a coding mode, wherein the coding mode includes a planar coding mode and an angular coding mode; obtaining coding context information for a present child node, entropy encoding the present child node based on the obtained coding context information to produce encoded data for the bitstream.