High Efficiency Wiener Filter-based Point Cloud Quality Enhancement for MPEG G-PCC

TL;DR

The paper tackles the challenge of improving attribute quality and rate-distortion performance for dynamic point clouds in MPEG G-PCC. It presents a Wiener-filter-based framework comprising BWF, CIWF, VCWF, and M-KNN, enabling fast, local-statistics–driven quality enhancement integrated into both encoder and decoder pipelines. Key contributions include a Morton-code–based fast nearest-neighbor search, coefficients inheritance across GOFs, and variance-based Luma categorization to tailor filtering, yielding notable BD-rate improvements across Luma and Chrominance components, with perceptual gains confirmed subjectively. The approach balances performance and complexity, offering a practical path to higher-quality dynamic point-cloud coding within the G-PCC standardscape.

Abstract

Point clouds, which directly record the geometry and attributes of scenes or objects by a large number of points, are widely used in various applications such as virtual reality and immersive communication. However, due to the huge data volume and unstructured geometry, efficient compression of point clouds is very crucial. The Moving Picture Expert Group is establishing a geometry-based point cloud compression (G-PCC) standard for both static and dynamic point clouds in recent years. Although lossy compression of G-PCC can achieve a very high compression ratio, the reconstruction quality is relatively low, especially at low bitrates. To mitigate this problem, we propose a high efficiency Wiener filter that can be integrated into the encoder and decoder pipeline of G-PCC to improve the reconstruction quality as well as the rate-distortion performance for dynamic point clouds. Specifically, we first propose a basic Wiener filter, and then improve it by introducing coefficients inheritance and variance-based point classification for the Luma component. Besides, to reduce the complexity of the nearest neighbor search during the application of the Wiener filter, we also propose a Morton code-based fast nearest neighbor search algorithm for efficient calculation of filter coefficients. Experimental results demonstrate that the proposed method can achieve average Bjøntegaard delta rates of -6.1%, -7.3%, and -8.0% for Luma, Chroma Cb, and Chroma Cr components, respectively, under the condition of lossless-geometry-lossy-attributes configuration compared to the latest G-PCC encoding platform (i.e., geometry-based solid content test model version 7.0 release candidate 2) by consuming affordable computational complexity.

Figures (13)

• Figure 1: Encoding framework of GeS-TM. The modules within the dotted boxes are optional to perform.
• Figure 2: Results of octree segmentation, where the octree structure is divided into six layers. The blocks, marked in red and labeled as A, B, C, and D, are selected for local stationary analysis.
• Figure 3: Mean and variance distribution of the 816 blocks.
• Figure 4: Mean, variance and autocovariance ($\gamma _{1}$) distribution of subblocks in block A, B, C and D.
• Figure 5: Illustration of potential nearest neighbor positions. The brown block denotes the current point position, the blue blocks represent its coplanar nearest neighbor positions which are not empty, while the grey blocks represent its coplanar nearest neighbor positions which are empty.