Heterogeneous Implementation of a Novel Indirect Visual Odometry System

Cheng Hung Lin, Wei Yen Wang, Shen Ho Liu, Chen-Chien James Hsu, Chiang Heng Chien

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

In this paper, a novel parallel indirect visual odometry (VO) system is proposed based on a newly designed map management method, key-frame selection, and a camera pose correction model, where the speeded-up robust features (SURF) algorithm is used to extract features from an image, and a linear exhaustive search (LES) algorithm is introduced to match features. To minimize computation time, a key-frame selection mechanism is proposed to distinguish key frames among the input images. Moreover, map management is proposed to filter out unstable landmarks and add features for a reliable estimation of the relative camera pose. In addition, estimation accuracy is improved by the camera pose correction model. To enhance computational efficiency of the VO system, the proposed approach is implemented on a TX2 embedded system with multiple graphics processing units (GPUs), taking advantage of a heterogeneous parallel computing architecture. To validate the performances of the proposed system, several experiments are conducted using an ASUS Xtion 3D camera and a laptop. Average errors of pose estimations are compared with those via the conventional VO to show the effectiveness of the proposed VO system. Thanks to the proposed approach, a real-time and reliable VO system can be established with low cost, low power consumption, high processing efficiency, and portability. The experimental results show that based on heterogeneous computing, the required computation time of the overall system with GPUs is approximately 80 times faster than that with only a CPU, when dealing with 80 features in the environment. To the best of our knowledge, this is the first released paper that implements an indirect VO on a CPU/GPU heterogeneous computing platform.

Original languageEnglish
Article number8666020
Pages (from-to)34631-34644
Number of pages14
JournalIEEE Access
Volume7
DOIs
Publication statusPublished - 2019 Jan 1

Fingerprint

Cameras
Program processors
Parallel processing systems
Computational efficiency
Embedded systems
Electric power utilization
Processing
Graphics processing unit
Costs
Experiments

Keywords

  • GPU
  • SURF
  • Visual odometry
  • heterogeneous computing
  • perspective-three-points (P3P)

ASJC Scopus subject areas

  • Computer Science(all)
  • Materials Science(all)
  • Engineering(all)

Cite this

Heterogeneous Implementation of a Novel Indirect Visual Odometry System. / Lin, Cheng Hung; Wang, Wei Yen; Liu, Shen Ho; Hsu, Chen-Chien James; Chien, Chiang Heng.

In: IEEE Access, Vol. 7, 8666020, 01.01.2019, p. 34631-34644.

Research output: Contribution to journalArticle

Lin, Cheng Hung ; Wang, Wei Yen ; Liu, Shen Ho ; Hsu, Chen-Chien James ; Chien, Chiang Heng. / Heterogeneous Implementation of a Novel Indirect Visual Odometry System. In: IEEE Access. 2019 ; Vol. 7. pp. 34631-34644.
@article{93d0d89ff978408887023353cce9877c,
title = "Heterogeneous Implementation of a Novel Indirect Visual Odometry System",
abstract = "In this paper, a novel parallel indirect visual odometry (VO) system is proposed based on a newly designed map management method, key-frame selection, and a camera pose correction model, where the speeded-up robust features (SURF) algorithm is used to extract features from an image, and a linear exhaustive search (LES) algorithm is introduced to match features. To minimize computation time, a key-frame selection mechanism is proposed to distinguish key frames among the input images. Moreover, map management is proposed to filter out unstable landmarks and add features for a reliable estimation of the relative camera pose. In addition, estimation accuracy is improved by the camera pose correction model. To enhance computational efficiency of the VO system, the proposed approach is implemented on a TX2 embedded system with multiple graphics processing units (GPUs), taking advantage of a heterogeneous parallel computing architecture. To validate the performances of the proposed system, several experiments are conducted using an ASUS Xtion 3D camera and a laptop. Average errors of pose estimations are compared with those via the conventional VO to show the effectiveness of the proposed VO system. Thanks to the proposed approach, a real-time and reliable VO system can be established with low cost, low power consumption, high processing efficiency, and portability. The experimental results show that based on heterogeneous computing, the required computation time of the overall system with GPUs is approximately 80 times faster than that with only a CPU, when dealing with 80 features in the environment. To the best of our knowledge, this is the first released paper that implements an indirect VO on a CPU/GPU heterogeneous computing platform.",
keywords = "GPU, SURF, Visual odometry, heterogeneous computing, perspective-three-points (P3P)",
author = "Lin, {Cheng Hung} and Wang, {Wei Yen} and Liu, {Shen Ho} and Hsu, {Chen-Chien James} and Chien, {Chiang Heng}",
year = "2019",
month = "1",
day = "1",
doi = "10.1109/ACCESS.2019.2904142",
language = "English",
volume = "7",
pages = "34631--34644",
journal = "IEEE Access",
issn = "2169-3536",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - JOUR

T1 - Heterogeneous Implementation of a Novel Indirect Visual Odometry System

AU - Lin, Cheng Hung

AU - Wang, Wei Yen

AU - Liu, Shen Ho

AU - Hsu, Chen-Chien James

AU - Chien, Chiang Heng

PY - 2019/1/1

Y1 - 2019/1/1

N2 - In this paper, a novel parallel indirect visual odometry (VO) system is proposed based on a newly designed map management method, key-frame selection, and a camera pose correction model, where the speeded-up robust features (SURF) algorithm is used to extract features from an image, and a linear exhaustive search (LES) algorithm is introduced to match features. To minimize computation time, a key-frame selection mechanism is proposed to distinguish key frames among the input images. Moreover, map management is proposed to filter out unstable landmarks and add features for a reliable estimation of the relative camera pose. In addition, estimation accuracy is improved by the camera pose correction model. To enhance computational efficiency of the VO system, the proposed approach is implemented on a TX2 embedded system with multiple graphics processing units (GPUs), taking advantage of a heterogeneous parallel computing architecture. To validate the performances of the proposed system, several experiments are conducted using an ASUS Xtion 3D camera and a laptop. Average errors of pose estimations are compared with those via the conventional VO to show the effectiveness of the proposed VO system. Thanks to the proposed approach, a real-time and reliable VO system can be established with low cost, low power consumption, high processing efficiency, and portability. The experimental results show that based on heterogeneous computing, the required computation time of the overall system with GPUs is approximately 80 times faster than that with only a CPU, when dealing with 80 features in the environment. To the best of our knowledge, this is the first released paper that implements an indirect VO on a CPU/GPU heterogeneous computing platform.

AB - In this paper, a novel parallel indirect visual odometry (VO) system is proposed based on a newly designed map management method, key-frame selection, and a camera pose correction model, where the speeded-up robust features (SURF) algorithm is used to extract features from an image, and a linear exhaustive search (LES) algorithm is introduced to match features. To minimize computation time, a key-frame selection mechanism is proposed to distinguish key frames among the input images. Moreover, map management is proposed to filter out unstable landmarks and add features for a reliable estimation of the relative camera pose. In addition, estimation accuracy is improved by the camera pose correction model. To enhance computational efficiency of the VO system, the proposed approach is implemented on a TX2 embedded system with multiple graphics processing units (GPUs), taking advantage of a heterogeneous parallel computing architecture. To validate the performances of the proposed system, several experiments are conducted using an ASUS Xtion 3D camera and a laptop. Average errors of pose estimations are compared with those via the conventional VO to show the effectiveness of the proposed VO system. Thanks to the proposed approach, a real-time and reliable VO system can be established with low cost, low power consumption, high processing efficiency, and portability. The experimental results show that based on heterogeneous computing, the required computation time of the overall system with GPUs is approximately 80 times faster than that with only a CPU, when dealing with 80 features in the environment. To the best of our knowledge, this is the first released paper that implements an indirect VO on a CPU/GPU heterogeneous computing platform.

KW - GPU

KW - SURF

KW - Visual odometry

KW - heterogeneous computing

KW - perspective-three-points (P3P)

UR - http://www.scopus.com/inward/record.url?scp=85063772639&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85063772639&partnerID=8YFLogxK

U2 - 10.1109/ACCESS.2019.2904142

DO - 10.1109/ACCESS.2019.2904142

M3 - Article

AN - SCOPUS:85063772639

VL - 7

SP - 34631

EP - 34644

JO - IEEE Access

JF - IEEE Access

SN - 2169-3536

M1 - 8666020

ER -