3D Point Cloud Motion Compensation for Efficient Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is a demand for improving encoding efficiency in three-dimensional data encoding processes, particularly for three-dimensional point clouds, which are used in applications such as autonomous vehicles and infrastructure inspection, where the massive amount of data necessitates efficient compression and transmission.
Innovation Solution
A method for encoding three-dimensional point cloud data involves selecting a moving method, determining a second area based on the selected method, moving a second point cloud to the first area, and encoding geometry information using the moved point cloud, along with generating a bitstream that includes the encoded geometry information and moving information, where the moving methods include rotation and translation of the point cloud about a vertical axis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If point cloud data is compressed using traditional encoding methods, then data transmission efficiency is improved, but encoding accuracy and prediction precision deteriorate
Solution Approach 1:
The patent applies preliminary action by performing moving compensation on point cloud data before encoding. The decoder pre-processes the point cloud data by applying rotation and translation transformations based on motion information, ensuring that the prediction process starts with already-aligned data, thereby improving both encoding accuracy and transmission efficiency
Solution Approach 2:
The patent introduces motion information as an intermediary element that mediates between the encoder and decoder. This motion information contains rotation and translation parameters that are transmitted separately from the point cloud data, allowing the decoder to accurately reconstruct and align the point cloud without transmitting redundant alignment data, thus improving both compression efficiency and accuracy
2Measurement precision
If more motion information is transmitted to improve prediction accuracy, then encoding precision is improved, but data amount and complexity increase
Solution Approach 1:
The patent extracts motion information (rotation and translation parameters) from the point cloud data and handles it separately. By taking out the motion compensation parameters as independent metadata, the system improves prediction accuracy without increasing the complexity of the main point cloud data structure, as the motion parameters are compact and efficiently encoded
3Measurement precision
If point cloud data is rotated and translated to match reference areas, then prediction accuracy is improved, but processing time and computational load increase
Solution Approach 1:
The patent performs rotation and translation transformations as preliminary actions before the main encoding process. By pre-aligning the point cloud data using motion compensation, the system improves prediction accuracy while minimizing processing time during the actual encoding phase, as the computationally intensive transformations are done once in advance using efficiently encoded motion parameters
Data Source
AI summary
A three-dimensional data encoding method of encoding geometry information of a first three-dimensional point cloud located in a first area includes: selecting a moving method from among moving methods; determining a second area based on the moving method selected and the first area; moving a second three-dimensional point cloud located in the second area determined, to the first area, using a method in accordance with the moving method; encoding the geometry information of the first three-dimensional point cloud, based on encoded geometry information of the second three-dimensional point cloud moved to the first area; and generating a bitstream including the geometry information of the first three-dimensional point cloud encoded and moving information indicating the moving method. The moving methods include a method for rotating the second three-dimensional point cloud about a vertical axis of a mobile body that includes a sensor configured to generate the first three-dimensional point cloud.