Point Cloud Spatial Quantization for Fractional Coordinate Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing point cloud compression methods struggle with efficient encoding and decoding of non-integer coordinate values, leading to inefficiencies in transmitting large volumes of 3D data for applications like VR/AR, particularly in dynamic scenes where millions of point clouds are generated per second.
Innovation Solution
The proposed solution involves primary and secondary quantization positions within quantization cells, using a combination of coordinate transformation, voxelization, octree geometry encoding, and arithmetic encoding to optimize the compression process, ensuring efficient encoding and decoding of point clouds with fractional coordinates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If traditional compression methods are used for point clouds with non-integer coordinates, then the transmission of large volumes of 3D data becomes feasible, but the encoding and decoding efficiency deteriorates due to the complexity of handling fractional coordinates
Solution Approach 1:
The patent segments the point cloud data processing into distinct stages: coordinate transformation to convert fractional coordinates to integer coordinates, voxelization to discretize the transformed coordinates, and separate encoding paths for occupied and unoccupied voxels. This segmentation allows each stage to handle specific aspects of the data, improving overall processing efficiency while maintaining the ability to handle large volumes of point cloud data with non-integer coordinates
Solution Approach 2:
The patent introduces an intermediary coordinate transformation step that acts as a mediator between the original fractional coordinate system and the final compressed representation. By transforming fractional coordinates to integer coordinates through a deterministic function before compression, the system enables efficient integer-based processing while preserving the ability to reconstruct original positions, thus resolving the efficiency problem without sacrificing data fidelity
2Measurement precision
If high fidelity representation with thousands or millions of points is maintained, then the accuracy of point cloud representation is improved, but the transmission bandwidth requirement increases significantly
Solution Approach 1:
The patent extracts and processes occupied and unoccupied voxel information separately through distinct encoding paths. By extracting only the necessary information about occupied voxels (which contain actual point data) and using efficient entropy coding for unoccupied voxels, the system maintains high fidelity representation of all points while significantly reducing the transmission bandwidth required compared to transmitting complete point cloud data
Solution Approach 2:
The patent changes the parameter representation from raw fractional coordinates to transformed integer coordinates through a deterministic coordinate transformation. This parameter change enables more efficient compression while preserving the ability to reconstruct original positions with high precision, thus maintaining measurement precision while reducing transmission data volume
3Manufacturing precision
If deterministic quantization is applied to handle fractional coordinates, then the reconstruction accuracy is improved, but the complexity of the encoding process increases
Solution Approach 1:
The patent applies preliminary coordinate transformation and voxelization before the main compression process. By pre-processing the fractional coordinates through deterministic transformation to integer coordinates and organizing them into a voxel grid structure in advance, the system establishes a foundation that simplifies subsequent encoding operations while ensuring reconstruction accuracy is maintained throughout the compression pipeline
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An encoding apparatus (110) for compression of a point cloud (100) is disclosed. The encoding apparatus comprises a processing circuitry (111) configured to generate based on the point cloud (100) a quantized point cloud. Each point (101) of the point cloud (100) is located within one of a plurality of quantization cells and at least one quantization cell defines a primary quantization position and at least one secondary quantization position. The processing circuitry (111) is configured to generate the quantized point cloud by shifting each point (101) of the point cloud (100) located within the at least one quantization cell depending on its position within the at least one quantization cell either to the primary quantization position or to the at least one secondary quantization position. The processing circuitry (111) is further configured to encode the quantized point cloud. Moreover, a corresponding decoding apparatus (130) is disclosed.