面向室内三维模型重建的神经辐射场网络

打开文本图片集

中图分类号：TP391 文献标志码：A 文章编号：1008-0562（2025）05-0627-07

Neural radiation field network for indoor 3D model reconstruction

XUEChenxia'

（1.Material Supply Center of Baorixile Energy Company Limited， China Energy InvestmentCorporation，Hulunbuir O21ooo，China;2.Ordos Institute ofLiaoning Technical University， Ordos017000， China）

Abstract：Aiming at the problems of insufficient robustness and poor reconstruction effct of weak texture regions in indoor 3Dreconstruction，a new indoor 3Dreconstruction method M-HashRecon is proposed basedon the principle of neural radiation field.The algorithmutilizes a pointcloud selection module to extract the key point cloud information，and introduces multi-resolution hash coding to realize the multi-scale feature index of the point cloud.The residual module is designed to optimize the performance of the model and improve the training efficiencyof the deep network.Experiments are carredout in four typical scenarios of ScanNet dataset， and the experimental results and the convergence ofthe model are analyzed.The research results show that the Fscore comprehensive index of the algorithm is significantlybeter than that of the comparison algorithm，and the reconstruction accuracy of multiple scenes is high and the stability is good.The research conclusions can provide reference for the design of subsequent high-precision indoor three-dimensional reconstruction system.

Keywords： indoor 3D reconstruction;deep guided sampling；multi-resolution hash coding;residual module; ScanNet dataset

0引言

随着科技的不断发展，智能化技术不断进步，利用计算机视觉和传感技术，将室内环境中的物体、结构和布局信息转换为高质量的三维模型已成为研究的热点。（剩余12046字）