Point Cloud Predictive Geometry Coding for Azimuthal Residuals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing point cloud compression techniques require dependencies on radius values for parsing azimuthal angle residuals, leading to delayed parsing and increased decoding time.
Innovation Solution
Parsing or signaling of azimuthal angle residuals for a current point in a point cloud is performed independently of the radius value of the current or previous point, allowing parallel processing and reducing decoding time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If parsing of azimuth residual value requires reconstructed radius values, then parsing accuracy may be maintained, but decoding time increases and parallel processing is prevented
Solution Approach 1:
The patent segments the parsing process into independent components: azimuth residual value parsing is separated from radius value reconstruction. This allows the azimuth parsing to proceed independently without waiting for radius reconstruction, enabling parallel processing while maintaining necessary accuracy through the use of predictor values and residual corrections.
Solution Approach 2:
The patent applies preliminary action by using predictor values (pre-computed azimuth values) before the actual parsing of residual values. The predictor values are generated in advance based on previous point data, allowing the residual parsing to start immediately without waiting for complete radius reconstruction, thus reducing decoding time while maintaining accuracy.
2Reliability
If parsing of azimuth residual value waits for radius reconstruction, then data dependency is maintained, but processing efficiency decreases
Solution Approach 1:
The patent introduces an intermediary mechanism using predictor values and residual values. Instead of directly depending on reconstructed radius values, the system uses predictor azimuth values (computed from previous points) as intermediaries, with residual values serving as corrections. This intermediary approach maintains data reliability while eliminating the bottleneck of sequential processing.
Solution Approach 2:
The patent changes the parameter representation from direct radius-dependent azimuth values to a transformed representation using predictor values and residuals. This parameter transformation allows the parsing to proceed independently of radius reconstruction status, improving processing efficiency while maintaining the mathematical relationship necessary for accuracy through the residual correction mechanism.
3Measurement precision
If sequential parsing is used to maintain data accuracy, then parsing precision is preserved, but decoding speed decreases
Solution Approach 1:
The patent segments the parsing operations into independent parallel tasks: predictor value computation, residual value parsing, and radius reconstruction can proceed simultaneously. Each segment maintains its own data dependencies locally, allowing parallel execution while preserving overall parsing precision through the coordinated use of predictor and residual values.
Solution Approach 2:
The patent performs preliminary computation of predictor azimuth values before the main parsing loop. These predictor values are pre-computed based on previous point data and stored, allowing the main parsing process to immediately use them without waiting for sequential radius reconstruction, thus increasing decoding speed while maintaining precision through residual corrections.
Data Source
AI summary
A method of processing a point cloud includes determining that angular mode is enabled for encoding or decoding a current point of points in the point cloud, and parsing or signaling information for an azimuthal angle residual value for the current point independent of a radius value of the current point or a radius value of a previous point of the point cloud that is previous to the current point in decoding order. The azimuthal angle residual value is based on a difference between an azimuthal angle value of the current point and a predictor azimuthal angle value of the current point, and the azimuthal angle value of the current point is indicative an azimuthal angle of the current point based on a laser used to capture the points of the point cloud.