摘要
现有的图像拼接方法在处理虚拟现实环境下的散乱点云图像时,存在三维几何特征利用不充分的问题,导致拼接精度较低,难以满足虚拟现实场景的高精度需求。基于上述背景,文章研究了虚拟现实环境下散乱点云图像智能化拼接处理方法。该方法利用双边滤波与体素下采样对原始点云图像进行预处理,结合基于K近邻的法线估计与ISS评分机制对点云特征进行捕捉,并利用主成分分析实现点云全局粗配准。继而采用ICP算法结合法线夹角筛选与曲率加权误差函数进行精配准,优化局部拼接精度,并通过高斯渲染优化与基于K近邻的带宽调节函数增强拼接效果。实验结果表明,与2种对比方法相比,该方法提升了散乱点云的拼接精度。
Existing image stitching methods face the issue of insufficient utilization of three-dimensional geometric features when processing scattered point cloud images in virtual reality environments,leading to low stitching accuracy and failing to meet the high-precision requirements of virtual reality scenes.Based on this background,an intelligent method for stitching scattered point cloud images in virtual reality environments is studied.This method preprocesses the original point cloud image using bilateral filtering and voxel downsampling,then captures point cloud features through normal estimation based on K-nearest neighbors and the ISS scoring mechanism.Principal component analysis is used to achieve global coarse registration of the point cloud.Next,ICP algorithm combined with normal angle selection and curvature-weighted error function is employed for fine registration,optimizing local stitching accuracy.Also,Gaussian rendering optimization and bandwidth adjustment functions based on K-nearest neighbors enhance the stitching effect.Experimental results show that compared to 2 contrasting methods,this method improves the stitching accuracy of scattered point clouds.
作者
杨泽慧
岳阳
YANG Zehui;YUE Yang(Guangzhou College of Technology and Business,Foshan 528000,China)
出处
《无线互联科技》
2025年第15期19-22,共4页
Wireless Internet Science and Technology
基金
广州工商学院质量工程建设项目,项目名称:基于ADDIE模型的UI设计课程情境式教学改革研究,项目编号:JXGG202409
广州工商学院质量工程建设项目,项目名称:数字化环境设计教学研究,项目编号:JXGG20231071。
关键词
虚拟现实
散乱点云图像
智能化拼接处理
局部拼接优化
实时渲染优化
virtual reality
scattered point cloud image
intelligent splicing processing
local splicing optimization
real-time rendering optimization
