Open AccessCalibration of AOTF-based 3D measurement system using multiplane model based on phase fringe and BP neural network
Author(s) -
Huijun Zhao,
Shaoguang Shi,
Hongzhi Jiang,
Ying Zhang,
Zefu Xu
Publication year - 2017
Publication title -
optics express
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.394
H-Index - 271
ISSN - 1094-4087
DOI - 10.1364/oe.25.010413
Subject(s) - pinhole (optics) , calibration , subpixel rendering , computer science , optics , distortion (music) , artificial neural network , projection (relational algebra) , hyperspectral imaging , camera resectioning , phase (matter) , artificial intelligence , pinhole camera model , structured light 3d scanner , lens (geology) , process (computing) , computer vision , pixel , camera auto calibration , scanner , algorithm , physics , amplifier , computer network , bandwidth (computing) , quantum mechanics , operating system
A specifically designed imaging system based on an acousto-optic tunable filter (AOTF) can integrate hyperspectral imaging and 3D reconstruction. As a result of the complicated optical structure, the AOTF imaging system deviates from the traditional pinhole model and lens distortion form, causing difficulty to achieve precise camera calibration. The influencing factors leading to the deviation are discussed and a multiplane model (MPM) is proposed with phase fringe to produce dense mark points and a back propagation neural network to obtain subpixel calibration. Experiments show that MPM can reduce the back projection error efficiently compared with the pinhole model. A 3D reconstruction process is conducted based on the calibration result to verify the feasibility of the proposed method.