点云中混合像元的探测与剔除算法

摘要

针对地面三维激光扫描仪扫描过程中产生混合像元的问题，通过对混合像元产生原理和相关特性的分析，提出了一种用于探测和剔除混合像元的算法，并且编制相应的计算机程序予以实现。同时以球体作为实验扫描对象，用混合像元剔除算法对扫描过程中产生的混合像元予以剔除，并比较了混合像元剔除前后点云拟合模型与实际物体在尺寸上及拟合精度的差异，从而验证了该算法的有效性和实用性。混合像元探测与剔除算法可以为后期进行的点云配准、特征提取和曲面建模等工作提供更精确的点云。%For 3D terrestrial laser scanner produces mixed pixel problems during the scanning, after the analysis of the principle of mixed pixel and related features, this paper proposes a algorithm for detecting and eliminating the mixed pixel, and programs the cor-responding computer program to be realized.Furthermore the experiment uses a ball as an experimental scanning object and applies the algorithm to remove mixed pixel in dense point cloud.Compared different sizes and fitting accuracy of point cloud fitting model be-fore and after removing the mixed pixel with the actual object, thus verify the effectiveness and practicality of the algorithm which can provide a more accurate point cloud for point cloud post work like registration, feature extraction and surface modeling etc.