Point Cloud Geometry Encoding With Sensor-Based Radius Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing point cloud codecs fail to combine simplicity, low latency, and high compression performance, particularly when encoding sparse geometry data sensed by sensors on moving vehicles, and do not adapt to the order of sensing effectively.
Innovation Solution
A method of encoding and decoding point cloud geometry data using a coarse representation in a two-dimensional space, where each point is associated with a sensor index and a sample index, selecting a predicting radius from points sensed by different sensors to improve prediction accuracy and reduce residual radius dynamics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If existing point cloud codecs are used to encode sparse geometry data, then compression is achieved, but compression performance is insufficient and complexity increases
Solution Approach 1:
The patent segments the point cloud data processing into distinct stages: organizing points by sensor and time, separating radius prediction from coordinate encoding, and dividing the encoding into coarse and fine components. This segmentation allows each stage to be optimized independently, improving compression performance without proportionally increasing overall complexity.
Solution Approach 2:
The patent performs preliminary organization of point cloud data by sensor index and sample index before encoding, and pre-calculates prediction radii from previously decoded points. This preliminary action structures the data in a way that enables more efficient compression during the actual encoding phase, reducing the computational burden during real-time processing.
2Loss of information
If complex prediction methods are used to improve compression, then compression performance increases, but latency increases
Solution Approach 1:
The prediction mechanism uses previously decoded points within the same data structure to generate prediction radii, making the system self-sufficient without requiring external reference data or complex inter-frame predictions. This self-service approach reduces processing overhead and latency while maintaining compression effectiveness.
Solution Approach 2:
The patent applies prediction only to the radius component of point coordinates, rather than all spatial dimensions, and uses a simplified prediction model based on previous points. This partial application of prediction reduces computational complexity and latency while still achieving improved compression performance compared to encoding all coordinates without prediction.
3Ease of manufacture
If traditional encoding methods are used, then encoding is straightforward, but residual radius dynamics are high reducing efficiency
Solution Approach 1:
The patent changes the parameter being encoded from absolute radius values to residual radii (differences between actual and predicted radii). This parameter transformation reduces the dynamic range of values requiring encoding, improving encoding efficiency while maintaining simplicity through the use of differential encoding techniques.
4Loss of information
If high compression is achieved through complex methods, then bitrate is reduced, but implementation complexity increases
Solution Approach 1:
The encoding process is segmented into distinct, manageable components: data organization by sensor/time, radius prediction, residual calculation, and final encoding. Each segment can be implemented and optimized independently, reducing overall implementation complexity while achieving high compression through the cumulative effect of all segments.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
There is provided methods and apparatus of encoding/decoding point cloud geometry 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 (120) a data (Ipred) representative of a predicting radius (rpred), and encoding (130), into the bitstream, a residual radius (rres) between a radius (r1) associated with an occupied coarse point and the predicting radius (rpred).