Robot-based 3-D machine vision using circular-feature normal estimation and multi-dimensional image fusion
Journal
IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM
Journal Volume
2019-July
Pages
248-253
Date Issued
2019
Author(s)
Abstract
Reconstruction of 3-D images from 2-D images is an essential problem in vision systems. Circular features represent one of the most common quadratic-curved features, which have been presented for 3-D reconstruction such as machined features in aerospace and electronics industries. The circular features may deviate from their nominal geometric properties due to potential machining errors. In this paper, a robot-based 3-D machine vision method is developed to perform the 3-D reconstruction of machined circular features with the goal of automatically inspecting and reconstructing 3-D objects using machine vision. An innovative algorithm is developed for estimating the normal vector of a circular machined feature on a free-form surface and reconstructing a 3-D surface profile with the machined circular feature using multi-dimensional data fusion. The surface between the projected 2-D circular feature and the identified 3-D surface profilometry is reconstructed via extrapolation by NURBS surface fitting. Some experimental results were successfully demonstrated that the proposed method is both accurate and effective.
SDGs
Type
conference paper