3D Data Decoding Using Mesh Prediction and Arithmetic Residual Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing 3D data encoding schemes, such as those described in NPL 1, face inefficiencies in encoding and decoding mesh displacements, mesh motion information, and base meshes due to dependencies on syntax elements and contexts, leading to suboptimal encoding efficiency.
Innovation Solution
A 3D data decoding apparatus and encoding apparatus are designed to improve encoding efficiency by incorporating a mesh prediction unit to derive prediction positions for base mesh vertex positions and an arithmetic decoder/encoder that uses context to efficiently decode and encode prediction residuals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If arithmetic encoding scheme is used for mesh displacements, mesh motion information, and base mesh, then encoding is performed, but encoding efficiency is poor because performance depends on syntax elements and contexts
Solution Approach 1:
The patent applies preliminary action by performing mesh prediction to generate prediction values for base mesh vertex positions before the actual arithmetic encoding process. The prediction residual (difference between actual and predicted values) is then encoded, which simplifies the data and improves encoding efficiency by reducing the complexity and dependency on syntax elements and contexts.
2Manufacturing precision
If arithmetic encoding with context dependency is used, then encoding is performed, but encoding quality is suboptimal due to inefficiencies in handling prediction residuals
Solution Approach 1:
The patent extracts the prediction component from the base mesh vertex position data, separating it into a prediction value (derived from neighboring vertices) and a prediction residual. By encoding only the residual rather than the full vertex position, the patent improves encoding quality while maintaining efficiency, as the residual contains less information and can be encoded more effectively.
Data Source
AI summary
A 3D data decoding apparatus for decoding encoded data includes a mesh prediction unit configured to derive a prediction position of a base mesh vertex position from the encoded data, and an arithmetic decoder configured to arithmetically decode a prediction residual. The arithmetic decoder decodes a part of a beginning of a prefix of an absolute value of a coefficient of the prediction residual by using a context and adds the prediction position and the prediction residual to derive the base mesh vertex position.