Premium
Localization and measurement method of continuous casting slab model based on binocular vision
Author(s) -
Fang Jianzhong,
Xu Sixiang,
Yang Yu,
Wang Yang
Publication year - 2020
Publication title -
microwave and optical technology letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.304
H-Index - 76
eISSN - 1098-2760
pISSN - 0895-2477
DOI - 10.1002/mop.32017
Subject(s) - subpixel rendering , artificial intelligence , intersection (aeronautics) , slab , computer vision , zernike polynomials , feature (linguistics) , computer science , autoencoder , distortion (music) , machine vision , mathematics , optics , engineering , pixel , artificial neural network , physics , amplifier , computer network , linguistics , philosophy , structural engineering , bandwidth (computing) , wavefront , aerospace engineering
As the corner points of continuous casting slab model are not obvious, the traditional feature points detection methods cannot detect the effective feature points, so these methods cannot complete the localization and measurement of continuous casting slab model. A binocular vision localization and measurement method based on edges fitting was proposed. The acquired photos of continuous casting slab model were preprocessed including de‐distortion and stereo correction. And then the Canny operator and Zernike moment were used to detect the subpixel points of the model edges. These subpixel points that have been obtained were clustered and fitted, and the intersection points of fitting curves were calculated. Finally, using the fundamental matrix to match the intersection points, and the parallax information was obtained to locate the slab model. The experiment shows that the average measurement error of model is 1.31%, which can meet the precision requirements of production.