Robust statistical approaches for circle fitting in laser scanning three-dimensional point cloud data
作者：
Highlights：
• Two robust circle fitting algorithms are proposed in point cloud data.
• The new methods fit robust circle in the presence noise, and high percentage of scattered and clustered outliers.
• The proposed methods fit and reconstruct circles for partial as well as full arc data.
• They are more accurate and robust than existing robust statistical methods like LTS and LMS, pattern recognition technique: LTSD, and computer vision techniques like RANSAC.
• The algorithms potential include building information modeling, product quality control, arboreal assessment and road asset monitoring.
摘要
•Two robust circle fitting algorithms are proposed in point cloud data.•The new methods fit robust circle in the presence noise, and high percentage of scattered and clustered outliers.•The proposed methods fit and reconstruct circles for partial as well as full arc data.•They are more accurate and robust than existing robust statistical methods like LTS and LMS, pattern recognition technique: LTSD, and computer vision techniques like RANSAC.•The algorithms potential include building information modeling, product quality control, arboreal assessment and road asset monitoring.
论文关键词：3D modeling,Feature extraction,Object detection,Point cloud processing,Remote sensing,Robust statistics,Surface fitting
论文评审过程：Received 1 September 2017, Revised 1 February 2018, Accepted 8 April 2018, Available online 10 April 2018, Version of Record 22 April 2018.
论文官网地址：https://doi.org/10.1016/j.patcog.2018.04.010
