TVMC: Time-Varying Mesh Compression via Multi-Stage Anchor Mesh Generation

TL;DR

TVMC tackles the challenge of compressing time-varying meshes by introducing a multi-stage anchor mesh pipeline that ensures topology consistency while accurately modeling inter-frame motion. A fast octree-based topology alignment produces an initial anchor aligned to the previous frame, which is then refined by a Kalman-filter motion estimator into a coarse anchor, and finally by Quadric Error Metrics into a high-fidelity fine anchor. Adaptive displacement quantization further improves rate–distortion by allocating bits according to local geometric complexity. On MPEG V-DMC sequences, TVMC achieves BD-rate gains of $10.2\%$ to $16.9\%$ with acceptable overhead, demonstrating robust performance across diverse dynamic geometries.

Abstract

Time-varying meshes, characterized by dynamic connectivity and varying vertex counts, hold significant promise for applications such as augmented reality. However, their practical utilization remains challenging due to the substantial data volume required for high-fidelity representation. While various compression methods attempt to leverage temporal redundancy between consecutive mesh frames, most struggle with topological inconsistency and motion-induced artifacts. To address these issues, we propose Time-Varying Mesh Compression (TVMC), a novel framework built on multi-stage coarse-to-fine anchor mesh generation for inter-frame prediction. Specifically, the anchor mesh is progressively constructed in three stages: initial, coarse, and fine. The initial anchor mesh is obtained through fast topology alignment to exploit temporal coherence. A Kalman filter-based motion estimation module then generates a coarse anchor mesh by accurately compensating inter-frame motions. Subsequently, a Quadric Error Metric-based refinement step optimizes vertex positions to form a fine anchor mesh with improved geometric fidelity. Based on the refined anchor mesh, the inter-frame motions relative to the reference base mesh are encoded, while the residual displacements between the subdivided fine anchor mesh and the input mesh are adaptively quantized and compressed. This hierarchical strategy preserves consistent connectivity and high-quality surface approximation, while achieving an efficient and compact representation of dynamic geometry. Extensive experiments on standard MPEG dynamic mesh sequences demonstrate that TVMC achieves state-of-the-art compression performance. Compared to the latest V-DMC standard, it delivers a significant BD-rate gain of 10.2% ~ 16.9%, while preserving high reconstruction quality. The code is available at https://github.com/H-Huang774/TVMC.

Figures (8)

• Figure 1: Left: In V-DMC, unaligned topology between the target base mesh and the reference base mesh prevents motion tracking, making inter-frame coding infeasible. Right: A toy example showcasing the effectiveness of the proposed multi-stage anchor mesh generation in TVMC. The initial anchor mesh maintains the reference topology but poorly approximates the target geometry, resulting in low reconstruction quality and a large size. The coarse anchor mesh then corrects motion-induced distortions, improving accuracy and reducing displacement. Finally, the fine anchor mesh is further optimized to enhance geometric fidelity, yielding the highest reconstruction quality and minimal displacement. The orange arrow indicates an example of the residual displacement for a vertex. The Size corresponds to the compressed displacement bitstream size, while 'Displacement' denotes the example displacement magnitude per mesh.
• Figure 2: A toy example of V-DMC.
• Figure 3: Detailed architecture of inter-frame coding in the V-DMC standard for mesh sequences with consistent topology. $D$ represents the displacements.
• Figure 4: Overview of the proposed TVMC framework. Initially, a fast topology alignment module generates an initial anchor mesh whose topology aligns with the reference base mesh $B_{t-1}$, while preserving the primary geometric characteristics of the target input mesh $M_t$. A Kalman filter-based motion estimation module then generates a coarse anchor mesh $B_{t,c}$ by compensating inter-frame motions. Subsequently, a QEM-based refinement module optimizes the coarse anchor mesh, producing a fine anchor mesh $B_{t,f}$ with improved geometric fidelity. Inter-frame motion vectors $m$ computed between $B_{t-1}$ and $B_{t,f}$ are encoded using arithmetic coding. Finally, displacements $D$ between the subdivided mesh $M_s$ derived from $B_{t,f}$ and $M_t$ are adaptively quantized and encoded using video coding techniques, resulting in efficient reconstruction of the input mesh $\widehat{M}_t$.
• Figure 5: Example of coarse anchor mesh generation for the second and third frames.