JZUS - Journal of Zhejiang University SCIENCE

Frontiers of Information Technology & Electronic Engineering 2015 Vol.16 No.2 P.152-165

Design of an enhanced visual odometry by building and matching compressive panoramic landmarks online

Author(s): Wei Lu, Zhi-yu Xiang, Ji-lin Liu
Affiliation(s): Institute of Information and Communication Engineering, Zhejiang University, Hangzhou 310027, China; more
Corresponding email(s): lwhfh01@zju.edu.cn, xiangzy@zju.edu.cn
Key Words: Visual odometry, Panoramic landmark, Landmark matching, Compressed sensing, Adaptive compressive feature

Share this article to： More <<< Previous Article \|Next Article >>>

Abstract: Efficient and precise localization is a prerequisite for the intelligent navigation of mobile robots. Traditional visual localization systems, such as visual odometry (VO) and simultaneous localization and mapping (SLAM), suffer from two shortcomings: a drift problem caused by accumulated localization error, and erroneous motion estimation due to illumination variation and moving objects. In this paper, we propose an enhanced VO by introducing a panoramic camera into the traditional stereo-only VO system. Benefiting from the 360° field of view, the panoramic camera is responsible for three tasks: (1) detecting road junctions and building a landmark library online; (2) correcting the robot’s position when the landmarks are revisited with any orientation; (3) working as a panoramic compass when the stereo VO cannot provide reliable positioning results. To use the large-sized panoramic images efficiently, the concept of compressed sensing is introduced into the solution and an adaptive compressive feature is presented. Combined with our previous two-stage local binocular bundle adjustment (TLBBA) stereo VO, the new system can obtain reliable positioning results in quasi-real time. Experimental results of challenging long-range tests show that our enhanced VO is much more accurate and robust than the traditional VO, thanks to the compressive panoramic landmarks built online.

The paper presents a method for enhancing the visual odometry using two kinds of vision sensors, like stereo cameras and panoramic (360 degree-omnidirectional) camera. The proposed method aims to reduce the error of motion estimation based on matching the landmarks in panoramic images with database in the library as the robot is revisiting the scenes, which focuses on special scenes such as road junctions and buildings. Authors utilize the panoramic camera for estimating the azimuthal rotation of robot when the stereo VO (visual odometry) could not sufficiently provide reliable position estimation results. The experimental results demonstrate the efficiency and effectiveness of the proposed method under varied conditions. It would be a good idea to compare the result with other landmark matching localization schemes. This comparison would be more useful if the authors are able to quantify the benefit (may be in terms of frequency and accuracy) of the proposed work. The repeatability of the proposed scheme should also be commented. In general, the paper is fairly well written. Overall quality of the research is acceptable.

基于在线建立与匹配压缩全景路标的增强型视觉里程计

目的：高效精确定位是移动机器人智能导航的先决条件。传统视觉定位系统，如视觉里程计（VO）和同时定位与三维重建（SLAM）算法，存在两点不足：一是由累积定位误差引起的漂移问题，二是由光照变化和移动物体导致的错误运动估计结果。
创新：通过引入全景相机到传统双目VO系统，提出一种增强型VO，高效利用全景相机360˚视场角信息。（1）在线建立路口场景压缩全景路标库；（2）机器人以任意方向重新访问路标时，对定位结果进行全局校正；（3）当双目立体VO不能提供可靠定位信息时对航向角估计结果进行校正；（4）为高效利用信息量较多的全景图像，引入压缩感知概念并提出一种自适应压缩特征。
方法：首先，在压缩亮度特征基础上，增加压缩SURF特征提高其描述能力，通过分析特征区分度，使压缩特征可以根据具体图像特点自适应调节，最终构建自适应压缩特征（ACF,图2），该特征计算速度快（表3）、描述能力强（图6、7，表1），有效提高全景图像信息利用效率。然后，使用ACF对全景路标图像进行描述，提出一种任意方向的路标图像匹配算法，若当前全景图像与路标图像匹配成功，则对当前定位结果进行全局位姿校正（图4），抑制大范围环境中定位路径漂移问题（图10、11）。最后，介绍基于图像片匹配的航向角鲁棒估计方法，当双目视觉里程计因特征跟踪质量差而导致运动估计结果不稳定时，对局部运动估计结果进行校正，提高运动估计的精度（图9）。
结论：提出的增强型视觉里程计系统可以准实时提供可靠定位结果，极大抑制大范围挑战性环境中传统VO漂移问题和运动估计错误问题。实验结果显示，所提算法大幅度提高传统VO的准确性和鲁棒性。

关键词：视觉里程计；全景路标；路标匹配；压缩感知；自适应压缩特征

Reference

[1]Bay, H., Tuytelaars, T., van Gool, L., 2006. SURF: speeded up robust features. Proc. 9th European Conf. on Computer Vision, p.404-417.

[2]Cai, X., Zhang, Z., Zhang, H., et al., 2014. Soft consistency reconstruction: a robust 1-bit compressive sensing algorithm. arXiv:1402.5475 (preprint).

[3]Candes, E.J., Tao, T., 2005. Decoding by linear programming. IEEE Trans. Inform. Theory, 51(12):4203-4215.

[4]Donoho, D.L., 2006. Compressed sensing. IEEE Trans. Inform. Theory, 52(4):1289-1306.

[5]Durrant-Whyte, H., Bailey, T., 2006. Simultaneous localization and mapping: part I. IEEE Robot. Autom. Mag., 13(2):99-110.

[6]Fischler, M.A., Bolles, R.C., 1981. Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography. Commun. ACM, 24(6):381-395.

[7]Fraundorfer, F., Scaramuzza, D., 2012. Visual odometry: part II. Matching, robustness, optimization, and applications. IEEE Robot. Autom. Mag., 19(2):78-90.

[8]Galvez-López, D., Tardos, J.D., 2012. Bags of binary words for fast place recognition in image sequences. IEEE Trans. Robot., 28(5):1188-1197.

[9]Geiger, A., Lenz, P., Urtasun, R., 2012. Are we ready for autonomous driving? The KITTI vision benchmark suite. Proc. IEEE Conf. on Computer Vision and Pattern Recognition, p.3354-3361.

[10]Horn, B.K.P., 1987. Closed-form solution of absolute orientation using unit quaternions. JOSA A, 4(4):629-642.

[11]Konolige, K., Agrawal, M., Solà, J., 2011. Large-scale visual odometry for rough terrain. Proc. 13th Int. Symp. on Robotics Research, p.201-212.

[12]Liu, Y., Zhang, H., 2012. Visual loop closure detection with a compact image descriptor. Proc. IEEE/RSJ Int. Conf. on Intelligent Robots and Systems, p.1051-1056.

[13]Lu, W., Xiang, Z., Liu, J., 2013. High-performance visual odometry with two-stage local binocular BA and GPU. Proc. IEEE Intelligent Vehicles Symp., p.1107-1112.

[14]Munguia, R., Grau, A., 2007. Monocular SLAM for visual odometry. Proc. IEEE Int. Symp. on Intelligent Signal Processing, p.1-6.

[15]Nistér, D., Naroditsky, O., Bergen, J., 2004. Visual odometry. Proc. IEEE Computer Society Conf. on Computer Vision and Pattern Recognition, p.652-659.

[16]Scaramuzza, D., Fraundorfer, F., 2011. Visual odometry (tutorial). IEEE Robot. Autom. Mag., 18(4):80-92.

[17]Se, S., Lowe, D., Little, J., 2002. Global localization using distinctive visual features. Proc. IEEE/RSJ Int. Conf. on Intelligent Robots and Systems, p.226-231.

[18]Singh, G., Košechá, J., 2010. Visual loop closing using gist descriptors in Manhattan world. ICRA Omnidirectional Vision Workshop.

[19]Sivic, J., Zisserman, A., 2003. Video Google: a text retrieval approach to object matching in videos. Proc. 9th IEEE Int. Conf. on Computer Vision, p.1470-1477.

[20]Stewénius, H., Engels, C., Nistér, D., 2006. Recent developments on direct relative orientation. ISPRS J. Photogr. Remote Sens., 60(4):284-294.

[21]Sünderhauf, N., Protzel, P., 2011. BRIEF-Gist—closing the loop by simple means. Proc. IEEE/RSJ Int. Conf. on Intelligent Robots and Systems, p.1234-1241.

[22]Wang, Y., 2013. Navigational Road Modeling Based on Omnidirectional Multi-camera System. PhD Thesis, Zhejiang University, Hangzhou, China (in Chinese).

[23]Wright, J., Yang, A.Y., Ganesh, A., et al., 2009. Robust face recognition via sparse representation. IEEE Trans. Patt. Anal. Mach. Intell., 31(2):210-227.

[24]Wu, C., 2007. SiftGPU: a GPU Implementation of Scale Invariant Feature Transform (SIFT). Available from http://cs.unc.edu/~ccwu/siftgpu/.

[25]Zhang, K., Zhang, L., Yang, M.H., 2012. Real-time compressive tracking. Proc. 12th European Conf. on Computer Vision, p.864-877.

Open peer comments: Debate/Discuss/Question/Opinion

<1>

Similar articles

- Go to

基于在线建立与匹配压缩全景路标的增强型视觉里程计

Darkslateblue:Affiliate; Royal Blue:Author; Turquoise:Article

Reference