Point Cloud Entropy Coding with Adaptive Neighbor Occupancy Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current point cloud compression techniques are inefficient and lack effective encoding and decoding processes, hampering the adoption and deployment of point cloud data in applications such as autonomous vehicles and virtual reality, due to the large datasets and complex geometric representations involved.
Innovation Solution
The method involves encoding and decoding point clouds using a tree structure, where nodes are split into sub-volumes, and entropy encoding is applied based on the occupancy patterns of neighboring nodes, selecting from multiple probability distributions to optimize compression, and updating these distributions dynamically during the encoding and decoding processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional entropy coding methods are used for point cloud compression, then the coding complexity remains low, but the compression efficiency is insufficient
Solution Approach 1:
The patent applies context-adaptive probability distributions that are locally optimized for each node's occupancy pattern. Different probability distributions are selected based on the specific geometric characteristics of neighboring nodes, allowing the encoding to adapt to local geometric correlations rather than using a uniform approach throughout the entire point cloud.
Solution Approach 2:
The patent dynamically updates probability distributions during the encoding process based on previously encoded occupancy patterns. The context model evolves as encoding progresses, adapting to the statistical properties of the point cloud data being processed, which improves compression efficiency without requiring a complete reanalysis of the data structure.
2Productivity
If simple occupancy encoding is used, then the coding complexity is low, but the compression performance is poor due to ignoring geometric correlations
Solution Approach 1:
The patent implements a context model that uses feedback from previously encoded occupancy patterns to inform the encoding of current nodes. The probability distributions are updated based on the occupancy patterns of neighboring nodes, creating a feedback loop that continuously refines the encoding strategy to match the actual geometric correlations present in the data.
Solution Approach 2:
The patent performs preliminary analysis of occupancy patterns in neighboring nodes before encoding the current node. By pre-selecting appropriate probability distributions based on the occupancy status of adjacent nodes, the encoding process is optimized in advance, capturing geometric correlations before the actual encoding occurs.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
Methods and devices for encoding a point cloud. A current node associated with a sub-volume is split into further sub-volumes, each further sub-volume corresponding to a child node of the current node, and, at the encoder, an occupancy pattern is determined for the current node based on occupancy status of the child nodes. A probability distribution is selected from among a plurality of probability distributions based on occupancy data for a plurality of nodes neighbouring the current node. The encoder entropy encodes the occupancy pattern based on the selected probability distribution to produce encoded data for the bitstream and updates the selected probability distribution. The decoder makes the same selection based on occupancy data for neighbouring nodes and entropy decodes the bitstream to reconstruct the occupancy pattern.