Point Cloud Attribute Prediction Using Neighborhood Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The inaccuracy of determining neighboring points in point cloud attribute encoding leads to poor encoding and decoding performance, particularly in the transmission of large volumes of point cloud data.
Innovation Solution
Determine M neighborhood blocks corresponding to a current point in a reference picture and use attribute information from these blocks to calculate an attribute prediction value for the current point during encoding and decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional neighboring point determination methods are used in point cloud attribute encoding, then the encoding process is simple, but the accuracy of attribute prediction is poor
Solution Approach 1:
The patent divides the reference picture into multiple neighborhood blocks (first neighborhood block, second neighborhood block, third neighborhood block) around the current point. Each block contains specific neighboring points at different spatial positions and distances, allowing for more comprehensive attribute prediction by segmenting the search space into structured regions rather than treating all neighboring points uniformly.
Solution Approach 2:
The patent applies different selection criteria to different neighborhood blocks based on their local characteristics. For example, the first neighborhood block may contain points closer to the current point while the second and third blocks contain points at different orientations or distances. This local differentiation allows the system to select the most appropriate neighboring points for prediction based on the specific local geometry and attribute variation patterns.
2Measurement precision
If more neighboring points are considered for attribute prediction, then prediction accuracy improves, but computational complexity increases
Solution Approach 1:
The reference picture is segmented into multiple neighborhood blocks with defined spatial relationships to the current point. This segmentation allows the system to efficiently search within each block rather than searching the entire reference picture, reducing computational complexity while still considering multiple relevant neighboring points for accurate prediction.
Solution Approach 2:
The patent changes the parameter of neighborhood selection by defining multiple blocks with different spatial characteristics (distance, orientation, position) around the current point. This allows the system to adaptively select neighboring points based on the specific prediction needs while maintaining controlled computational complexity through the structured block definition.
Data Source
AI summary
A point cloud decoding method is provided. The point cloud decoding method includes: determining M neighborhood blocks corresponding to a current point in a reference picture of the current point, wherein the current point is a point in a point cloud whose attribute information is to be decoded, M being a positive integer; and determining an attribute prediction value of the current point based on attribute information of points comprised in the M neighborhood blocks.