Point Cloud Geometry Coding with Ordered Coarse Points
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing point cloud compression technologies fail to effectively combine encoding and decoding simplicity, low latency, and high compression performance, particularly for sparse geometry data sensed by spinning sensors, which is crucial for applications like autonomous driving where real-time data processing is essential.
Innovation Solution
A method of encoding and decoding point cloud geometry data using ordered coarse points in a two-dimensional space, where each point is represented by a sensor index and a sample index, allowing for the selection of a predicting radius from different sensor indices to improve prediction accuracy and reduce residual radius dynamics, thereby enhancing compression efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional point cloud compression methods are used, then encoding and decoding simplicity is maintained, but compression performance is insufficient and latency is high
Solution Approach 1:
The patent segments the point cloud data processing into distinct stages: geometry-based compression for sparse data and attribute-based compression for dense data. This segmentation allows each stage to be optimized independently, improving overall compression performance while maintaining manageable complexity for real-time processing
Solution Approach 2:
The patent transforms the traditional 3D point cloud representation into a 2D projection space for processing. By projecting points onto a 2D plane and using 2D spatial prediction, the method reduces computational complexity and enables more efficient compression algorithms to be applied, thereby reducing latency while improving compression ratios
2Productivity
If geometry-based compression is applied to sparse point cloud data, then compression efficiency improves, but handling transitions between objects becomes difficult
Solution Approach 1:
The patent introduces an occupancy map as an intermediary data structure that tracks which spatial regions contain points. This occupancy map serves as a mediator between geometry-based compression and object transitions, enabling efficient detection and handling of object boundaries without compromising compression efficiency
Solution Approach 2:
The patent performs preliminary classification of points into geometry-based and attribute-based groups before compression. By pre-identifying sparse regions suitable for geometry-based compression and preparing occupancy information in advance, the method handles object transitions more smoothly while maintaining high compression efficiency
3Reliability
If multiple sensors are used to sense point cloud data, then data completeness improves, but encoding complexity increases
Solution Approach 1:
The patent merges data from multiple sensors by projecting all sensor points into a unified 2D coordinate space. This merging approach consolidates multiple sensor inputs into a single processed representation, improving data completeness while reducing encoding complexity through unified processing rather than separate handling of each sensor
Data Source
AI summary
Methods and apparatus of encoding/decoding point cloud geometry may use data sensed by at least one sensor associated with a sensor index. The point cloud geometry data is represented by ordered coarse points occupying some discrete positions of a set of discrete positions of a two-dimensional space, each occupied coarse point is associated with a radius based on a distance of the point of the point cloud from a referential. The encoding/decoding method comprises encoding/decoding a data representative of a predicting radius, and encoding, into the bitstream, a residual radius between a radius associated with an occupied coarse point and the predicting radius.