Proposing Camera Calibration Method Using PPO (Proximal Policy Optimization) for Improving Camera Pose Estimations

Haitham Al-Jabri, Takafumi Matsumaru

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This paper highlights camera orientation estimation accuracy and precision, as well as proposing a new camera calibration technique using a reinforcement learning method named PPO (Proximal Policy Optimization) in offline mode. The offline mode is used just for extracting the camera geometry parameters that are used for improving accuracy in real-time camera pose estimation techniques. We experiment and compare two popular techniques using 2D vision feedbacks and evaluate their accuracy beside other considerations related to real applications such as disturbance cases from surrounding environment and pose data stability. First, we use feature points detection ORB (Oriented FAST and Rotated BRIEF) and BF (Brute-Force) matcher to detect and match points in different frames, respectively. Second, we use FAST (Features from Accelerated Segment Test) corners and LK (Lucas-Kanade) optical flow methods to detect corners and track their flow in different frames. Those points and corners are then used for the pose estimation through optimization process with the: (a) calibration method of Zhang using chessboard pattern and (b) our proposed method using PPO. The results using our proposed calibration method show significant accuracy improvements and easier deployment for end-user compared to the pre-used methods.

Original languageEnglish
Title of host publication2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages790-795
Number of pages6
ISBN (Electronic)9781728103761
DOIs
Publication statusPublished - 2019 Mar 11
Event2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018 - Kuala Lumpur, Malaysia
Duration: 2018 Dec 122018 Dec 15

Publication series

Name2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018

Conference

Conference2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018
CountryMalaysia
CityKuala Lumpur
Period18/12/1218/12/15

Fingerprint

Calibration
Cameras
Optical flows
Reinforcement learning
Feedback
Geometry
Learning
Experiments

ASJC Scopus subject areas

  • Biotechnology
  • Artificial Intelligence
  • Human-Computer Interaction

Cite this

Al-Jabri, H., & Matsumaru, T. (2019). Proposing Camera Calibration Method Using PPO (Proximal Policy Optimization) for Improving Camera Pose Estimations. In 2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018 (pp. 790-795). [8665088] (2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ROBIO.2018.8665088

Proposing Camera Calibration Method Using PPO (Proximal Policy Optimization) for Improving Camera Pose Estimations. / Al-Jabri, Haitham; Matsumaru, Takafumi.

2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018. Institute of Electrical and Electronics Engineers Inc., 2019. p. 790-795 8665088 (2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Al-Jabri, H & Matsumaru, T 2019, Proposing Camera Calibration Method Using PPO (Proximal Policy Optimization) for Improving Camera Pose Estimations. in 2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018., 8665088, 2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018, Institute of Electrical and Electronics Engineers Inc., pp. 790-795, 2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018, Kuala Lumpur, Malaysia, 18/12/12. https://doi.org/10.1109/ROBIO.2018.8665088
Al-Jabri H, Matsumaru T. Proposing Camera Calibration Method Using PPO (Proximal Policy Optimization) for Improving Camera Pose Estimations. In 2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018. Institute of Electrical and Electronics Engineers Inc. 2019. p. 790-795. 8665088. (2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018). https://doi.org/10.1109/ROBIO.2018.8665088
Al-Jabri, Haitham ; Matsumaru, Takafumi. / Proposing Camera Calibration Method Using PPO (Proximal Policy Optimization) for Improving Camera Pose Estimations. 2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 790-795 (2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018).
@inproceedings{15fa59b778814d78bf9704aa1eef6e5b,
title = "Proposing Camera Calibration Method Using PPO (Proximal Policy Optimization) for Improving Camera Pose Estimations",
abstract = "This paper highlights camera orientation estimation accuracy and precision, as well as proposing a new camera calibration technique using a reinforcement learning method named PPO (Proximal Policy Optimization) in offline mode. The offline mode is used just for extracting the camera geometry parameters that are used for improving accuracy in real-time camera pose estimation techniques. We experiment and compare two popular techniques using 2D vision feedbacks and evaluate their accuracy beside other considerations related to real applications such as disturbance cases from surrounding environment and pose data stability. First, we use feature points detection ORB (Oriented FAST and Rotated BRIEF) and BF (Brute-Force) matcher to detect and match points in different frames, respectively. Second, we use FAST (Features from Accelerated Segment Test) corners and LK (Lucas-Kanade) optical flow methods to detect corners and track their flow in different frames. Those points and corners are then used for the pose estimation through optimization process with the: (a) calibration method of Zhang using chessboard pattern and (b) our proposed method using PPO. The results using our proposed calibration method show significant accuracy improvements and easier deployment for end-user compared to the pre-used methods.",
author = "Haitham Al-Jabri and Takafumi Matsumaru",
year = "2019",
month = "3",
day = "11",
doi = "10.1109/ROBIO.2018.8665088",
language = "English",
series = "2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "790--795",
booktitle = "2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018",

}

TY - GEN

T1 - Proposing Camera Calibration Method Using PPO (Proximal Policy Optimization) for Improving Camera Pose Estimations

AU - Al-Jabri, Haitham

AU - Matsumaru, Takafumi

PY - 2019/3/11

Y1 - 2019/3/11

N2 - This paper highlights camera orientation estimation accuracy and precision, as well as proposing a new camera calibration technique using a reinforcement learning method named PPO (Proximal Policy Optimization) in offline mode. The offline mode is used just for extracting the camera geometry parameters that are used for improving accuracy in real-time camera pose estimation techniques. We experiment and compare two popular techniques using 2D vision feedbacks and evaluate their accuracy beside other considerations related to real applications such as disturbance cases from surrounding environment and pose data stability. First, we use feature points detection ORB (Oriented FAST and Rotated BRIEF) and BF (Brute-Force) matcher to detect and match points in different frames, respectively. Second, we use FAST (Features from Accelerated Segment Test) corners and LK (Lucas-Kanade) optical flow methods to detect corners and track their flow in different frames. Those points and corners are then used for the pose estimation through optimization process with the: (a) calibration method of Zhang using chessboard pattern and (b) our proposed method using PPO. The results using our proposed calibration method show significant accuracy improvements and easier deployment for end-user compared to the pre-used methods.

AB - This paper highlights camera orientation estimation accuracy and precision, as well as proposing a new camera calibration technique using a reinforcement learning method named PPO (Proximal Policy Optimization) in offline mode. The offline mode is used just for extracting the camera geometry parameters that are used for improving accuracy in real-time camera pose estimation techniques. We experiment and compare two popular techniques using 2D vision feedbacks and evaluate their accuracy beside other considerations related to real applications such as disturbance cases from surrounding environment and pose data stability. First, we use feature points detection ORB (Oriented FAST and Rotated BRIEF) and BF (Brute-Force) matcher to detect and match points in different frames, respectively. Second, we use FAST (Features from Accelerated Segment Test) corners and LK (Lucas-Kanade) optical flow methods to detect corners and track their flow in different frames. Those points and corners are then used for the pose estimation through optimization process with the: (a) calibration method of Zhang using chessboard pattern and (b) our proposed method using PPO. The results using our proposed calibration method show significant accuracy improvements and easier deployment for end-user compared to the pre-used methods.

UR - http://www.scopus.com/inward/record.url?scp=85064120902&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85064120902&partnerID=8YFLogxK

U2 - 10.1109/ROBIO.2018.8665088

DO - 10.1109/ROBIO.2018.8665088

M3 - Conference contribution

AN - SCOPUS:85064120902

T3 - 2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018

SP - 790

EP - 795

BT - 2018 IEEE International Conference on Robotics and Biomimetics, ROBIO 2018

PB - Institute of Electrical and Electronics Engineers Inc.

ER -