一种基于神经网络的三维点云可见性快速预测方法及系统

By constructing a point cloud visibility prediction system based on octree convolutional U-Net and lightweight MLP, the problems of low computational efficiency and poor robustness in 3D point cloud visibility determination are solved, and efficient and accurate point cloud visibility prediction is achieved.

CN120689857BActive Publication Date: 2026-07-17PEKING UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
PEKING UNIV
Filing Date
2025-05-26
Publication Date
2026-07-17

Smart Images

  • Figure CN120689857B_ABST
    Figure CN120689857B_ABST
Patent Text Reader

Abstract

本发明公布了一种基于神经网络的三维点云可见性快速预测方法及系统，构建包含特征提取器和可见性预测器的神经网络架构，其中特征提取器采用3D U‑Net结构，用于提取点云的视图无关特征；可见性预测器采用轻量级的多层感知器，用于根据提取的特征和视图方向，预测三维点云中每个点的可见性。设计多视点并行推理与动态特征复用机制，显著提升多视点场景下的推理速度；设计合成数据驱动的端到端训练策略，生成带有真实可见性标签的训练数据。本发明具有高效性、鲁棒性和泛化性，能够有效克服三维点云可见性预测中的噪声、低密度和复杂几何结构的技术问题，显著提升点云可见性预测的准确性和效率，适用于实时渲染、视角优化和表面重建等应用场景。
Need to check novelty before this filing date? Find Prior Art