Geometric Point Cloud Intra Prediction via Level-Dependent Neighbor Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In geometry-based point cloud compression, the existing intra prediction methods require all 26 neighbor nodes for prediction, leading to increased computational complexity and reduced prediction efficiency.
Innovation Solution
The proposed intra prediction method determines a current quantity of neighbor nodes based on the current level in the octree partition, using only a subset of neighbor nodes for prediction, thereby reducing computational complexity while maintaining prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all 26 neighbour nodes are used for intra prediction, then prediction accuracy is improved, but computational complexity increases and prediction efficiency decreases
Solution Approach 1:
The patent applies partial action by selecting only a subset of neighbour nodes (current quantity less than 26) for intra prediction based on the current level in the octree partition. This reduces the number of neighbour nodes processed from the maximum 26 to a level-dependent subset, thereby reducing computational complexity while maintaining acceptable prediction accuracy for different spatial resolutions
2Measurement precision
If all 26 neighbour nodes are used for intra prediction, then prediction accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent reduces computational complexity by processing only a partial set of neighbour nodes determined by the current level. The current quantity of neighbour nodes is set to a first quantity for certain levels and a second quantity for other levels, where these quantities are less than 26, thereby reducing the computational burden while maintaining prediction accuracy
3Ease of manufacture
If a fixed number of neighbour nodes is used for intra prediction, then implementation is simplified, but prediction accuracy deteriorates for different spatial resolutions
Solution Approach 1:
The patent introduces dynamics by making the number of neighbour nodes (current quantity) variable rather than fixed. The current quantity is dynamically determined based on the current level in the octree partition, allowing the system to adapt to different spatial resolutions and point cloud densities, thereby maintaining prediction accuracy across varying conditions
Data Source
AI summary
An intra prediction method and a decoder are provided. The method includes the following. Coordinate information of a current node and a current level corresponding to the current node are determined. A current quantity is determined based on the current level, where the current quantity is positively correlated with the current level. Current neighbour nodes are determined based on the coordinate information of the current node, where the current neighbour nodes are neighbour nodes of the current quantity among neighbour nodes of a parent node of the current node. Intra prediction is performed on the current node based on the coordinate information of the current node and the current neighbour nodes.