PUFM++: Point Cloud Upsampling via Enhanced Flow Matching
By: Zhi-Song Liu , Chenhang He , Roland Maier and more
Recent advances in generative modeling have demonstrated strong promise for high-quality point cloud upsampling. In this work, we present PUFM++, an enhanced flow-matching framework for reconstructing dense and accurate point clouds from sparse, noisy, and partial observations. PUFM++ improves flow matching along three key axes: (i) geometric fidelity, (ii) robustness to imperfect input, and (iii) consistency with downstream surface-based tasks. We introduce a two-stage flow-matching strategy that first learns a direct, straight-path flow from sparse inputs to dense targets, and then refines it using noise-perturbed samples to approximate the terminal marginal distribution better. To accelerate and stabilize inference, we propose a data-driven adaptive time scheduler that improves sampling efficiency based on interpolation behavior. We further impose on-manifold constraints during sampling to ensure that generated points remain aligned with the underlying surface. Finally, we incorporate a recurrent interface network~(RIN) to strengthen hierarchical feature interactions and boost reconstruction quality. Extensive experiments on synthetic benchmarks and real-world scans show that PUFM++ sets a new state of the art in point cloud upsampling, delivering superior visual fidelity and quantitative accuracy across a wide range of tasks. Code and pretrained models are publicly available at https://github.com/Holmes-Alan/Enhanced_PUFM.
Similar Papers
Efficient Point Clouds Upsampling via Flow Matching
CV and Pattern Recognition
Makes blurry 3D shapes sharp and clear.
Towards Stable and Structured Time Series Generation with Perturbation-Aware Flow Matching
Machine Learning (CS)
Creates realistic data that changes suddenly.
Coupled Flow Matching
Machine Learning (Stat)
Makes complex data simpler, then rebuilds it perfectly.