A stereo visual odometry based on SURF feature and three consecutive frames

Visual Odometry is the process of estimating 6 DOF motion of a vehicle equipped with a single or multiple cameras. This technique has many potential applications, such as autonomous navigation and driving assistant in intelligent transportation field, which is an important part of smart city. In this paper, the basic framework of stereo visual odometry is reviewed, and a new 3D position estimation method in three consecutive frames is proposed. The main contributions of this paper include: (1) SURF of stereo image sequences is used in the process of detecting and matching features; (2) by saving historical matching result, feature matching is performed on three consecutive frames instead of just two frames without additional computation; (3) in the 3D-to-2D motion estimating step, 3D position is estimated from three consecutive frames instead of just two frames, thereby we can obtain more accurate results. We apply our method on the KITTI datasets, and the results show an accurate trajectory estimation over several hundred meters.

[1]  Gordon Wyeth,et al.  Single camera vision-only SLAM on a suburban road network , 2008, 2008 IEEE International Conference on Robotics and Automation.

[2]  Larry H. Matthies,et al.  Two years of Visual Odometry on the Mars Exploration Rovers , 2007, J. Field Robotics.

[3]  Philippe Bonnifait,et al.  An experiment of a 3D real-time robust visual odometry for intelligent vehicles , 2009, 2009 12th International IEEE Conference on Intelligent Transportation Systems.

[4]  Zhengyou Zhang,et al.  A Flexible New Technique for Camera Calibration , 2000, IEEE Trans. Pattern Anal. Mach. Intell..

[5]  Christopher G. Harris,et al.  3D positional integration from image sequences , 1988, Image Vis. Comput..

[6]  Julius Ziegler,et al.  StereoScan: Dense 3d reconstruction in real-time , 2011, 2011 IEEE Intelligent Vehicles Symposium (IV).

[7]  Andrew W. Fitzgibbon,et al.  Bundle Adjustment - A Modern Synthesis , 1999, Workshop on Vision Algorithms.

[8]  Larry Matthies,et al.  Stereo vision and rover navigation software for planetary exploration , 2002, Proceedings, IEEE Aerospace Conference.

[9]  Massimo Bertozzi,et al.  Real-time obstacle detection using stereo vision for autonomous ground vehicles: A survey , 2014, 17th International IEEE Conference on Intelligent Transportation Systems (ITSC).

[10]  Richard I. Hartley,et al.  Theory and Practice of Projective Rectification , 1999, International Journal of Computer Vision.

[11]  Tom Drummond,et al.  Machine Learning for High-Speed Corner Detection , 2006, ECCV.

[12]  R. Goecke,et al.  Visual Vehicle Egomotion Estimation using the Fourier-Mellin Transform , 2007, 2007 IEEE Intelligent Vehicles Symposium.

[13]  Friedrich Fraundorfer,et al.  Visual Odometry Part I: The First 30 Years and Fundamentals , 2022 .

[14]  Carlo Tomasi,et al.  Good features to track , 1994, 1994 Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.

[15]  Paul Newman,et al.  Continually improving large scale long term visual navigation of a vehicle in dynamic urban environments , 2012, 2012 15th International IEEE Conference on Intelligent Transportation Systems.

[16]  Mahmoud Belhocine,et al.  SIFT and SURF Performance Evaluation for Mobile Robot-Monocular Visual Odometry , 2014 .

[17]  Sergiu Nedevschi,et al.  Fast vision based ego-motion estimation from stereo sequences — A GPU approach , 2011, 2011 14th International IEEE Conference on Intelligent Transportation Systems (ITSC).

[18]  David G. Lowe,et al.  Distinctive Image Features from Scale-Invariant Keypoints , 2004, International Journal of Computer Vision.

[19]  Larry H. Matthies,et al.  Visual odometry on the Mars Exploration Rovers , 2005, 2005 IEEE International Conference on Systems, Man and Cybernetics.

[20]  Thomas S. Huang,et al.  Motion and structure from feature correspondences: a review , 1994, Proc. IEEE.

[21]  Luc Van Gool,et al.  SURF: Speeded Up Robust Features , 2006, ECCV.