Point Cloud Filtering Using Nearest Neighbor Coefficients
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing geometry-based point cloud compression (G-PCC) encoding and decoding framework suffers from severe distortion due to basic reconstruction methods, leading to significant differences between the reconstructed and initial point clouds, which affects the overall quality of the compressed point cloud.
Innovation Solution
An encoding and decoding method that determines filtering coefficients based on the initial and reconstructed point clouds, filters the reconstructed point cloud using nearest points, and encodes the filtering identification and coefficients into a bitstream to optimize the point cloud quality, considering PSNR and rate-distortion trade-offs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If basic reconstruction is performed on the initial point cloud, then the encoding process is simple, but the difference between reconstructed and initial point cloud is large, leading to severe distortion
Solution Approach 1:
The patent applies preliminary filtering action by determining filtering coefficients before final reconstruction. The encoder filters the reconstructed point cloud using K nearest points and filtering coefficients to reduce distortion before output, preventing quality degradation rather than correcting it after the fact.
Solution Approach 2:
The patent changes the parameters of the reconstructed point cloud by applying filtering coefficients to K target points (including the first point and K-1 nearest adjacent points). This transforms the reconstructed point cloud parameters to better match the initial point cloud, reducing distortion while maintaining encoding efficiency.
2Manufacturing precision
If filtering coefficients are encoded and transmitted in the bitstream, then the reconstructed point cloud quality is improved, but the bitrate increases
Solution Approach 1:
The patent applies partial filtering action by selecting only K target points (where K is a small integer greater than 1) for filtering rather than processing all points in the point cloud. This partial application of filtering reduces the quality improvement benefit while significantly reducing the bitrate overhead for transmitting filtering coefficients.
Solution Approach 2:
The encoder determines filtering identification information and filtering coefficients in advance before encoding them into the bitstream. This preliminary determination allows the decoder to efficiently reconstruct the filtered point cloud without requiring complex real-time calculations, reducing overall system complexity while maintaining quality.
3Manufacturing precision
If K nearest points are used for filtering, then the point cloud quality is improved, but the encoding and decoding complexity increases
Solution Approach 1:
The patent applies local filtering quality by focusing the filtering operation on K target points (the first point and its K-1 nearest adjacent points) rather than the entire point cloud. This localized approach improves quality at critical locations while minimizing the overall computational complexity of the filtering process.
Solution Approach 2:
The encoder performs preliminary determination of filtering identification information and filtering coefficients before transmission. The decoder then simply applies these pre-determined coefficients to the K target points, significantly reducing decoding complexity compared to performing complex optimization calculations in real-time.
Data Source
AI summary
A decoding method includes the following. The bitstream is decoded to determine filtering identification information, where the filtering identification information is used to determine whether to filter a reconstructed point cloud. When the filtering identification information indicates to filter the reconstructed point cloud, the bitstream is decoded to determine filtering coefficients. K target points corresponding to a first point in the reconstructed point cloud are filtered with the filtering coefficients to determine a filtered point cloud corresponding to the reconstructed point cloud, where the K target points include the first point and (K−1) nearest points adjacent to the first point, K is an integer greater than 1, and the first point represents any point in the reconstructed point cloud.