Point Cloud Motion Coding via Absolute Value Binarization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional point cloud coding techniques face inefficiencies in coding motion information, particularly due to inaccurate quantization of translation vectors and lack of predictive coding, which affects motion compensation and coding length, especially when dealing with negative motion parameters.
Innovation Solution
The proposed method improves motion parameter coding by quantizing translation vectors with more precise control and using predictive coding techniques, binarizing motion parameters based on their distribution probability to reflect absolute values, rather than using complement formats.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If complement format is used to binarize negative motion parameters, then all negative values can be represented, but coding efficiency deteriorates due to not reflecting distribution probability
Solution Approach 1:
The patent changes the binarization parameter representation from complement format to absolute value format. Specifically, motion parameters are binarized to reflect their absolute values, and a separate sign bit indicates whether the original value was negative. This parameter change allows the coding to adapt to the distribution probability of motion parameters, improving coding efficiency while maintaining the ability to represent all negative values.
2Device complexity
If conventional quantization is used for translation vectors, then coding process is simple, but coding precision deteriorates affecting motion compensation
Solution Approach 1:
The patent introduces dynamic adaptive quantization for translation vectors. Instead of using fixed quantization steps, the quantization process adapts based on the characteristics of the motion parameters. The quantization step size is adjusted dynamically, allowing finer precision for small motion values and coarser quantization for large motions, thereby improving overall coding precision while maintaining reasonable complexity.
3Device complexity
If predictive coding is not used, then coding process is straightforward, but temporal redundancy increases affecting compression ratio
Solution Approach 1:
The patent applies predictive coding by performing motion parameter prediction before actual coding. The predictor uses previously decoded motion parameters to estimate current motion parameters, and only the prediction residuals are encoded. This preliminary prediction action removes temporal redundancy before the main coding process, improving compression ratio while adding manageable complexity through the prediction step.
Data Source
AI summary
Embodiments of the present disclosure provide a method for point cloud coding. The method comprises: obtaining, during a conversion between a current frame of a point cloud sequence and a bit-stream of the point cloud sequence, motion information of the current frame; determining a binarized representation of the motion information, the binarized representation at least reflecting an absolute value of the motion information; and performing the conversion based on the binarized representation of the motion information. Compared with the conventional solution, the proposed method can advantageously improve the accuracy of motion information coding and coding quality.