Point Cloud Geometric Smoothing for Patch Boundary Artifacts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing point cloud compression technologies face challenges in addressing compression artifacts at patch boundaries, leading to distorted geometry and reduced quality in reconstructed 3D scenes, particularly in immersive applications like virtual reality and autonomous driving.
Innovation Solution
A geometric smoothing strategy is applied iteratively using various smoothing filters and an adaptive erosion scheme on the occupancy map to recover distorted geometry, ensuring high-quality reconstruction of point clouds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If compression is applied to point cloud data, then data transmission efficiency is improved, but geometry distortion and compression artifacts appear at patch boundaries
Solution Approach 1:
The patent extracts and processes patch boundary regions separately from the interior regions. By identifying patch boundaries in the occupancy map and applying targeted smoothing operations only to boundary points, the method removes compression artifacts where they occur most frequently without unnecessarily processing already-accurate interior points, thus resolving the contradiction between compression efficiency and geometry accuracy.
Solution Approach 2:
The patent applies different processing strategies to different regions: boundary points receive adaptive smoothing treatment while interior points retain their original compressed values. This local differentiation allows the system to improve geometry accuracy specifically where compression artifacts manifest (at boundaries) while maintaining overall compression efficiency by not over-processing interior regions.
2Manufacturing precision
If smoothing filters are applied to reduce compression artifacts, then geometry quality is improved, but computational complexity increases
Solution Approach 1:
The patent applies smoothing filters selectively only to patch boundary points rather than uniformly to all points in the point cloud. By identifying boundary regions in the occupancy map and limiting smoothing operations to these specific locations, the method improves geometry quality where artifacts occur while significantly reducing computational complexity compared to global smoothing approaches.
Solution Approach 2:
The patent applies smoothing operations to a subset of points (boundary points) rather than all points. This partial action approach focuses computational resources on the critical regions where compression artifacts manifest, achieving sufficient geometry quality improvement without the excessive computational cost of processing the entire point cloud.
3Manufacturing precision
If adaptive erosion scheme is applied on occupancy map, then boundary distortion is reduced, but processing time increases
Solution Approach 1:
The patent performs adaptive erosion on the occupancy map before reconstructing the point cloud. By preprocessing the occupancy map to remove boundary distortions in advance, the method ensures that subsequent point cloud reconstruction and smoothing operations work with already-corrected boundary information, reducing overall processing time compared to correcting distortions after reconstruction.
Solution Approach 2:
The patent segments the occupancy map processing into distinct stages: adaptive erosion of boundary regions, followed by point cloud reconstruction, and then targeted smoothing. This segmentation allows each operation to focus on specific tasks with optimized processing, reducing total processing time while maintaining boundary accuracy.
Data Source
Figure 1A~1B
Figure 2
Figure 3
AI summary
A method of video decoding performed in a video decoder includes receiving compressed geometric data corresponding to a three dimensional (3D) space. The method further includes receiving a low resolution occupancy map associated with the data cloud. The method further includes converting the low resolution occupancy map to a high resolution occupancy map. The method further includes reconstructing a point cloud using the compressed geometric data and the high resolution occupancy map. The method further includes performing a smoothing operation on one of the (i) reconstructed point cloud and (ii) the high resolution occupancy map.