RAHT Inter Prediction with AC Scaling for Point Cloud Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing point cloud attribute inter prediction in region-adaptive hierarchical transform (RAHT) assumes a correlation coefficient of 1, which is rarely true, leading to inefficient predictions, especially for inter-prediction in point-clouds.
Innovation Solution
Scale the reference AC value to derive inter-prediction in RAHT, applying scaling factors based on various criteria such as motion, texture, and geometric differences, and signal these factors to the decoder.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If inter prediction in RAHT assumes a correlation coefficient of 1, then the prediction process is simple, but the prediction accuracy deteriorates because this assumption is rarely true in actual point-clouds
Solution Approach 1:
The patent changes the fixed parameter (correlation coefficient = 1) to a variable parameter by introducing scaling factors that adapt to actual point cloud characteristics. The scaling factors are derived from motion information, texture complexity, and geometric differences, allowing the prediction process to adjust to varying correlation levels in different regions and contexts.
Solution Approach 2:
The patent introduces dynamic scaling factors that vary across different regions and prediction contexts rather than using a static correlation coefficient. These scaling factors are computed based on local characteristics such as motion vectors, texture complexity, and geometric differences, making the prediction process adaptive and dynamic.
2Measurement precision
If scaling factors are applied to AC values for inter prediction, then prediction accuracy improves by adapting to actual correlation, but device complexity increases due to additional scaling operations and factor signaling
Solution Approach 1:
The patent applies different scaling factors to different regions and prediction contexts based on local characteristics. Instead of using a uniform scaling approach, the method computes scaling factors specific to each prediction unit or region based on local motion information, texture complexity, and geometric differences, thereby optimizing prediction accuracy where needed while avoiding unnecessary complexity elsewhere.
Solution Approach 2:
The patent selectively applies scaling operations only where beneficial rather than universally. Scaling factors are computed and applied based on specific conditions such as motion magnitude, texture complexity thresholds, and geometric difference measures, avoiding unnecessary scaling operations in regions where they would not improve prediction accuracy.
3Reliability
If scaling factors are signaled to the decoder, then the decoder can perform accurate reconstruction, but the bitstream size increases due to additional signaling data
Solution Approach 1:
The patent signals scaling factors selectively rather than for all prediction units. Scaling factors are transmitted only for regions or contexts where they provide significant improvement in prediction accuracy, determined by thresholds on motion magnitude, texture complexity, or geometric difference. This partial signaling approach balances reconstruction accuracy with bitstream efficiency.
Solution Approach 2:
The patent changes the representation of scaling factors from precise floating-point values to quantized or rounded values suitable for efficient coding. The scaling factors are transformed into a compact representation that can be signaled with fewer bits while maintaining sufficient accuracy for reconstruction, thereby reducing bitstream overhead.
Data Source
AI summary
A mechanism for processing video data is disclosed. The mechanism may include determining to scale an alternating current (AC) value to derive an inter prediction mode in a region-adaptive hierarchical transform (RAHT). A conversion is performed between a visual media data and a bitstream based on the inter prediction mode.