Mesh Displacement Lifting Offsets for Bias-Aware Dynamic Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video-based dynamic mesh coding technologies face inefficiencies due to bias in transform coefficients, leading to suboptimal coding efficiency and accuracy, particularly in extended reality applications.
Innovation Solution
Implementing a lifting transform with an offset to compensate for non-zero mean distributions in transform coefficients, combined with hierarchical flag mechanisms and delta coding to enhance encoding and decoding processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a lifting transform is applied to mesh displacement vectors, then compression efficiency is improved, but bias in transform coefficients occurs leading to reduced coding accuracy
Solution Approach 1:
The patent applies parameter changes by introducing an offset parameter to the lifting transform. This offset is determined based on the mean of transform coefficients and is applied to adjust the coefficients before quantization. By changing the parameter distribution through this offset adjustment, the transform coefficients are centered closer to zero, reducing bias while maintaining compression efficiency. The offset value is signaled in the bitstream to enable accurate reconstruction at the decoder.
2Device complexity
If transform coefficients are quantized directly without offset adjustment, then encoding complexity is reduced, but coding efficiency deteriorates due to non-zero mean distribution
Solution Approach 1:
The patent implements preliminary action by performing offset adjustment on the transform coefficients before quantization. This preliminary step centers the coefficient distribution around zero, which optimizes the subsequent quantization process. The offset is calculated from the mean of the transform coefficients and applied systematically prior to quantization, ensuring that the quantized coefficients achieve better compression efficiency without requiring complex post-processing.
3Measurement precision
If offset values are signaled for every patch, then coding accuracy is improved, but data transmission requirements increase
Solution Approach 1:
The patent applies local quality by selectively signaling offset values based on patch-specific characteristics. Rather than uniformly applying offsets to all patches, the system determines whether each patch requires offset adjustment based on its transform coefficient distribution. This selective approach allows accurate reconstruction where needed while minimizing overhead in cases where the offset is zero or unnecessary, optimizing the trade-off between accuracy and data transmission.
Data Source
AI summary
Certain aspects of the disclosure provide for encoding and decoding of mesh data using a lifting transform and signaling lifting offsets to address lifting transform bias. A three-flag signaling mechanism can be employed that implements sequence, frame, and patch flags to regulate the processing and transmission of control parameters. Delta coding can also be performed to calculate and transmit a delta value rather than an actual offset value for inter-patch and merge-patch modes that include a reference from which the offset can be determined based on the delta value. Furthermore, support is provided for variable subdivision iteration counts, with patch-specific processing that enables the independent compression of different geometric regions.