Point Cloud Encoding Using Image Motion Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional encoding methods for 3D point clouds, such as those used in sensing systems for automatic operation vehicles, face inefficiencies due to the need for iterative calculations and increased computational load when estimating motion between frames, particularly when both intra-frame and inter-frame predictions are used.
Innovation Solution
An encoding system that includes a processor or circuit configured to determine the prediction method for point cloud encoding based on information from image encoding, selecting between intra-frame and inter-frame prediction using a motion vector, intensity, and color value differences to reduce computational load and enhance encoding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If iterative closest point (ICP) algorithm is used to estimate motion between frames, then motion estimation accuracy is improved, but calculation amount increases
Solution Approach 1:
The patent applies preliminary action by performing image encoding first to obtain motion vectors, which are then used as initial estimates for point cloud motion compensation. This preliminary motion estimation from the 2D image guide reduces the search space and computational complexity of the subsequent point cloud motion estimation, avoiding the need for full iterative ICP algorithms while maintaining accuracy.
Solution Approach 2:
The patent uses the motion vector from image encoding as an intermediary to transfer motion information to the point cloud encoding process. This intermediary motion parameter bridges the 2D image and 3D point cloud data, enabling efficient motion compensation without direct iterative calculation between point cloud frames.
2Productivity
If both intra-frame and inter-frame prediction modes are implemented, then encoding efficiency is improved, but device complexity increases
Solution Approach 1:
The patent implements dynamic prediction mode selection where the encoding system adaptively switches between intra-frame and inter-frame prediction modes based on the characteristics of the point cloud data and motion conditions. This dynamic approach optimizes encoding efficiency for different scenarios while managing complexity through conditional logic rather than always maintaining both modes at full complexity.
3Manufacturing precision
If motion compensation is performed for all points in the point cloud, then encoding accuracy is improved, but calculation amount increases
Solution Approach 1:
The patent applies local quality by performing motion compensation selectively based on point characteristics and local motion patterns. Rather than uniformly applying motion compensation to all points, the system identifies regions or points that benefit most from motion compensation and applies the processing locally, maintaining encoding accuracy where needed while reducing overall calculation amount.
Data Source
AI summary
To perform encoding processing on an image and encoding processing on a point cloud, and to reduce an amount of calculation of the point cloud encoding processing, in an encoding system, a first encoding unit configured to encode an image; and a second encoding unit configured to encode a point cloud are provided, and the second encoding unit performs encoding processing based on information that has been used for the encoding processing in the first encoding unit, for example, the magnitude of a motion vector and the like, to reduce a calculation amount of point cloud encoding processing.