TY - CHAP
T1 - A Linear Approach to Absolute Pose Estimation for Light Fields
AU - Nousias, Sotiris
AU - Lourakis, Manolis
AU - Keane, Pearse
AU - Ourselin, Sebastien
AU - Bergeles, Christos
N1 - Funding Information:
Acknowledgements: This work was supported by an EPSRC Centre for Doctoral Training in Medical Imaging [EP/L016478/1], an AMS Springboard Award [SBF001/1002], an ERC Starting Grant [714562], and EU’s H2020 Programme [GA No 826506 sustAGE].
Publisher Copyright:
© 2020 IEEE.
Copyright:
Copyright 2021 Elsevier B.V., All rights reserved.
PY - 2020/11
Y1 - 2020/11
N2 - This paper presents the first absolute pose estimation approach tailored to Light Field cameras. It builds on the observation that the ratio between the disparity arising in different sub-aperture images and their corresponding baseline is constant. Hence, we augment the 2D pixel coordinates with the corresponding normalised disparity to obtain the Light Field feature. This new representation reduces the effect of noise by aggregating multiple projections and allows for linear estimation of the absolute pose of a Light Field camera using the well-known Direct Linear Transformation algorithm. We evaluate the resulting absolute pose estimates with extensive simulations and experiments involving real Light Field datasets, demonstrating the competitive performance of our linear approach. Furthermore, we integrate our approach in a state-of-the-art Light Field Structure from Motion pipeline and demonstrate accurate multi-view 3D reconstruction.
AB - This paper presents the first absolute pose estimation approach tailored to Light Field cameras. It builds on the observation that the ratio between the disparity arising in different sub-aperture images and their corresponding baseline is constant. Hence, we augment the 2D pixel coordinates with the corresponding normalised disparity to obtain the Light Field feature. This new representation reduces the effect of noise by aggregating multiple projections and allows for linear estimation of the absolute pose of a Light Field camera using the well-known Direct Linear Transformation algorithm. We evaluate the resulting absolute pose estimates with extensive simulations and experiments involving real Light Field datasets, demonstrating the competitive performance of our linear approach. Furthermore, we integrate our approach in a state-of-the-art Light Field Structure from Motion pipeline and demonstrate accurate multi-view 3D reconstruction.
UR - http://www.scopus.com/inward/record.url?scp=85101431778&partnerID=8YFLogxK
U2 - 10.1109/3DV50981.2020.00077
DO - 10.1109/3DV50981.2020.00077
M3 - Conference paper
AN - SCOPUS:85101431778
T3 - Proceedings - 2020 International Conference on 3D Vision, 3DV 2020
SP - 672
EP - 681
BT - Proceedings - 2020 International Conference on 3D Vision, 3DV 2020
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 8th International Conference on 3D Vision, 3DV 2020
Y2 - 25 November 2020 through 28 November 2020
ER -