双目视觉具有近距离测距精度高、获取的图像内容丰富等优点，是目前无人船自主避障领域的研究热点，立体匹配是双目视觉测距的技术关键。然而，无人船体的不规则晃动，增大了双目图像间海面障碍物的尺度、亮度差异，易造成匹配歧义。为提高立体匹配的准确性，本文提出一种改进的AKAZE特征匹配方法。通过提取、匹配非线性尺度空间中障碍物的轮廓角点，从而可准确测得海面障碍物的位置信息。利用本文所提方法研发出一种具有防水、防盐等性能的适用于无人船海上工作环境的宽基线双目视觉系统SV-100。测试实验结果表明，该系统在100 m内的测距误差小于6 m。
Stereo vision has the advantages of high ranging accuracy in short-range, rich content of obtained images. Therefore, it is a research hotspot in the field of obstacle avoidance for unmanned surface vehicles (USVs). Stereo matching is the key of stereo vision ranging. However, the irregular shaking of hull increases the scale and brightness difference of obstacles between binocular images, which is easy to cause matching errors. To improve the accuracy of stereo matching, an improved AKAZE features method is proposed in this paper. By extracting and matching contour corner points of obstacles in a nonlinear scale space, the position of the obstacle can be measured accurately. Based on the method proposed in this paper, a wide-baseline stereo vision system named SV-100 with water and salt resistance is developed, which is suitable for USVs working environment. The test experiment of the system was carried out. The results verified that the ranging error of the system within 100 m was less than 6 m.
2019,41(12): 118-122 收稿日期：2019-07-11
DOI：10.3404/j.issn.1672-7649.2019.12.024
分类号：U664.82
基金项目：国家重点研发计划资助项目（2017YFC1405203）；中央高校基本科研业务费专项资助（19CX05003A-1，17CX02079）
作者简介：李方旭(1995-),男,硕士研究生,研究方向为无人船双目视觉海面障碍物检测
参考文献：
[1] 王石, 张建强, 杨舒卉, 等. 国内外无人艇发展现状及典型作战应用研究[J]. 火力与指挥控制, 2019, 44(2): 13–17
[2] 田军, 佘亚军, 蔡龙. 针对持续性跟踪无人艇的探测技术[J]. 舰船科学技术, 2013, 35(4): 98–101
[3] 李小毛, 张鑫, 王文涛, 等. 基于3D激光雷达的无人水面艇海上目标检测[J]. 上海大学学报(自然科学版), 2017, 23(1): 27–36
[4] MA H, SMART E, AHMED A, et al. Radar image-based positioning for USV under GPS denial environment[J]. IEEE Transactions on Intelligent Transportation Systems, 2017: 1–9
[5] LARSON J, BRUCH M, EBKEN J. Autonomous navigation and obstacle avoidance for unmanned surface vehicles[J]. Proceedings of SPIE - The International Society for Optical Engineering, 2006, 6230: 623007–623007-12
[6] SHIN B S, MOU X, MOU W, et al. Vision-based navigation of an unmanned surface vehicle with object detection and tracking abilities[J]. Machine Vision and Applications, 2017, 29: 95–112
[7] SINISTERRA A J, DHANAK M R, VON ELLENRIEDER K. Stereovision-based target tracking system for USV operations[J]. Ocean Engineering, 2017, 133: 197–214
[8] MUHOVIC J, MANDELJC R, BOVCON B, et al. Obstacle tracking for unmanned surface vessels using 3-D point cloud[J]. IEEE Journal of Oceanic Engineering, 2019: 1–13
[9] 韩敏, 闫阔, 秦国帅. 基于改进KAZE的无人机航拍图像拼接算法[J]. 自动化学报, 2019, 45(2): 305–314
[10] ALCANTARILLA P F, NUEVO J, BARTOLI A. Fast explicit diffusion for accelerated features in nonlinear scale spaces[J]. In Proceedings of the British Machine Vision Conference, Bristol, England, 2013: 13.1–13.11
[11] ROSTEN E, DRUMMOND T. Faster and better: a machine learning approach to corner detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2008, 32(1): 105–119
[12] ALAPHI A, ORTIZ R, VANDERGHEYNST P. Freak: fast retina key-point[J]. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Providence, RI, USA, 2012: 510–517
