3D Point Encoding Using Geometry Image Depth Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Image-based point cloud compression techniques face inefficiencies and quality issues when projecting 3D points with complex surfaces, leading to increased bit rates and resource wastage due to duplicated reconstructed points.
Innovation Solution
A method that assigns a dummy attribute to pixels with identical depth values in geometry images, avoiding the encoding of fake duplicated points, and reconstructs 3D points only when depth values differ, optimizing bit rate and resource usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If image-based point cloud compression techniques project 3D points onto 2D images, then compression efficiency is improved, but duplicated reconstructed points increase bit rate and resource consumption
Solution Approach 1:
The patent extracts and identifies duplicated reconstructed points from the projected point cloud data. By detecting points that map to the same 2D image coordinates, the system separates useful unique point data from redundant duplicated data, encoding only the unique points to reduce bit rate while maintaining compression efficiency.
Solution Approach 2:
The patent discards duplicated reconstructed points during the encoding process by identifying and removing redundant point data that maps to the same 2D coordinates. The unique point information is preserved and encoded, while duplicate information is discarded to reduce bit rate consumption without losing essential 3D point cloud data.
2Productivity
If image-based point cloud compression techniques project 3D points onto 2D images, then compression efficiency is improved, but resource wastage increases due to duplicated points
Solution Approach 1:
The patent extracts duplicated point information during the compression process by comparing 3D point projections onto 2D image planes. By identifying points that occupy the same 2D coordinates, the system separates redundant data from essential data, processing and encoding only unique points to reduce computational resource consumption while maintaining compression efficiency.
Solution Approach 2:
The patent discards duplicated reconstructed points during encoding by detecting redundant point mappings to the same 2D coordinates. This discarding of duplicate information reduces the amount of data requiring processing, storage, and transmission, thereby reducing energy and computational resource consumption while preserving compression efficiency through efficient encoding of unique points only.
3Reliability
If all reconstructed points are encoded including duplicates, then complete point cloud data is preserved, but bit rate increases unnecessarily
Solution Approach 1:
The patent extracts and identifies duplicated reconstructed points by comparing their 2D image coordinates. By separating unique point data from duplicate data, the system encodes only the essential unique point information while discarding redundant duplicates, thereby reducing bit rate while maintaining complete representation of the unique 3D point cloud structure.
Solution Approach 2:
The patent discards duplicated reconstructed points during the encoding process by detecting points that map to identical 2D coordinates. This selective discarding of duplicate information reduces the total bit rate required for encoding while preserving all unique point cloud data, achieving efficient compression without losing essential 3D geometric information.
Data Source
AI summary
The present embodiments relate to a method for encoding 3D points whose geometry is represented by geometry images and attribute is represented by an attribute image. The method checks whether the depth value of a pixel in a first of said geometry images and the depth value of a co-located pixel in a second of said geometry images are not the same (not identical values). When the depth value of a pixel in said first geometry image and the depth value of the co-located pixel in said second geometry image are not the same, then the method assigns (encodes), attribute of a 3D point defined from 2D spatial coordinates of said pixel in said first geometry image and the depth value of said co-located pixel in the second geometry image.