点云上采样方法综述

A Survey of Point Cloud Upsampling

点云是三维视觉领域常用的数据。然而由于现实条件的局限性,传感器通常只能获得稀疏或带有噪声的点云。点云上采样旨在从稀疏点云中生成稠密点云,为后续分类、分割等任务提供高质量的数据,是当前的重要课题之一。基于此,在介绍各种上采样方法的基础上,着重对基于深度学习实现点云上采样的经典算法进行总结,根据点云上采样的难点和面临的问题,对点云上采样技术未来的发展趋势进行展望。

Point cloud is a common data in the field of three-dimensional vision.However,due to the limitations of realistic conditions,sensors can only obtain sparse or noisy point clouds.Point cloud sampling aims to generate dense point clouds from sparse point clouds and provide high-quality data for subsequent classification,segmentation and other tasks,which is one of the important topics at present.On the basis of the introduction of various up-sampling methods,this paper focuses on the deep learning based on the implementation of point cloud sampling classical algorithms are summarized,according to the difficulties and problems of point cloud sampling,the future development trend of point cloud sampling technology is forecast.