Point Cloud Compression via 2D Projection and Patch Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Point clouds generated by sensors like LIDAR systems and 3-D cameras contain large amounts of spatial and attribute information, making them costly and time-consuming to store and transmit, which limits their use in real-time applications due to significant storage and network resource requirements.
Innovation Solution
A system that compresses point cloud data by dividing it into patches, generating patch images with spatial and depth information, and encoding these images using video encoding standards like HEVC or JPEG, allowing for efficient storage and transmission.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If point cloud data is stored and transmitted in its original format, then complete spatial and attribute information is preserved, but storage space and transmission time increase significantly
Solution Approach 1:
The point cloud data is divided into multiple patches based on spatial proximity and surface normal similarity. Each patch is processed independently through projection and encoding, enabling efficient compression while preserving local geometric and attribute details. This segmentation allows the system to manage large point clouds in manageable units without losing overall information integrity.
Solution Approach 2:
The patent creates 2D projection images as simplified representations of 3D point cloud patches. These projection images capture essential spatial and attribute information in a compressed format that requires significantly less storage space while maintaining the ability to reconstruct the original point cloud data when needed.
2Loss of information
If point cloud data is stored and transmitted in its original format, then complete spatial and attribute information is preserved, but transmission time increases significantly
Solution Approach 1:
By dividing the point cloud into patches and processing them independently, the system can transmit data in smaller, parallelizable units. This reduces the total transmission time while maintaining complete information through the preservation of both geometric projection data and attribute information for each patch.
Solution Approach 2:
The patent transforms 3D point cloud data into 2D projection images with associated depth and attribute maps. This parameter transformation from three-dimensional coordinates to two-dimensional projections with depth information enables significantly faster transmission while preserving all necessary spatial and attribute data for reconstruction.
3Quantity of substance
If point cloud data is compressed using traditional methods, then storage space is reduced, but information loss increases and real-time processing becomes difficult
Solution Approach 1:
The patent projects 3D point cloud patches onto 2D planes while preserving depth information through depth maps and normal vectors. This dimensional transformation enables efficient compression by representing 3D data in 2D format without losing spatial relationships or attribute information, achieving both compression and information preservation simultaneously.
Solution Approach 2:
By processing point clouds in localized patches rather than as a whole, the system can apply compression techniques that preserve local geometric fidelity and attribute accuracy. Each patch maintains its complete information set independently, ensuring no information loss even as overall data size is reduced through compression.
4Productivity
If point cloud data is compressed, then storage and transmission efficiency improves, but processing complexity increases
Solution Approach 1:
Dividing the point cloud into patches enables parallel processing of multiple independent units. This segmentation reduces the complexity of any single processing step while improving overall productivity through parallelization. Each patch can be projected, encoded, and transmitted independently, simplifying the processing pipeline.
Solution Approach 2:
The patent replaces complex 3D point cloud processing with 2D image processing techniques. By transforming the problem from three-dimensional geometry manipulation to two-dimensional projection and encoding, the system leverages mature, efficient image processing algorithms to reduce processing complexity while maintaining storage and transmission efficiency.
Data Source
AI summary
A system comprises an encoder configured to compress attribute information and/or spatial for a point cloud and/or a decoder configured to decompress compressed attribute and/or spatial information for the point cloud. To compress the attribute and/or spatial information, the encoder is configured to convert a point cloud into an image based representation. Also, the decoder is configured to generate a decompressed point cloud based on an image based representation of a point cloud. The encoder generates an occupancy map and may also encode the occupancy map as an image based representation. In some embodiments, a video encoder encodes image based representations of spatial information for the points of the point cloud, image based representations of attribute values for points of the point cloud, and an image based representation of an occupancy map for the spatial and attribute images.