An Efficient Feature Point Reconstruction Method for Robotic Visual Servoing Based on Multiple Cameras

Qiang Fu ； Zi-Liang Zheng ； Shu-Yu Zhang

《淡江理工學刊》 23卷1期 (2020/03) Pp. 47-54

https://doi.org/10.6180/jase.202003_23(1).0006

摘要

This paper deals with the problem of reconstructing the three-dimensional (3D) feature points for robotic visual servoing. Based on the epipolar geometry constraint and the ray distance constraint, a 3D feature point reconstruction method is proposed based on multiple cameras. Simulation and real experiments are performed to show the effectiveness of the proposed method (the reconstruction accuracy reaches millimeter level). Compared to traditional feature point reconstruction methods, the proposed method is not only suitable for conventional cameras but also suitable for fish-eye cameras, which expands the application range.

關鍵字

3D Reconstruction ； Point Matching ； Epipolar Geometry ； Ray Distance ； Multiple Cameras