Edge–Point Cloud Fusion for Geometric Fitting of Cylinder Parameters Using Single-View RGB-D Data

Huayan Zhang, Jiaxin Liu, Zhongkui Wang

Sensors, 2026

We propose an edge–point cloud fusion method for cylinder fitting from single-view RGB-D data. By incorporating 2D image-edge constraints with 3D point cloud information in a unified optimization framework, our method reduces the effect of point cloud noise and improves fitting accuracy and robustness.

Cylinder Fitting Using Orthonormal Representation in Noisy Point Clouds

Huayan Zhang, Yang Tian, Shugen Ma, Zhongkui Wang

IEEE Transactions on Instrumentation and Measurement, 2025

We propose a least-squares cylinder fitting method for noisy point clouds. By reformulating the fitting problem with an orthonormal representation and adopting a coarse-to-fine joint optimization strategy, our method improves fitting accuracy and robustness under noisy conditions significantly.