3D Generative Model Encoding for Arbitrary-View Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The massive amount of data in three-dimensional point clouds necessitates efficient compression for accumulation and transmission, particularly in applications like computer vision and three-dimensional map data, where existing methods are inadequate for reducing data volume.
Innovation Solution
An encoding device and decoding device utilizing circuitry and memory to generate and decode bitstreams from three-dimensional data generative models, allowing output of two-dimensional images from arbitrary viewpoints, thereby compressing data for storage and transmission.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If three-dimensional point cloud data is used to represent spatial information, then the completeness and accuracy of three-dimensional representation is improved, but the data volume increases significantly requiring efficient compression
Solution Approach 1:
The patent creates a compressed three-dimensional map data structure that copies only the essential spatial relationships and positional information from complete point cloud data, enabling efficient storage and transmission while preserving the ability to retrieve accurate spatial information when needed
Solution Approach 2:
The patent extracts and stores only the critical spatial coordinates and positional data from comprehensive three-dimensional point cloud information, separating essential navigation data from redundant details to reduce data volume while maintaining measurement precision for spatial representation
2Reliability
If comprehensive three-dimensional map data is stored for autonomous navigation, then the reliability of navigation decisions is improved, but the storage capacity and transmission bandwidth requirements increase
Solution Approach 1:
The patent segments three-dimensional map data into discrete coordinate sets representing different spatial features and navigation elements, organizing data by functional importance rather than storing complete point clouds, which reduces storage requirements while maintaining navigation reliability through selective data retrieval
Solution Approach 2:
The patent transforms comprehensive three-dimensional point cloud data into a parameterized coordinate representation that captures essential spatial relationships using fewer parameters, changing the data structure from dense point clouds to optimized coordinate sets that reduce storage volume while preserving navigation decision reliability
Data Source
AI summary
An encoding device includes circuitry and memory coupled to the circuitry. In operation, the circuitry: obtains a first three-dimensional data generative model corresponding to a first time and a second three-dimensional data generative model corresponding to a second time; and generates a bitstream by encoding the first three-dimensional data generative model obtained and the second three-dimensional data generative model obtained. When receiving viewpoint information including a viewpoint and a line-of-sight direction, each of the first three-dimensional data generative model and the second three-dimensional data generative model outputs a two-dimensional image of a subject as viewed from the viewpoint and the line-of-sight direction.