基于PRPNet的三维表面重建方法  

3D Surface Reconstruction Method Based on PRPNet

作　　者：雷懂 宋俊锋 叶振[2] LEI Dong;SONG Junfeng;YE Zhen(School of In f ormation Science and Technology,Zhejiang Sci-Tech University,Hangzhou 310018,China;School of Mathematics and Computer Science,Lishui University,Lishui 323000,China)

机构地区：[1]浙江理工大学信息科学与工程学院,浙江杭州310018 [2]丽水学院数学与计算机学院,浙江丽水323000

出　　处：《软件工程》2024年第6期59-62,67,共5页Software Engineering

摘　　要：隐式神经网络用于三维表面重建时,存在重建物体的结构不准确、表面缺乏局部细节等问题,针对此问题,文章提出了一种基于PRPNet(点云残差编码网络)的三维表面重建方法。首先采用更深的网络结构且加入残差模块挖掘点云潜在的结构信息,加入PointMateBase模块,以增强局部细节表示能力;其次使用特征权重网络获取查询点的占用概率;最后通过区域增长的Marching Cubes算法提取三维表面。实验结果表明,PRPNet模型在ShapetNet和Synthetic Rooms数据集上的精度较DpConvONet模型相应数据集上的精度分别提升了2.5百分点和2.6百分点,能够有效提升三维表面重建性能。When implicit neural network is applied to 3D surface reconstruction,there are problems such as inaccurate structure of reconstructed objects and lack of local details on the surface.To address these problems,this paper proposes a 3D surface reconstruction method based on PRPNet(Point Cloud Residual Encoding Network).Firstly,a deeper network structure with a residual module is utilized to explore the latent structural information of point clouds,and a PointMateBase module is incorporated to enhance the representation capability of local details.Secondly,a feature weight network is employed to obtain the occupancy probability of query points.Finally,the 3D surface is extracted using the region-growing Marching Cubes algorithm.Experimental results demonstrate that the accuracy of PRPNet model on the ShapetNet and Synthetic Rooms datasets has improved by 2.5 percentage points and 2.6 percentage points respectively compared to DpConvONet model,indicating the effectiveness of enhancing the performance of 3D surface reconstruction.

关 键 词：三维表面重建 隐式神经网络 点云 残差模块 PointMateBase模块 特征权重网络 Marching Cubes算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]